Professional Documents
Culture Documents
Amin
Amin
fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
1
Abstract—The stability of offshore wind power network con- distortion of output voltage, minimize filter design, and high
nected through a High Voltage dc (HVDC) transmission line can efficiency [1] - [3]. The stability of such offshore wind power
be challenging since a strong ac collection (ACC) bus might not network connected through HVDC transmission line is a
be available, when there is no rotating machine connected in that
bus. In addition, the synchronization unit (Phase-Locked-Loop) critical problem [4] - [6]. Field experience has shown that wind
has shown to have a significant impact in achieving satisfactory farms connected through HVDC system are suffering from
performance. To tackle this problem, this paper has proposed a sub-synchronous oscillation, harmonic resonance, and poor
wind energy conversion system (WECS) controller for such ACC system performance, since there is no strong grid or rotating
bus based on the synchronverter concept. A synchronverter -an machine connected to the ac collection (ACC) bus [7]. The
inverter without PLL that mimics the synchronisation mechanism
inherent to Synchronous Generator (SG)- is introduced in the resonance caused by long transmission cables [8], passive fil-
grid side of WECS Voltage source converter (VSC) where the ters [9], transformers [10], and series compensated capacitors
wind farms are connected to ac network through Modular [12] are investigated in existing literature. The possible causes
Multilevel Converter (MMC)-based HVDC system. In order of the oscillatory phenomena resulting from interaction of dq-
to determine the stability of the interconnected system, an frame WECS converter controller and HVDC system converter
impedance-based stability method is adopted. The impedances
of both the wind power inverter and MMC-HVDC converter controller are investigated in [4], [11]. Moreover, WECS and
are analytically derived and the analytical model is verified by HVDC converters require a dedicated synchronization unit,
comparing the frequency responses obtained from numerical for example, PLL which has significant impact on the system
simulation. The detailed analysis and the results presented show stability [13]- [14]. A slow synchronization unit could directly
the benefits of this controller and its potential for stability. affect the control performance and degrade the system stability
The results highlights the synchronverter’s ability in keeping
better performance compared to PLL based dq-domain control but a complex synchronization unit, on the other hand, adds
in point of stability and control in integrating offshore wind farm significant computational burden to the controller [15], [16].
through MMC-based HVDC system, since the impedance of the Several researches have been conducted to develop the
synchronverter reflects a simple R-L characteristic. On the other control for power electronics converters having no dedicated
hand, the impedance of PLL based dq-domain control impedance synchronization unit and at the same time emulate the inertia
is inductive above 2 kHz and reflects composite characteristics
below 2 kHz with different resonance points and with higher and damping as a synchronous machine [17], [18], [19].
impedance magnitude at low frequencies making it more vulner- One approach of controlling VSC is known as synchronverter
able to voltage instability. Finally, time domain simulation results [19]. Synchronverter is an inverter without PLL that mim-
are presented to validate the theoretical analysis and to show how ics the synchronisation mechanism inherent to Synchronous
the self-synchronisation impacts on the system performance. Generator (SG). At initial development of synchronverter, it
Index Terms—Modular Multi-level Converter (MMC), High requires a PLL for initial synchronization and later a self-
Voltage dc (HVDC), Wind Energy Conversion System (WECS), synchronized synchronverter has been developed where the
Synchronverters, Offshore Grids, wind farm. inverter can synchronize with the grid without a dedicated
synchronization unit [20], [21]. Synchronverter based control
I. I NTRODUCTION becomes a promising technique in various applications, such
as integrating distributed generation [22], HVDC transmission
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
2
converter mimics synchronous generator and can automatically time domain simulation results are presented to validate the
synchronize itself before connecting and can track the ACC theoretical analysis and to show how the self-synchronisation
bus frequency after connection [20]. impacts on the system performance.
Synchronverter based inverters can relatively easy integrate The rest of the paper is organized as follows. Section II
distributed generation; however determining the stability of describes the modeling and control of the wind power inverter.
synchronverter is not easy due to non-linearities of the con- In addition, an impedance model of this converter will be
troller. Examples of non-linearities are calculation of real and derived and validated by comparing with the frequency re-
reactive power, coupling between the frequency and the field sponses obtained by numerical simulation. Section III presents
excitation current loops, initial self-synchronization frequency the control and impedance model derivation of MMC-HVDC
loops. Continuous efforts have been made to investigate the converter. Section IV discusses the stability analysis and time
stability of such system by different approaches. One approach domain simulation. Finally, Section V concludes this work.
for stability study of power electronics converters is based on
state-space and eigenvalue. State-space and Eigenvalue based II. W IND E NERGY C ONVERSION S YSTEM M ODELLING
stability analysis has been performed for synchronverter in AND C ONTROL
[27], [28]; however this method requires detailed modeling
The wind farm consists of Wind turbine generators (WTGs)
of each components of wind farm and MMC-HVDC system
based on two-level full power back-to-back converters. The
and does not support local control development at individual
grid side VSC of WECS is decoupled with generator side
terminals. Another approach of stability analysis is based on
VSC by the dc-link capacitor, therefore, the turbine mechanical
the impedance [5], [13], [29]- [31]. Impedance-based stability
system including the generator and generator side VSC can be
analysis method which is first presented in [29], is the simple
replaced by a power source [4]. For simplification of analysis,
method for analyzing the stability of such interconnected
an aggregated model of wind farm is assumed in which the
system of wind farms and MMC-HVDC system. Therefore,
wind farm is lumped into one unit of same generation capacity
the impedance-based stability method is adopted to determine
as the wind farm. The wind farms are connected to ACC bus
the stability of the interconnected system. Once the source
via step-up transformer.
and load impedances are analytically obtained, the Generalized
Nyquist Stability Criterion (GNC) is applied to predict the
stability of the interconnected system. The frequency do- A. Synchronverter-based WECS control
main impedance characteristics both for the wind farm and Control systems that enables self-synchronization of grid-
HVDC system from offshore ACC point are presented and connected converters is a research topic of high interest.
the analytical model is verified by comparing the frequency One of the first implementations called Virtual Synchronous
responses obtained from numerical simulation. The Nyquist Machine (VSM) was presented in [32]. In [18] it was shown
plot of impedance ratio (minor-loop gain) indicates potential that the VSM-concept is equivalent to droop controllers in
instability of the interconnected system. converter-based microgrids. Another implementation called
The detailed analysis and the results presented show the Synchronverter was proposed in [19]. Other similar concepts
benefits of this controller and its potential for stability. The can be found in [33]- [35]. The main difference between
results highlights the synchronverter’s ability in keeping good implementations lies in the control system details, for example
performance in point of stability and control in integrating off- the presence of current controller and fault-ride-through capa-
shore wind farm through MMC-based HVDC system. Finally, bility. Another important function is the initial synchronization
method, i.e. the ability to connect to an existing grid without
any harmful transient or overcurrent.
ib In this work, a control strategy based on synchronverter
b axis is adopted for WECS where grid-side VSCs of WECS run
vb
rotor field
axis as synchronous machine. The implementation is based on
the synchronverter proposed in [19]. For the purpose of
ia
Lf . The phase terminal voltage, vabc = [va vb vc ]T can be
written as
vc
Ls Rs diabc
dt
c axis where, iabc = [ia ib ic ]T is the stator phase currents vector;
Rs and Ls = L + M are the stator winding resistance and
Fig. 1: Structure of three-phase round-rotor SG inductance, respectively and eabc = [ea eb ec ]T is the back
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
3
i
vdc Ls vabc vo ,abc
Power +
source Cdc
-
Cwf vo ,abc
Sf
PI
vdc Dp
g
ST g
vdc ,ref PI
1 g 1
Tm,ref Js s
PWM vo ,abc
Te
Q Eqn. 2
Qref
1
Eqn. 4
Eqn. 6
eabc SI 1
1
Ks Mg iabc Rs sLs
vg 2
v
Dq Mag
SQ vo ,abc
g,ref
Fig. 2: Synchronverter (Grid-side WECS VSC) including both the electrical circuit part and electronic part.
electromotive force (EMF) due to the rotor movement and can Fig. 2 can be called electronic part of the synchronverter
be given by which controls the switches of the power part. Such a control
eabc = Mg θ̇g sinθ
f g (2) structure is assumed to be equivalent to an SG with a capacitor
bank connected in parallel with the stator terminal [19] .
where Mg is the flux field; θg is the rotor angle and Voltage in (2) corresponds to the back EMF of virtual rotor.
f g = [sinθg sin(θg − 2π ) sin(θg + 2π )].
sinθ
The inverter switches are operated such that over a switching
3 3 period, the converter outputs are to be equal to eabc as given
The mechanical part of the machine can be written by in (2) and it is achieved by a PWM technique.
The swing equation for the synchronverter can be given by
J θ̈g = Tm − Te − Dp θ̇g (3)
1
(Tm − Te − Dp θ̇g )
θ̈g = (7)
where J is the moment of inertia of all the parts rotating with J
rotor; Tm is the mechanical torque; Te is the electromagnetic where the mechanical torque, Tm is a control input and
torque and Dp is a damping factor. The electromagnetic electrical torque, Te depends on iabc and θg according to (4).
torque, Te can be found from the energy stored in the magnetic To have similar behaviour as SG, the following frequency
field of the machine and can be given by droop control loop is proposed
D E
Te = Mg iabc , sinθ
f g (4) Tm = Tm,ref + Dp (ωg − θ˙g ) (8)
3
where < ., . > denotes the conventional inner product in R . where Tm,ref is mechanical torque applied to the rotor. The
The real power, P and reactive power, Q generated by SG can grid-side WECS VSC converter control the WECS dc link
be given by, respectively voltage and reactive power. The WECS dc link voltage can
D E be controlled by controlling the mechanical torque and it is
P = Mg θ̇g iabc , sinθ
f g (5) generated by PI-controller as shown in Fig. 2 and can be given
by
Q = Mg θ̇g hiabc , cosθ
f gi . (6) kivdc
Tm,ref = (kpvdc + )(vdc − vdc,ref ) (9)
The synchronverter concept is developed based on the SG s
model (1)-(6). The electrical circuit of synchronverter (grid- where kpvdc and kivdc are the proportional and integral gain
side WECS VSC inverter) including the controller part is given of the dc voltage controller, respectively.
in Fig. 2. The electrical circuit part consists of power switch, In order to regulate the field excitation, Mg , the reactive
L − C filter and step-up transformer. The lower part of the power is controlled by a voltage droop control loop using
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
4
2 2
10 10
Z
SYNC Zw, +ve Seq
Magnitude [Ω]
Z
TOTAL
Magnitude [Ω]
0
10 0
10
Zw, −ve Seq
−2 ZSYNC
10
−2
10
100 200
Angle [o]
100
Angle [o]
0 Zw, −ve Seq
0
−100
0 1 2 3
10 10 10 10 −100 Zw, +ve Seq
Frequency [Hz]
−200 0 1 2 3
10 10 10 10
Fig. 3: Impedance model verification of Syncronverter (Solid Frequency [Hz]
line is model prediction and point-line is from numerical
simulation).
Fig. 4: Comparison of Impedance frequency response between
Synchronverter control mode and dq-domain control mode
voltage droop coefficient, Dq . The control of reactive power is with PLL obtained from numerical simulation.
shown in the lower part of Fig. 2 where the inner loop is the
voltage (amplitude) loop and outer loop is the reactive power
loop. The time constant, τv of voltage loop can be estimated The reactive power of the synchronverter is assumed to
K
by τv ≈ θ̇D as variation θ̇ is very small where K follows if be zero which gives cosθg0 = 0 and sinθg0 = 1. Hence,
q
τv and Dq have been chosen. Magnetic field excitation, Mg linearizing (2) at sinθg0 = 1 and inserting (12) and (14), ea
and reactive power reference, Qm,ref can be given by can be obtained
Gv
1
Mg = (Qg,ref − Q) (10)
z }| !{
Ks Dq θ̇g0 3RHvdc Mg0 I0 Mg0 Dq I0
ea = − + + ṽa
Qm,ref = Qref + Dq (vg,ref − vg ) (11) Ks 2Vdc0 (Js + Dp ) Ks (Js + Dp )
where vg is the output voltage magnitude. Gi
z }| !{
2
3RHvdc Mg0 V0 (Mg0 )
B. Impedance Model of Synchronverter + − ĩa (15)
2Vdc0 (Js + Dp ) (Js + Dp )
In order to apply the impedance-based stability analysis,
deriving the impedance model is a prerequisite. The ac This converter does not have any PLL and it is assumed that
impedance model of synchronverter can be obtained from (1) there is no coupling term. Therefore, the positive and negative
by dividing the phase voltage by current, therefore eabc needs sequence impedance is assumed to be equal. Linearizing (1)
to be presented in term of phase voltage and current. and inserting (15), the impedance model of the syncronverter
It is assumed that both reactive power generation and can be obtained as
reference are zero. Applying linearization on (7)-(11), small- ṽa Rs + sLs − Gi
˜ ZSY N C = − = . (16)
signal model of θ̇g can be written as ĩa 1 − Gv
˜ Hvdc (s) Dq ia0 sinθg0 Mg0 sinθg0 Including the filter capacitor Cf w and transformer impedance,
θ̇g = ṽdc + ṽg − ĩa
(Js + Dp ) Ks (Js + Dp ) (Js + Dp ) ZT,W ECS , the total impedance of Syncronverter is given by
(12)
ZSY N C
where Hvdc (s) = kpvdc + kivdc /s. ZT OT AL = + ZT,W ECS . (17)
1 + sCf w ZSY N C
For a defined wind speed, output wind power characteristics
is similar as the behaviour of constant power load (CPL) and The impedance model derived analytical is verified by numer-
it can be represented by a negative impedance, R in steady- ical simulation in MATLAB Simulink. A perturbation current
state. For single-phase system, power balance between ac and (5% of rated current) in abc-domain at different frequency
dc side can be written as from 1 Hz to 5 kHz is injected in shunt and the voltage is
v2 measured. Fast Fourier Transform (FFT) tool from SimPow-
Pph = dc = va ia (13) erSystem is used to analyze the different harmonic voltages
3R
and currents. The impedance is calculated by dividing the
and applying linearization, it can be written as
voltage by current at each frequency. The impedance frequency
3R response of syncronverter is shown in Fig. 3. The analytical
ṽdc = (V0 ĩa + I0 ṽa ). (14)
2Vdc0 model has good agreement both magnitude and phase with
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
5
WFMMC
iDC iu , a BA SM iu ,b BA SM1
iu ,c A Leg ACC LT RT Vs ,abc
1 B SM1 Submodule I s ,abc
Vdc
Arm A A A
B SM 2 B SM 2 B SM 2 A C
vdc ,u A A A B kf
B SM N B SM N B SM N * ref
v abc vabc Modulation
La La La H vac (s) and Voltage
balancing
Ra Ra Ra
is ,a Fig. 6: Control of WFMMC: ac voltage control.
Ra R is ,b
a Rais ,c
La La La
haviour and high impedance magnitude could result in voltage
instability problems.
A A A
B SM1 B SM1 B SM1
III. MMC-HVDC S YSTEM M ODELING AND C ONTROL
A A A
A. System configuration
vdc ,l B SM 2 B SM 2 B SM 2
The MMC-based HVDC system comprises converter trans-
A A A
former, wind farm side MMC (WFMMC), subsea dc cable
B SM N
il ,a B SM N
il ,b B SM N
il ,c and grid side MMC (GSMMC). The point-to-point connection
HVDC system installed in the purpose of integrating wind
farm must have dc and ac voltage control objectives. The
Fig. 5: Basic structure of MMC topology for MMC-HVDC GSMMC converter regulates the HVDC-link dc voltage. An
system. energy based dc voltage control has been implemented on
the GSMMC. The energy based control implementation for
MMC has been discussed in [37] and this control structure
impedance frequency responses obtained by numerical simu- will not be further discussed due to out of scope of this study.
lation which validates the correctness of derived impedance Assuming the dc voltage controller performance is satisfactory
model. and provides constant dc voltage at WFMMC converter dc
As can be seen from the impedance in Fig. 3, the impedance side, the GSMMC-HVDC converter can be replaced with a
characteristics of synchronverter alone is similar to the be- constant dc voltage source in the impedance model derivation
haviour of simple RL circuit. It is due to having similar of the WFMMC [4]. The WFMMC behaves a voltage source
control structure as an SG. In comparison with decoupled dq- at ac terminal for the wind farms.
domain control structure, it has less cascaded control structure A basic structure of MMC topology for MMC-based HVDC
which reduces the interaction between controllers and as a system is depicted in Fig. 5. Each phase leg of MMC consists
consequence, the impedance behaviour becomes similar to a of one upper and one lower arm i.e., an upper arm represented
simple RL circuit instead of being a higher order system. by subscript ’u’ and a lower arm represented by ’l’, connected
Including the filter capacitance and transformer inductance, in series with dc terminal. Each arm has N number of identical
it becomes a LCL circuit and resonance peak depends on the sub-modules (SMs) and series connected arm inductor La
circuit physical parameters. and also its equivalent resistor Ra to represent the losses
within the arm. The arm inductors suppress the high frequency
A control structure of wind power inverter in dq-domain
components from iu,abc and il,abc . The SMs provide two
including a PLL is discussed in [4], [11] and the impedance
different voltage level at its terminal depending on the stage of
frequency response in sequence-domain of the inverter in wind
the complementary switches. A sorting algorithm is adopted
power inverter application is presented in [4]. Fig. 4 shows
to balance and maintain the SMs capacitor voltages at vdc /N .
a comparison of impedance frequency responses between
synchronverter control mode and PLL based synchronization
in dq-domain control mode. Both positive and negative se- B. Control of WFMMC
quence impedances are presented for dq-domain control. As Many researches have been focused on modeling [38]-
can be seen from the impedance of dq-domain control, at [41] and controlling the MMC for example, Model predictive
frequency higher than 2 kHz, it is inductive; however, below control [42], [43], energy based control [37], PR regulator
2 kHz the impedance curves are composite and consist of in stationary frame [5], PI controller in synchronous rotating
different resonance points. In addition, the magnitudes of the frame [44], [45]. PR regulator in stationary abc-frame and
impedances are significantly higher than for synchronverter PI controller in synchronous rotating (dq-) frame are simple
control mode. With different resonance points in impedance methods and easy to implement compared to other methods. In
characteristics the system is more prone to oscillatory be- theory, the PR regulator in stationary frame is equivalent to the
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
6
Magnitude [Ω]
3
of the PR regulator over PI-regulator is that the need for 10
synchronous dq transformation in three phase system can be 10
2
Angle [o]
WFMMC to provide an ac voltage source for the wind farm. 100
The controller is implemented in abc-reference frame and a 0
PR-controller is used to achieve the zero steady state error for −100
sinusoidal quantities [5]. The modulation voltage reference in
−200 0
stationary abc-frame can be given by 10 10
1
10
2
10
3
Frequency [Hz]
ref ∗
vabc = Hvac (s) × (vabc − vabc ) + kf vabc (18)
Fig. 7: Frequency response of MMC impedance model (Solid-
∗ line is model prediction and points are from numerical simu-
where vabc is the reference ac voltage in abc-domain; kf is the
feed forward gain of ac voltage and Hvac is the PR-controller lation).
transfer function as
kr,vac s
Hvac = kp,vac + 2 2 (19) where,
s + ω1f
2N ic0 (kf − Hv ) 1 R0 iref
where kp,vac and kr,vac are the proportional and resonance Gg =2 − − c
− 2(kf − Hv )
coefficient of the ac voltage controller, respectively. sCvd 2 vd
A circulating current suppression controller (CCSC) is kf N m(kf − Hv )is0
+
also implemented to limit the circulating current through 2sCvd
the switches. The circulating current flowing in each phase N 1 Ra iref
2
kf N m2
c
consists of a dc component and low-order harmonic com- Gis =sLa + Ra + − +
sC 2 vd 4sC
ponents, mainly the negative sequence component with the
N 1 Ra iref
frequency twice the fundamental [48]- [50]. Therefore, in c H i
cc s0
Gic = − − kf m
order to restrain the second order harmonic current, a CCSC sC 2 vd vd
is adopted in abc-reference. The reference voltage generated kf N m 2(Hcc − Ra )ic0
− 1+ .
by the CCSC can be expressed as 2sC vd
vcref = Hcc (s)(iref
c − ic ) + Ra iref
c (20)
The circulating current suppression depends on the proper
where Hcc (s) is a PR-controller with resonance frequency selection of CCSC gain, therefore the PR coefficient has
2xω1f ; iref
c is the circulating current reference and Ra iref
c significant impact on the impedance of the MMC. Even
is introduced to compensate the voltage drop on the parasitic CCSC suppresses the 2nd harmonic components of circulating
resistance in the steady-state. current, in steady-state circulating current can have some
ripples that need to be considered. Hence, it is necessary to
C. Impedance model of WFMMC find an expression of circulating current in terms of ac voltage
In this subsection, with ac voltage closed-loop control an and current. Combining (18), (20) and (23) and applying
impedance model of WFMMC will be derived which includes linearization in frequency domain, and rearranging, it can be
dynamic of the CCSC. Internal dynamics of MMC can be obtained
described as third order system and are given by [51] GB GC
ĩc = ṽg + ĩs (25)
P
GA GA
v ref is 2v ref
C dvc
=− s + 1− c ic (21)
N dt vd vd where
C dvc∆ 2v ref is 2v ref ic N Ra iref
2ic0 (Ra − Hcc )
c 1
= 1− c − s (22) GA =sLa + Ra − − 1−
N dt vd 2 vd sC 2vd 4 vd
P P kf N mRa is0 kf N m2
dic vd vc v ref vc v ref v ∆ − + + Hcc
La = − + c + s c − Ra ic (23) 4sCvd 4sC
dt 2 4 2vd 2vd
N is0 (Hv − kf ) Ra iref
c 1 kf N mic0 (Hv − kf )
Combining (18), (20), (21) and (22) and applying linearization GB = − +
sCvd 2vd 4 2sCvd
in frequency domain, and rearranging, it can be obtained [5] ref
kf N m kf N mRa iref
kf N m Ra ic 1 c
GC = − − + − .
Gg ṽg + Gis ĩs + ĩc Gic = 0 (24) 2sC 2vd 4 8sC 4sCvd
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
7
Offshore Onshore
SM1 SM1 SM1 SM1 SM1
Ls Dc cable SM1
Fig. 8: Simplified configuration of interconnected system of wind farm and HVDC system.
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
8
5 1 0.02
0.015
∞
∞ 0.5 0.01 Unit Circle
Unit Circle ∞ Unit Circle (−1, j0)
Imaginary Axis
Imaginary Axis
Imaginary Axis
ω
ω 0.005 ω
0 0 0
−0.005
−0.5 −0.01
801 Hz
−0.015
−5 −1 −0.02
−3 −2 −1 0 1 2 3 −10 −8 −6 −4 −2 0 2 4 −3 −2 −1 0 1 2
Real Axis Real Axis Real Axis
B. Stability Analysis of the Interconnected System shows the Nyquist plot of minor-loop gain. As can be seen,
In order to determine the stability of the interconnected the Nyquist plot of the minor-loop gain does not encircle the
system an impedance-based method in frequency domain is point (-1, j0), therefore the system will operate stably for this
adopted. Fig. 9 shows the small-signal equivalent impedance condition.
model of the interconnected system, where ZW F M M C is the An example case is presented to show that it can predict
impedance of WFMMC defined in (27) including transformer instability of the interconnected system. The filter inductor of
leakage inductance and ZW P represents the impedance of wind power inverter is purposely increased to 0.5 mH while
the wind turbine inverter in synchronverter control mode. The it is 0.05 mH (0.14 pu) initially. By increasing this inductor
impedance of a wind farm can be obtained by we are reducing the power rating of the wind power inverter.
However, the impedance is calculated for base power condition
ZW P,i = ZT OT AL + Zac,cable (28) with 5 MW wind power output and resulting Nyquist plot of
where i=1,2,3,... is the number of wind farms and Zac,cable minor loop gain is shown 10 (b). As can be seen from zoom
is the ac cable impedance. The investigated system under this view of Nyquist plot ( Fig. 10 (b)) in 10 (c), the minor loop
study has two wind farms, hence the total impedance can be gain encircles the point (-1, j0), the interconnected system is
given by predicted to be unstable. The minor loop gain crosses the unit
circle at three different frequencies, however the crossing point
1 1 at frequency around 801 Hz is the closest point to (-1, j0), the
ZW P = 1/ + . (29)
ZW P 1 ZW P 2 system can have an oscillation at this frequency.
The WFMMC is modeled by its Thevenin equivalent circuit
in the form of ideal voltage source, (Vs ) in series with C. Time domain simulation
the WFMMC impedance, ZW F M M C while the aggregated In order to validate the above theoretical analysis, the
wind power inverter is modeled by its Norton equivalent interconnected system as shown in Fig. 8 has been built in
circuit consisting of current source (Iw ) in parallel with the MATLAB/Simulink environment association with the Sim-
wind power inverter impedance, ZW P defined in (29) which PowerSystem Block-set with detailed switching model of
includes the capacitive filter inductance, transformer leakage WECS VSC and MMC with 24 SMs per leg. The SMs are
inductance and cable impedance. based on the half-bridge module. A balancing algorithm is
From Fig. 9, the voltage at ACC point can be derived as used to balance the SMs capacitor voltage. A Matlab code
Vs (s) 1 has been implemented to generate the PWM signal for the
vACC (s) = −1 (30) switches. The simulation parameters are the same as the
ZW F M M C (s) 1 + ZW F M M C (s)ZW P (s) analysis done in the frequency domain using Nyquist stability
which resembles the close-loop transfer function of negative criteria and the parameters of synchronverter-based WECS
feedback control system and the minor loop feedback gain can grid side VSC and MMC converter station are given in Table
be defined as I and II in Appendix.
−1 To validate the proposed synchronverter control, case studies
L(s) = ZW F M M C (s)ZW P (s) (31)
to show the initial self-synchronisation property of the system
hence, the stability of the interconnected system can be have been included. The detailed self-synchronization of syn-
determined by checking whether the minor loop gain, [L(s)] chronverter in power control mode including the droop control
satisfies the Nyquist stability criterion. has been discussed [20]. The synchronverter based WECS
Based on the parameter listed in table I and II, the inverter regulates the dc link voltage instead of power in this
impedance of the WFMMC and wind power inverter are case which requires one additional switch, ST for initial self-
calculated for full wind power output condition and Fig. 10 (a) synchonization as shown in Fig. 2. The control system has
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
9
200 200
200
100
150 100
0
100 50
−100
50 0
0.2 0.22 0.24
−200 0
1.9 1.95 2 2.05 2.1
0 0.5 1 1.5 2
200 40
50 A step reduction
CB turn on of wind power
100 20
Current (A)
200 −20
0 18.3 0.5 1 1.5 2
100 18.2
Voltage (kV)
18.4
0 18.1
0.2 0.22 0.24
−100 18.2
−200
0 0.1 0.2 0.3 0.4 0.5
18
100 0 0.5 1 1.5 2
Time (s)
50
Current (A)
Fig. 13: Wind farm side MMC arm voltages, arm currents,
0
circulating currents and cell voltages for WFMMC.
−50
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
10
1.1 1.1
Ref. Ref.
Meas. Meas.
1.05 1.05
1
1
0.04 pu step increase
0.04 pu step increase 0.95 of reference voltage
0.95 of WECS dc voltage 0.07 pu step deacrease
0.07 pu step decrease of wind power
of WECS output power 0.9
0 1 2 3 4
0.9 Time (s)
0 1 2 3 4
Time (s)
Fig. 16: The dc link voltage of dq frame control of WECS
Fig. 14: The dc link voltage of Synchronverter based WECS grid side VSC (Red dashed line is reference value and blue
(Red dashed line is reference value and blue solid line is solid line is measured value).
measured value).
0.4 0.04 pu step increase 0.07 pu step decrease component. The reactive power has good response during
of WECS dc voltage of WECS output power transient events, since the voltage of ac collection bus is
0.2 Q supported by the Synchronverter by means of controlling the
reactive power. As can be seen the behaviour of the wind
0 power inverter is a kind of first order system and is similar to
a SG. It does not have any oscillatory behaviour.
−0.2 The dq frame control is also implemented on the ACC
0 1 2 3 4
Time (s)
bus side WECS VSC to have a view of comparison of
performance with proposed technique. The electrical circuit
Fig. 15: The active and reactive power output of Synchron- parameters and rated power remain the same for both the
verter based WECS (Red line is the active power and black synchronverter control and the dq-based control. The inner-
dash-line is the reactive power). loop current control is assumed and in the outer-loop, a dc
voltage control is implemented in the dq frame control. The
current controller is tuned based on the modulus optimum
the circulating currents has been increased which has increased criteria and the outer-loop dc voltage controller is tuned
the SMs capacitor voltages, however there is no instability or based on the symmetrical optimum tuning criteria [52]. The
electrical oscillation in the system similar to the phenomena detailed control structure of the dq frame control is available
noticed in [5] without CCSC and the interconnected system in literature [6] and is not further discussed due to out of
of wind farms and HVDC operates stably. scope. A time domain simulation is carried out with dq frame
Second time domain simulation has been carried out to show control and the same transient disturbances are introduced. Fig.
the performance of the WECS dc link voltage controller and 16 shows response of dc link voltage in the dq-frame control
the resulting time domain responses are presented in Figures of the WECS VSC. In comparison with Synchronverter-based
14 and 15. As can be seen from time domain response, control, the dq-frame control tracks the reference voltage
the system operates stably as predicted by the Nyquist plot satisfactory; however when the active power from the wind
in previous subsection. Moreover, two transient events are turbine generator changes, the performance of the dc volt-
introduced to show the performance of the control designed age controller does not remain satisfactory. The dq-control
for the Synchronverter-based WECS. At 1 s, a step increase of structure has cascaded multiple control loops which result
0.04 pu dc voltage reference to the Synchronverter is applied oscillatory behaviour due to interaction of the WECS VSC
and at 2.5 s, a step reduction of 0.07 pu active power has been controller and HVDC converter controller [4]. The impedance
applied. The dc link voltage of the WECS is shown in Fig. 14. model in Fig. 4 and time domain simulation results in Figures
The performance of the dc voltage controller is satisfactory, it 14 and 16 clearly show that synchronverter based control has
follows the reference voltage. When the dc voltage reference better performance in integrating offshore wind farm through
is step-up 0.04 pu at 1 s, the voltage controller tracks the MMC-based HVDC system.
reference, the delay to follow the reference depends on the dc The wind power is varying continuously due to variation of
voltage PI-controller time response. The output power step the wind speed, therefore a variable wind profile has been
reduction is applied at 2.5 s which reduces the WECS dc used to study the performance of the synchronverter-based
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
11
6 FFT
Wind power (MW)
500
5.5
Voltage (kV)
5
0
4.5
4 −500
0 0.5 1 1.5 2
1.05 0.8 0.85 0.9 0.95 1
dc link voltage (pu)
1000
Current (A)
1 500
0.95 −500
0 0.5 1 1.5 2
Time (s)
−1000
0.8 0.85 0.9 0.95 1
Fig. 17: Response of WECS dc link voltage with variable wind Time (s)
power, (i) variable wind power and (ii) WECS dc link voltage.
(a)
800 Hz
WECS. Fig. 17 shows the applied variable wind power and
the responses of the dc link voltage of the WECS for the 600
variable wind power. Since the wind power is varying, the dc Magnitude (%)
link voltage is also varying accordingly and remains in the 400
operating range (0.95-1.05 pu). The variation of wind power
does not have any impact on the stability of the system. Since 200 50 Hz
the stability analysis has been performed for small-signal, the
system remains stable for the small-signal range. To ensure
0 200 400 600 800 1000
the stability of the system, the stability analysis has been Frequency (Hz)
performed for the entire operating range.
(b)
Time domain simulation has been carried out for the higher
value of the wind power inverter filter inductor. This inductor Fig. 18: (a) Three-phase ac voltages and currents at ACC bus
is purposely increased to 0.5 mH as described in previous for high value of filter inductor of WECS and (b) FFT of ac
subsection while the wind farm output remain the same at voltage.
the rated value. Fig.18 shows the three-phase ac voltages and
currents at ACC bus and FFT of the voltage. As can be seen the
system is unstable for this condition. The system operates few compared to decoupled dq frame, since the synchronverter
ms to s depending on the interconnected system parameters control does not require a dedicated synchronization unit and
in the simulation, however in the real world the system is it results in less cascaded control blocks.
equipped with many switchgear and protection system which In order to determine the stability of the interconnected
prevent the system operation. An FFT is performed to check system, an impedance based stability method is adopted. The
the frequency of the oscillation and the most dominant fre- impedances of both the wind power inverter and MMC-HVDC
quency is found around 800 Hz in unstable condition which is converter are analytically derived and analytical model is
predicted in frequency domain analysis in previous subsection. verified by comparing the frequency responses obtained from
numerical simulation. The detailed analysis and the results
V. C ONCLUSION presented show the benefits of this controller and its potential
The stability of the offshore wind power network connected for stability. The results highlights the synchronverter’s ability
through a HVDC transmission line is a critical problem since in keeping better performance in point of stability and control
there is no direct connection to a strong ACC bus. To tackle in integrating offshore wind farm through MMC-based HVDC
this problem, this paper has proposed a WECS controller system.
for such ACC bus based on the synchronverter concept. A
synchronverter is introduced in the grid side of WECS VSC
where the wind farms are connected to ac network through
MMC-based HVDC transmission system. The design of the
synchronverter is embedded in the VSC to mimic the way SG
synchronise. The control strategy supports the ACC bus volt-
age that helps to achieve stable system. The observed results
show that this controller minimizes the control interactions
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
12
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TEC.2017.2661540, IEEE
Transactions on Energy Conversion
13
[33] M. Ashabani, F. D. Freijedo, S. Golestan and J. M. Guerrero, ”Inducvert- Mohammad Amin (M’11) received the B.Sc. de-
ers: PLL-Less Converters With Auto-Synchronization and Emulated In- gree in electrical and electronic engineering from
ertia Capability,” in IEEE Transactions on Smart Grid, vol. 7, no. 3, pp. Chittagong University of Engineering and Technol-
1660-1674, May 2016. ogy, Chittagong, Bangladesh, and the M.Sc. degree
[34] M. Ashabani and Y. A. R. I. Mohamed, ”Novel Comprehensive Control in electric power engineering from Chalmers Uni-
Framework for Incorporating VSCs to Smart Power Grids Using Bidirec- versity of Technology, Gothenburg, Sweden in 2008
tional Synchronous-VSC,” in IEEE Transactions on Power Systems, vol. and 2011, respectively. He is working toward the
29, no. 2, pp. 943-957, March 2014. Ph.D. degree at Norwegian University of Science
[35] J. Alipoor, Y. Miura and T. Ise, ”Power System Stabilization Using and Technology (NTNU), Trondheim, Norway.
Virtual Synchronous Generator With Alternating Moment of Inertia,” in From 2011 to 2013, he was an Assistant Professor
IEEE Journal of Emerging and Selected Topics in Power Electronics, vol. with the Department of Electrical and Electronic En-
3, no. 2, pp. 451-458, June 2015. gineering, International Islamic University Chittagong (IIUC) at Chittagong.
[36] J. J. Grainger and W. D. Stevenson, Power System Analysis. New York: In 2015, he was a Ph.D. Visiting Scholar with the Wind Power Research
McGraw-Hill, 1994. Center, Shanghai Jiao Tong University, Shanghai, China. His current research
[37] G. Bergna Diaz, J. A. Suul and S. DArco, Small-signal state-space mod- interests include Power electronics converter control, stability and interaction
eling of modular multilevel converters for system stability analysis, 2015 analysis in VSC and MMC-based HVDC system and wind farm, FACTS, and
IEEE Energy Conversion Congress and Exposition (ECCE), Montreal, grid integration of wind/solar energy.
QC, 2015, pp. 5822-5829
[38] Z. Xu, H. Xiao and Z. Zhang, Selection methods of main circuit
parameters for modular multilevel converters, in IET Renewable Power
Generation, vol. 10, no. 6, pp. 788-797, 7 2016.
[39] Y. Li, Arm inductance and sub-module capacitance selection in modular
multilevel converter. Masters Thesis, University of Tennessee, 2013
[40] M. Zygmanowski, B. Grzesik and R. Nalepa, Capacitance and induc-
tance selection of the modular multilevel converter. 2013 15th European
Conf. on Power Electronics and Applications (EPE), 2013, pp. 110
[41] L. Cunico, G. Lambert and R. Dacol, Parameters design for modular
multilevel converter (MMC), 2013 Brazilian Power Electronics Conf.
(COBEP), pp. 264270, 2013
[42] M. Vatani, B. Bahrani, M. Saeedifard and M. Hovd, Indirect Finite Atle Rygg received the MSc degree in Electrical En-
Control Set Model Predictive Control of Modular Multilevel Converters, gineering from the Norwegian University of Science
in IEEE Trans. on Smart Grid, vol. 6, no. 3, pp. 1520-1529, May 2015. and Technology (NTNU) in 2011. From 2011 to
[43] J. Qin and M. Saeedifard, Predictive Control of a Modular Multilevel 2015 he was a research scientist at SINTEF Energy
Converter for a Back-to-Back HVDC System, in IEEE Transactions on Research in the field of power electronics. He is
Power Delivery, vol. 27, no. 3, pp. 1538-1547, July 2012. currently pursuing his PhD-degree at department of
[44] A. Jamshidifar and D. Jovcic, Small-Signal Dynamic DQ Model of engineering cybernetics at NTNU. His topic or re-
Modular Multilevel Converter for System Studies, in IEEE Transactions search is impedance based stability analysis of power
on Power Delivery, vol. 31, no. 1, pp. 191-199, Feb. 2016. electronic systems, where the aim is to contribute
[45] J. Beerten, G. B. Diaz, S. DArco and J. A. Suul, Comparison of to the fundamental understanding in this family of
small-signal dynamics in MMC and two-level VSC HVDC transmission methods.
schemes, 2016 IEEE International Energy Conference (ENERGYCON),
Leuven, Belgium, 2016, pp. 1-6.
[46] D. N. Zmood, D. G. Holmes and G. H. Bode, Frequency-domain analysis
of three-phase linear current regulators, in IEEE Transactions on Industry
Applications, vol. 37, no. 2, pp. 601-610, Mar/Apr 2001.
[47] R. Teodorescu, F. Blaabjerg, M. Liserre and P. C. Loh, Proportional-
resonant controllers and filters for grid-connected voltage-source convert-
ers, in IEE Proceedings Electric Power Applications, vol. 153, no. 5, pp.
750-762, September 2006.
[48] S. Rohner, S. Bernet, M. Hiller, and R. Sommer, Modelling, simulation
and analysis of a Modular Multilevel Converter for medium voltage
applications, in IEEE International Conference on Industrial Technology Marta Molinas (M’94) received the Diploma de-
(ICIT), pp. 775782, Mar. 2010. gree in electromechanical engineering from the Na-
[49] Q. Tu, Z. Xu, H. Huang, and J. Zhang, Parameter Design Principle tional University of Asuncion, Asuncion, Paraguay,
of the Arm Inductor in Modular Multilevel Converter Based HVDC, in in 1992; the Master of Engineering degree from
International Conference on Power System Technology (POWERCON), Ryukyu University, Japan, in 1997; and the Doc-
Oct. 2010. tor of Engineering degree from the Tokyo Institute
[50] Q. Tu, Z. Xu, and L. Xu, Reduced Switching-Frequency Modulation of Technology, Tokyo, Japan, in 2000. She was a
and Circulating Current Suppression for Modular Multilevel Converters, Guest Researcher with the University of Padova,
IEEE Trans. on Power Delivery, vol. 26, pp. 20092017, July 2011. Padova, Italy, during 1998. From 2004 to 2007, she
[51] L. Harnefors, A. Antonopoulos, S. Norrga, L. Angquist, and H.-P. was a Postdoctoral Researcher with the Norwegian
Nee, ”Dynamic Analysis of Modular Multilevel Converters,” in Industrial University of Science and Technology (NTNU) and
Electronics, IEEE Transactions on , vol.60, no.7, pp.2526-2537, July 2013 from 2008-2014 she has been professor at the Department of Electric
[52] C. Bajracharya, M. Molinas, J. A. Suul and T. Undeland, Understanding Power Engineering at the same university. She is currently Professor at the
of tuning techniques of converter controllers for VSC-HVDC, in Proceed- Department of Engineering Cybernetics, NTNU. Her research interests include
ings of NORPIE 2008, Helsinki, Finland, 09-11 June 2008 stability of power electronics systems, harmonics, instantaneous frequency,
and non-stationary signals from the human and the machine. She is Associate
Editor for the IEEE Journal JESTPE, IEEE PELS Transactions and Editor
of the IEEE Transactions on Energy Conversion. Dr. Molinas has been an
AdCom Member of the IEEE Power Electronics Society from 2009 to 2011.
0885-8969 (c) 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.