Modular Multilevel Converter Modelling, Control and

Analysis under Grid Frequency Deviations

Michal Sztykiel
1
, Rodrigo da Silva
1
, Remus Teodorescu
1
, Lorenzo Zeni
2,3
,
Lars Helle
3
and Philip Carne Kjaer
3
1
DEPARTMENT OF ENERGY TECHNOLOGY
2
DEPARTMENT OF WIND ENERGY
3
VESTAS WIND SYSTEMS A/S
Aalborg University Technical University of Denmark
Pontopidanstraede 101, 9220 Aalborg Frederiksborgvej 399, 4000 Roskilde Hedeager 42, 8200 Aarhus
Aalborg, Denmark Roskilde, Denmark Aarhus, Denmark
Email: msz, [email protected] Email: [email protected]
URL: http://www.et.aau.dk URL: http://www.vindenergi.dtu.dk URL: http://www.vestas.com
Keywords
(Modular multi-level converter, VSC-HVDC, offshore wind farms)
Abstract
A tool for component sizing for MMCs has been developed and tested through simulations in PLECS.
The steady-state behaviour under grid frequency deviations - interesting for offshore wind farm connec-
tions - has been analysed, providing insights in MMC characteristics and further testing the proposed
tool.
Introduction
In recent years, increasing attention has been drawn to the application of voltage source converter HVDC
(VSC-HVDC) for connection of large offshore wind power plants to the on-land grid. Furthermore, inter-
connections between distant and asynchronous networks look attractive in order to improve the economic
efciency and stability of power systems, thus offering more room for accommodating large shares of
uctuating renewable energy. The conjugation of both aspects may possibly lead to the construction of
vast multi-terminal HVDC grids with several offshore wind farms.
Recent advances in VSC-HVDC technology have shifted the focus from the classical two-level congu-
ration to the newmodular multi-level converter (MMC) concept [1, 2], which offers a number of technical
advantages, such as very low inherent harmonic voltage content and low switching losses. Such benets
appear to have recently outweighed volume and cost drawbacks.
The rst written footprint over the modular multilevel converter circuitry was introduced in 1975 [3]
based on series connection of full-bridge cells, while in 2002, a topology with an alternative module
structure was introduced by R. Marquardt [4], in which half-bridge modules were utilized. The reduced
number of switches, along with advances in power electronic technology, then facilitated the use of this
topology in HVDC systems.
Wind turbine manufacturers thus face the need for understanding the operation of this new converter
topology and, going beyond that, to investigate their behaviour and capabilities against variation of off-
shore voltage and frequency, which may naturally arise as a consequence of the offshore grid congu-
ration or, alternatively, be utilised as actual control signals [5]. For this reason, a rst analysis of the
operation of a MMC is presented in this article, focusing, for the time being, on steady-state harmonic
spectra.
Proposed topologies
More in detail, four main HVDC converter topologies have been so far proposed from the industrial
manufacturers that utilize modular multilevel conversion concept, and still intensive research takes place
in further improving their performance. These include following:
Modular Multilevel Converter with half-bridge modules (half-bridge MMC), g. 1(a): half bridge
cells consist of single IGBT modules with integrated free-wheeling diodes, DC capacitor, bypass thyristor
and a bypass electro-mechanical relay switch for cell protection [6].
Cascaded Two Level Converter with half-bridge modules (half-bridge CTL), g. 1(b): maintains sim-
ilar operation principle to MMC, but has an original half bridge cell design, in which single IGBTs are
replaced by valves made by eight series-connected press-packed IGBT stacks. Due to press-pack IGBT
short-circuit failure mode, no bypass thyristors or mechanical switches are required. In case of a single
IGBT failure, all remaining IGBTs continue to operate at slightly increased voltage level between mainte-
nance periods [2].
Modular Multilevel Converter with full-bridge modules (full-bridge MMC), g. 1(c): full-bridge mod-
ules are used in the cascaded topology, which operation principle remains similar to the previous cases.
Single IGBT modules are used in each cell that are additionally equipped with DC capacitor, by-pass thyris-
tor, protection switch and bleeder resistors.
Hybrid multilevel converter with full-bridge modules (Hybrid MMC), g. 1(d): by inter-connecting
each full-bridge module terminal with director IGBT switch, a voltage wave can be shaped in the desired
manner while at the same time allowing zero-voltage switching for the two-level operation [7].

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
a) b) c) d)
Half-bridge MMC Modular Cell
(~2.8 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Half-bridge CTL Modular Cell
(~18 kV)
IGBT Stack
DC Capacitor
Stack
x8 series-connected STAKPAK IGBTs
Full-bridge MMC Modular Cell
(~1.5 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Hybrid 2L / Full-bridge MMC Modular Cell
(~4.5 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Director
IGBT
(a) Half-bridge
MMC topology
- single phase
leg, schematic
representation.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
a) b) c) d)
Half-bridge MMC Modular Cell
(~2.8 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Half-bridge CTL Modular Cell
(~18 kV)
IGBT Stack
DC Capacitor
Stack
x8 series-connected STAKPAK IGBTs
Full-bridge MMC Modular Cell
(~1.5 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Hybrid 2L / Full-bridge MMC Modular Cell
(~4.5 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Director
IGBT
(b) Half-bridge CTL
topology - single
phase leg, schematic
representation.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
a) b) c) d)
Half-bridge MMC Modular Cell
(~2.8 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Half-bridge CTL Modular Cell
(~18 kV)
IGBT Stack
DC Capacitor
Stack
x8 series-connected STAKPAK IGBTs
Full-bridge MMC Modular Cell
(~1.5 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Hybrid 2L / Full-bridge MMC Modular Cell
(~4.5 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Director
IGBT
(c) Full-bridge MMC
topology - single
phase leg, schematic
representation.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
a) b) c) d)
Half-bridge MMC Modular Cell
(~2.8 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Half-bridge CTL Modular Cell
(~18 kV)
IGBT Stack
DC Capacitor
Stack
x8 series-connected STAKPAK IGBTs
Full-bridge MMC Modular Cell
(~1.5 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Hybrid 2L / Full-bridge MMC Modular Cell
(~4.5 kV)
Single IGBT
Module
By-pass
Thyristor
By-pass
Switch
DC
Capacitor
Director
IGBT
(d) Hybrid MMC topol-
ogy - single phase leg,
schematic representation.
Figure 1: Summary of proposed VSC-HVDC technologies.
Control Strategies
Besides the classical outer control loops also presented by older VSC-HVDC applications in which the
two level converter type is adopted, the control of multilevel converters in principle relies on the com-
bination of current, balancing and cell voltage controllers [8, 9, 10]. As control actions, the following
requirements are addressed:
Equal sharing of phase current between phases.
Control of the total energy in the converter - power balance between DC and AC power.
Energy balance between arms - limitation of oscillating circulating current.
The rst target is achieved by setting equal arm voltage targets, while the other two objectives can be
achieved by exploiting the fact that adding the same quantity to the target arm voltage of opposite arms
does not affect the AC quantities but only the total energy balance and the arm energy discrepancy. The
general control block diagram of the modular multilevel converter controller can be illustrated such as is
represented in g. 2.
. . .
Cell
N
Cell
N - 1
Cell
1
Cell
1
. . .
Cell
N - 1
Cell
N
U
p
p
e
r

A
r
m
L
o
w
e
r

A
r
m
+
D
C
-
D
C
Gate Signals
V V V
A A
Current
Controllers
V V V
Arm Balancing
Cell
Control
Cell
Control
Cell
Control
Cell
Control
Cell
Control
Cell
Control
Figure 2: MMC general control block diagram.
Besides the cell structure mentioned above, the current VSC-HVDC main vendors (Siemens, ABB and
Alstom Grid) also provide different approaches for control implementation based on g. 2. The two
main differences are referenced to the cells voltage controllers and the way in which the arm balancing
is performed [11, 12, 8, 2, 13, 14].
Cell Controllers
For the cell voltage balancing controllers, the rst approach requires the selection of a number of cells to
be switched on and off [11, 12]. The ring control systemis based on an accurate sorting algorithmwhich
measures the DC voltages in each cell and selects the appropriate number of cells which are bypassed. In
this case, the balancing methods based on cell selection can be faster and with lower level of complexity
when compared with a distributed voltage controller in each of the cells [8].
For the distributed cell control scheme, a PWM based modulation is presented. In each of the cells,
an error signal between the cell voltage measurement and set point value is generated and its result is
combined with the modulation signals coming from the balancing and current controllers. This signal is
then compared with a triangular carrier and the PWM pattern is applied directly to the switching devices.
Moreover, the PWM pulse number can be a non-integer value, this technique being applied to improve
the cell voltage balancing even further [2, 13].
Balancing Strategies
Two main strategies are applied by the main HVDC vendors to keep the voltages between upper and
lower arms balanced. The rst method is performed by a closed loop compensation of arm currents
and/or capacitor voltage measurements. By doing this, the ring pulses sent to the upper and lower arms
are slightly different for compensating the voltage differences. The approach is centralized and each cell
is only provided with a monitoring station that transmits data back to the arm control and accomplishes
what such control unit orders.
The alternative approach requires extra hardware in the circuit between the arm reactors and the converter
AC output. This component is a tunable LC-lter which limits 2nd harmonic voltage ripple and allows
lower losses and voltage ratings of power modules. Apart from the 2nd harmonic, 3rd harmonic is
damped enough to allow both lower voltage ratings and wye-connected transformer at the converter-
side [14].
System Benchmarking and Sizing
Design Engineering Studies
In order to optimally size the HVDC system at a rst step approximation, the following studies have
been carried out in this work:
Load ow analysis: allows to rate components for maximum current ampacity under normal operation.
Based on applied system diagram, load ow analysis is divided into external load ow (g. 3(a)) and internal
load ow (g. 3(b)), which includes only the analysis of the 2nd harmonic circulating current owing within
converter arms.
Harmonic analysis: allows optimal sizing of lters within the HVDC system. Harmonic analysis is utilized
via applied Discrete Fourier Transform of arm voltage staircase waveforms obtained through specic PWM
modulation technique.
DC short-circuit analysis: allows rating arm reactors for the reliable protection of half-bridge cells in case
of the DC short-circuit state.
System Description and Modelling
A schematic representation of the analysed HVDC system is presented on g. 3(a) and 3(b). The follow-
ing main parts can be derived within its internal design:
AC System: reects the external inuence of the AC-side on the HVDC system performance. AC system
is modelled with Thevenin equivalent AC voltage source V
AC
and a short-circuit impedance Z
AC
, which
can be measured at the point of common coupling (PCC).
DC System: reects the external inuence of the DC-side on the HVDC system. DC system is modelled
with Thevenin equivalent DC voltage V
DC
and load resistance R
DC
.
High Frequency Filter: used for reactive power compensation at the PCC. Implemented design is a 2nd
order high-pass lter modelled with C
f(AC)
, L
f(AC)
and R
f(AC)
parameters.
Power Transformer: required due to different voltage levels between AC and DC systems. It is modelled
with inductance L
tr
and resistance R
tr
accordingly for leakage ux distribution and copper losses.
Arm Reactors: used for limiting load and short-circuit arm currents to the desired value, which are re-
stricted by nominal ratings of the half-bridge cells. Each arm reactor is modelled with inductance L
arm
and
resistance R
arm
, which reects its copper losses.
Half-Bridge Cells: for load ow and harmonic analyses, cells are modelled with ideal switches and a DC
voltage source, which reects the electrical behaviour of the single cell capacitor. For DC short-circuit
analysis, cells are modelled with specied IGBT switches, which characteristics include their short circuit
withstand range.
DC Filter: used to provide harmonic path for the parasitic currents. In this manner, AC harmonic currents
on the DC side will not ow through DC lines and induce electromagnetic elds, which might generate
losses and interference in neighbouring communication systems. DC lter is modelled with a pair of ca-
pacitors C
f(DC)
interconnected between DC buses and ground. Capacitor losses are reected by equivalent
series resistance R
f(DC)
.
Assumptions
In order to simplify the rst step modelling of the MMC converter, the following assumptions were made:
1. All six arms are loaded equally on the DC-side, as they operate in identical manner. Such assumption allows
dividing common DC load resistance R
DC
into six corresponding arm DC load resistances R
DC(arm)
,
according to relation:
R
DC(arm)
=
3
2
R
DC
(1)
ZAC
ZN
Lf(AC) Rf(AC)
Cf(AC)
Rtr Ltr VAC
Arm
Reactors
Upper Arms
Half-Bridge
Cells
Larm
Rarm
Larm
Rarm
Larm
Rarm
Larm
Rarm
Larm
Rarm
Larm
Rarm
Power
Transformer
High Frequency
Filter
Lower Arms
Half-Bridge
Cells
AC System
RDC(arm) RDC(arm) RDC(arm)
DC Filter
Rf(DC)
Rf(DC)
Cf(DC)
Cf(DC)
RDC(arm) RDC(arm) RDC(arm)
(a) Circuit diagram for external load ow analysis.
Larm
Rarm
Larm
Rarm
Larm
Rarm
Larm
Rarm
Vcell Vcell Vcell
Iarm(2H) Iarm(2H) Iarm(2H)
(b) Circuit diagram for internal
load ow analysis.
Figure 3
2. Generated cell voltage ripple V
cell
consists entirely of the 2nd harmonic component. Consequently, in-
ternal load ow analysis can be derived and made separately only for the 2nd harmonic circulating current
enclosed through MMC phase legs (g. 3(b)).
3. Sizing for cell capacitors is made based on zero-, fundamental- and second-harmonic average arm current
values according to:
C
cell
=
I
arm(1H)
+I
arm(2H)
+I
arm(DC)
V
cell
(2)
All higher-order harmonic currents are considered negligible to inuence cell capacitance sizing.
4. Specic methodology for sizing DC lter is out of scope of this work. Hence, assumed required DC capac-
itance corresponds to 2 ms time constant.
Methodology
A generic algorithm structure for populating the HVDC system with optimal parameter values for the
rst step approximation is shown on g. 4. The whole procedure can be divided into three main parts:
Preliminary system conguration, which includes:
Power transformer rating - power transformer is rated for the nominal AC power S
AC
of the HVDC system
and a voltage ratio V
ratio
of
V
ratio
=
m V
ref
V
AC
=
m

3
V
DC
2

2
V
AC
(3)
where V
ref
is the reference arm voltage RMS value and m stands for the modulation index.
Cell sizing - number of cells per arm N
cells
can be calculated from the assumed nominal cell voltage V
cell
according to
N
cells
= ceil

V
DC
V
cell

(4)
Arm reactors sizing I - the rst criterion for sizing arm reactors is through limiting AC arm currents I
arm
to
the desired values, which need to be limited due to the strict cell IGBT and cell capacitor continuous current
ratings.
Based on MMC operation principle, the dominant impact of the 1st harmonic arm current I
arm(1H)
and
2nd harmonic arm current I
arm(2H)
is taken into account. Reactors tuning is performed by solving AC
circuit diagrams shown on gure 3(a). The nal arm current values which change with respect to the reactor
inductance L
arm
are obtained as a sum of their most relevant components
I
arm
(L
arm
)

= I
arm(1H)
(L
arm
) +I
arm(2H)
(L
arm
) (5)
derived from an external and internal load ow analyses.
1a. AC System Ratings
1b. DC System Ratings
1. Load Flow Analysis
1a. Nominal Cell Voltage
1b. Maximum Cell Voltage Ripple
1c. Nominal Device Current Ratings
1d. Transformer Winding Arrangement
Requirements
Equipment Design
2a. IGBT Specification
2b. Cell Capacitor Specification
1a. Power Transformer Rating
1b. Cell Sizing
1c. Arm Reactors Sizing I
2. DC Short-Circuit Analysis Engineering Studies
Procedure
3a. AC-Side Harmonic Requirements
3b. DC-Side Harmonic Requirements
3a. DC Filter Specification
3b. Modulation Technique Specification
2a. Arm Reactors Sizing II
2b. HF Filter Sizing I
3. Harmonic Analysis
3a. HF Filter Sizing II
3b. DC Filter Rating
PRELIMINARY CIRCUIT
CONFIGURATION
MAIN CIRCUIT
CONFIGURATION
DETAILED CIRCUIT
CONFIGURATION
Results
Figure 4: Generic algorithm structure for the 1st approximation optimal sizing of the HVDC System.
Main system conguration, which includes:
Arm reactors sizing II - the second criterion for sizing the arm reactors is through limiting the rate of change
of possible DC short-circuit current between arms. In order to protect the cell capacitors from extensive
fault current and minimize their short-circuit contribution; the applied IGBT switches in all cells should be
able to safely disconnect the capacitors during fault occurrence.
In this work, time required for safe IGBT turn-off operation under fault conditions includes: turn-off delay
time t
d(off)
, fall time t
f
and short-circuit withstand time t
SC
. The minimum inductance of arm reactors is
dened as
L
arm
=
dt
di
u

=
t
d(off)
+t
f
+t
SC
I
SC
(6)
where I
SC
is IGBT rated short-circuit current.
The nal value for the arm inductance is the higher one - chosen among the described sizing criteria.
HF lter sizing I - Final value of arm reactor inductance allows, when necessary, preliminary sizing of the
C
f(AC)
and L
f(AC)
. In order to assure reactive power compensation at PCC, C
f(AC)
is calculated from
C
f(AC)
= 6

I
arm
V
AC

2
L
arm
(7)
L
f(AC)
is applied to provide harmonic current resonant path, so that minimum current distortions are in-
jected into the grid
L
f(AC)
=
1

2f
res
C
f(AC)

=
1

2f
sw
C
f(AC)
(8)
L
f(AC)
is sized based on optimal estimation of the resonant frequency f
res
, which should be close to effec-
tive switching frequency f
sw
of the MMC converter.
Detailed system conguration is obtained, which includes:
HF lter sizing II - By applying specied PWM modulation technique to cell IGBT switches, a harmonic
analysis is provided in order to identify resonant frequencies at specic operating points and further size
lters for optimal HVDC system according to the external requirements. Resistance R
f(AC)
is sized based
on calculated quality factor q from
R
f(AC)
= 2 f
res
L
f(AC)
q (9)
Optimal q value is set for specic harmonics damping based on derived impedance characteristic plot in the
frequency domain. Requirements regarding AC-side harmonics damping are assumed to be provided by the
grid operator.
DC lter rating - the assumed DC lter must be rated for harmonic voltages and currents, which will
ow causing its extensive heating. For this reason, harmonic magnitude and phase spectra of current and
voltages are obtained, which allow deriving maximum voltage peak for insulation coordination along with
the resulting RMS current for thermal management.
Control Strategy
As mentioned above, the control of a MMC can be implemented in different ways and the control objec-
tives are variegated [4, 13]. In this work, a mixed approach has been used and the control blocks have
been developed in C-code. The main control features are as follows:
The grid current control perfectly resembles a classical VSC-converters control in dq reference frame [2].
The balancing action is performed by control of circulating current [4], suppressing its 2nd harmonic com-
ponent and controlling its DC part. The validation of the components sizing tool, however, required this
controller to be switched off. Its features will be presented in future work.
The cell voltage control is realised in a distributed manner, correcting the modulation index on every cell
based on the ltered cell voltage error [2].
Simulation Cases
Table I provides system parameter values derived according to the methodology presented above.
Element Parameter Description Symbol Unit Value
Modulation Carrier-based PWM, phase shifted
Cell PWM switching frequency f
cell
[Hz] 1250
HVDC System Nominal Power S
AC
[MV A] 22.5-j25.5
AC System Nominal voltage (RMS) V
AC
[kV ] 21.210
Frequency f
AC
[Hz] 505%
Phase impedance Z
AC
[m] 12.1+j121.3
Neutral point impedance Z
N
[m]
DC System DC system pole-to-pole voltage V
DC
[kv] 120
Nominal power P
DC
[MW] 36
Cell rating Nominal cell voltage V
cell
[kV ] 30
Number of cells per arm N
cell
4
Maximum cell voltage ripple V
cell
% 5%
Cell capacitance C
cell
[mF] 0.315
Cell IGBT turn-OFF delay time
1
t
d(OFF)
[s] 1.75
Cell IGBT fall time
1
t
f
[s] 0.71
Cell IGBT short circuit time
1
t
sc
[s] 10
Cell IGBT short circuit current
1
I
sc
[A] 12x200
Power Transformer Voltage ratio V
ratio
[kV : kV ] 1:1
Phase inductance L
tr
[mH] 0.014
Phase resistance R
tr
[m] 1.830
HF Filter Phase capacitance C
f(AC)
[F] 4.6230
Phase inductance L
f(AC)
[mH] 87.671
Phase resistance R
f(AC)
[] 1377.13
Arm Reactor Arm inductance L
arm
[mH] 50.00
Arm resistance R
arm
[m] 48.00
DC Filter Leg capacitance C
f(DC)
[F] 41.11
Leg resistance R
f(DC)
[m] 5.00
1
According to IGBT 5SMY 12N4500 datasheet.
Table I
Steady State Operation
A computer model has been developed in PLECS in order to validate the tool for sizing the components.
Three simulation cases have been made with regard to the grid frequency, which is nominal (50Hz), 5%
below nominal (47.5 Hz) and 5% above nominal (52.5 Hz). Chosen under- and over-frequency values
are close to maximum frequency operating ranges for Irish ESB grid code [15].
Nominal frequency f
AC
= 50 Hz (no frequency deviations)
Fig. 5 shows the steady-state arm and grid currents on phase a. Harmonic analysis of the arm currents
has been made in order to assess the accuracy of the proposed sizing tool. The table from g. 5 provides
arm current sequence values for dominant 0
th
-harmonic, fundamental and 2
th
-harmonic.
0.1 0.15 0.2 0.25
500
0
500
Time [s]
I

[
A
]

I
LOW
I
UP
I
g
Phase a arm and grid currents f
AC
= 50 Hz.
Harmonic
Theoretical Analysis Computer Analysis
+seq -seq 0seq +seq -seq 0seq
0
th
[A] 0.06 0.06 61.13 0.154 0.154 63.41
1
st
[A] 254 3.98 0.378 255 0.065 0.059
2
nd
[A] 11.4 65.8 12.3 0.058 54.72 0.185
Comparison of 0
th
, 1
st
and 2
nd
harmonic currents at f
AC
= 50 Hz.
Figure 5: Simulation results for f
AC
= 50 Hz.
On the left column of g. 9, harmonic spectra are observed for each sequence arm current - taken both
from computer analysis and theoretical analysis. It can be observed that harmonic peak regions for both
analyses are overlapping each other, which proves good accuracy of the sizing tool. Higher magnitudes
of the harmonics derived from the theoretical tool allow sizing components, thus providing reasonable
safety margin.
Fig. 6(a) shows deviations in cell voltage ripple level for different set of cell capacitances, both for values
obtained from (2) for theoretical analysis and measured across cell for computer analysis.

Fig.3

0
2
4
6
8
10
12
1.244 0.647 0.290 0.170
Ccell [mF]
CellVoltage Ripple at fAC =50.0 Hz [ %]
TheoreticalAnalysis ComputerAnalysis
(a) At f
AC
= 50 Hz.

Fig.6

0
2
4
6
8
10
12
1.244 0.647 0.290 0.170
Ccell [mF]
Cell VoltageRipple at fAC = 52.5 Hz [ %]
Theoretical Analysis ComputerAnalysis
(b) At f
AC
= 52.5 Hz.

Fig.9

0
2
4
6
8
10
12
1.244 0.647 0.290 0.170
Ccell [mF]
Cell VoltageRipple at fAC = 47.5 Hz [ %]
Theoretical Analysis ComputerAnalysis
(c) At f
AC
= 47.5 Hz.
Figure 6: Cell voltage ripple variations as a function of C
cell
.
Maximum differences in 1.7% of the ripple occurred for the C
cell
= 0.290 mF, which can be explained
by means of additional impact of higher harmonics, which provide circulating current causing voltage
unbalance.
High frequency deviations f
AC
= 52.5 Hz
For over-frequency case, the same gures are analysed and reported in g. 7. Reasonable accuracy
level is obtained over all dominant values (zero-seq. DC current, positive-seq. fundamental current and
negative-seq. circulating current).
Imposed harmonic spectra from the theoretical and computer analyses on the central column of g. 9
show good accuracy of the theoretical model for the over-frequency state. Similarly as for nominal
0.1 0.15 0.2 0.25
500
0
500
Time [s]
I

[
A
]

I
LOW
I
UP
I
g
Phase a arm and grid currents f
AC
= 52.5 Hz.
Harmonic
Theoretical Analysis Computer Analysis
+seq -seq 0seq +seq -seq 0seq
0
th
[A] 0.01 0.01 61.13 9.71 9.71 63.37
1
st
[A] 250 5.48 0.413 255 9.38 0.225
2
nd
[A] 5.25 67.7 0.819 12.29 53.16 0.135
Comparison of 0
th
, 1
st
and 2
nd
harmonic currents at f
AC
= 52.5 Hz.
Figure 7: Simulation results for f
AC
= 52.5 Hz.
frequency case, higher harmonic magnitudes for theoretical model provide reasonable safety margin for
component sizing.
Fig. 6(b) shows the cell voltage ripple deviation with regard to cell capacitance. It can be observed that
no particular impact is made from over-frequency operation for the accuracy of the theoretical model.
The largest difference for C
cell
= 0.290 mF is 1.5%.
Low frequency deviations f
AC
= 47.5 Hz
Again, the same gures are presented in g. 8 for low frequency at level of 47.5 Hz. Table from the
g. 8 lists dominant harmonic values, which are similar to the values for the previous cases. In the same
manner, harmonic spectra for under-frequency operation on the right column of g. 9 are accurately
overlapping each other, with higher harmonic magnitude values computed by theoretical sizing tool.
0.1 0.15 0.2 0.25
500
0
500
Time [s]
I

[
A
]

I
LOW
I
UP
I
g
Phase a arm and grid currents f
AC
= 47.5 Hz.
Harmonic
Theoretical Analysis Computer Analysis
+seq -seq 0seq +seq -seq 0seq
0
th
[A] 0.01 0.01 61.13 11.2 11.2 63.28
1
st
[A] 250 5.48 0.413 252 12.5 0.054
2
nd
[A] 5.25 67.7 0.819 10.78 53.02 0.214
Comparison of 0
th
, 1
st
and 2
nd
harmonic currents at f
AC
= 47.5 Hz.
Figure 8: Simulation results for f
AC
= 47.5 Hz.
It can be concluded from g. 6(c) that cell voltage ripple accuracy of the theoretical model is not affected
by under-frequency operation. Highest ripple difference of 2% is obtained for C
cell
= 0.290 mF.
Conclusion
A tool for quickly sizing electric components for modular multilevel converters has been developed and
its validity has been tested through simulations using the power electronics simulation tool PLECS.
Good accuracy of the derived method has been observed and the theoretical analysis slightly overesti-
mates the actual harmonic content of circulating arm currents, allowing for a safe design of the compo-
nents.
The inuence of the cell capacitance on its voltage ripple has been tested and a satisfying match was
found between theoretical analysis and simulated results.
Moreover, grid frequency deviations were introduced in order to investigate the steady-state behaviour
of the converter under such disturbance and the developed tool proved to offer satisfying results in this
regard as well. Future work will present the control strategy in more detail and provide results on transient
behaviour of the converter under the same disturbing events.
5
[ A ]
P
o
s
i
t
i
v
e

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

4
7
.
5

H
z
5
[ A ]
P
o
s
i
t
i
v
e

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

5
0

H
z
5
[ A ]
P
o
s
i
t
i
v
e

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

5
2
.
5

H
z
4 5
n i t u d e [ A ]
4 5
n i t u d e [ A ]
4 5
n i t u d e [ A ]
3 4
M a g n i t u
3 4
M a g n i t u
3 4
M a g n i t u
2 3
M
2 3
M
2 3
M
1 2
1 2
1 2
0 1
0 1
0 1
0
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
0
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
0
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
5
A ]
N
e
g
a
t
i
v
e

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

4
7
.
5

H
z
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
5
A ]
N
e
g
a
t
i
v
e

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

5
0

H
z
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
5
A ]
N
e
g
a
t
i
v
e

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

5
2
.
5

H
z
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
4 5
t u d e [ A ]
g
q
4 5
t u d e [ A ]
N
e
g
a
t
i
v
e

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C

5
0

H
z
4 5
t u d e [ A ]
N
e
g
a
t
i
v
e

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C

5
2
.
5

H
z
3 4
M a g n i t u d
3 4
M a g n i t u d e
3 4
M a g n i t u d e
3
M a
3
M a
3
M a
1 2
2
2
0 1
0 1
0 1
0
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
0
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y
[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
0
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y
[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
Z
e
r
o

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

4
7
.
5

H
z
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
Z
e
r
o

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

5
0

H
z
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

A
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
Z
e
r
o

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

5
2
.
5

H
z
5
u d e [ A ]
Z
e
r
o

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

4
7
.
5

H
z
5
u d e [ A ]
Z
e
r
o

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

5
0

H
z
5
u d e [ A ]
Z
e
r
o

S
e
q
u
e
n
c
e

A
r
m

C
u
r
r
e
n
t

a
t

f
A
C
=

5
2
.
5

H
z
4
M a g n i t u d e
4
M a g n i t u d e
4
M a g n i t u d e
3
M a g n
3
M a g n
3
M a g n
2
2
2
1
1
1
0
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y
[
H
z
]
C
o
m
p
u
t
e
r
a
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l
A
n
a
l
y
s
i
s
0
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y
[
H
z
]
C
o
m
p
u
t
e
r
a
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l
A
n
a
l
y
s
i
s
0
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
[
H
]
C
o
m
p
u
t
e
r
a
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l
A
n
a
l
y
s
i
s
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

a
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

a
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
4
0
0
0
6
5
0
0
9
0
0
0
1
1
5
0
0
1
4
0
0
0
1
6
5
0
0
F
r
e
q
u
e
n
c
y

[
H
z
]
C
o
m
p
u
t
e
r

a
n
a
l
y
s
i
s
T
h
e
o
r
e
t
i
c
a
l

A
n
a
l
y
s
i
s
Figure 9: Comparison of sequence current harmonic spectra for different values of f
AC
.
Acknowledgement
The works and developments required for the elaboration of this paper/article have been carried out par-
tially within OffshoreDCproject (www.offshoredc.dk) which belongs to the Top-level Research Initiative
funded by Nordic Energy Research under project no. TFI PK-int 02.
References
[1] A. Lesnicar and R. Marquardt, A new modular voltage source inverter topology, in Proc. EPE, vol. 3,
pp. 24, 2003.
[2] B. Jacobson, P. Karlsson, G. Asplund, L. Harnefors, and T. Jonsson, VSC-HVDC transmission with cas-
caded two-level converters, CIGR

E SC B4 Session 2010, 2010.

[3] R. Baker and L. Bannister, Electric power converter, Feb. 1 1975. US Patent 3,867,643.
[4] R. Marquardt, Current Rectication Circuit for Voltage Source Inverters with Separate Energy Stores Re-
places Phase Blocks with Energy Storing Capacitors, German Patent (DE10103031A1), vol. 25, 2002.
[5] L. Harnefors, Y. Jiang-Hafner, M. Hyttinen, and T. Jonsson, Ride-through methods for wind farms connected
to the grid via a VSC-HVDC transmission, in Proc. Nordic Wind Power Conference. Roskilde, Denmark,
2007.
[6] K. Friedrich, Modern HVDC PLUS application of VSC in Modular Multilevel Converter topology, in
Industrial Electronics (ISIE), 2010 IEEE International Symposium on, pp. 38073810, IEEE, 2010.
[7] M. Merlin, T. Green, P. Mitcheson, D. Trainer, D. Critchley, and R. Crookes, A new hybrid multi-level
Voltage-Source Converter with DC fault blocking capability, in AC and DC Power Transmission, 2010.
ACDC. 9th IET International Conference on, pp. 15, IET, 2010.
[8] D. Siemaszko, A. Antonopoulos, K. Ilves, M. Vasiladiotis, L. Angquist, and H. Nee, Evaluation of control
and modulation methods for modular multilevel converters, in Power Electronics Conference (IPEC), 2010
International, pp. 746753, IEEE, 2010.
[9] M. Hagiwara and H. Akagi, PWM control and experiment of modular multilevel converters, in Power
Electronics Specialists Conference, 2008. PESC 2008. IEEE, pp. 154161, IEEE, 2008.
[10] A. Antonopoulos, L. Angquist, and H. Nee, On dynamics and voltage control of the modular multilevel
converter, in Power Electronics and Applications, 2009. EPE09. 13th European Conference on, pp. 110,
IEEE, 2009.
[11] A. Lesnicar and R. Marquardt, An innovative modular multilevel converter topology suitable for a wide
power range, in Power Tech Conference Proceedings, 2003 IEEE Bologna, vol. 3, pp. 6pp, IEEE, 2003.
[12] S. Allebrod, R. Hamerski, and R. Marquardt, New transformerless, scalable modular multilevel converters
for HVDC-transmission, in Power Electronics Specialists Conference, 2008. PESC 2008. IEEE, pp. 174
179, IEEE, 2008.
[13] G. Asplund, Method for controlling a voltage source converter and a voltage converting apparatus, Jan. 8
2008. US Patent App. 12/811,806.
[14] L. Harnefors and B. Jacobson, Power converter with multi-level voltage output and harmonics lter, May 27
2010. WO Patent WO/2010/057,532.
[15] EirGrid, EirGrid Grid Code. www.eirgrid.com, 2011.