You are on page 1of 9

Electrical Engineering in Japan, Vol. 166, No.

4, 2009
Translated from Denki Gakkai Ronbunshi, Vol. 126-B, No. 6, June 2006, pp. 619626

New Control for HVDC System Connected to Large Windfarm

KENICHI TANOMURA, JUNICHI ARAI, YASUHIRO NORO, KIKUO TAKAGI,


and MASAKAZU KATO
Toshiba Corp. Power and Industrial Systems, Japan

voltage phase. In addition, some proposals have been made


regarding power transmission from windfarms [2, 3], and
such systems are being investigated at the experimental
level. As regards self-commutated HVDC transmission
from windfarms, the authors studied the case in which there
are no AC sources other than wind power generators on the
windfarm side [4, 5]; no other detailed reports are available.
In these previous studies, we applied conventional
inverter control, which is used for connection with general
AC power systems. In conventional inverter control, the AC
phase is detected, and the inverter-generated voltage is
adjusted accordingly [6, 7]. Just as in ordinary governorfree control of induction machines, active power control is
applied in direct proportion to the deviation from the fundamental frequency. However, when such control is applied
to autonomous wind farms, there are problems with start-up
and stable frequency control.
In the case of DC transmission systems connected
directly to windfarms, a communications system would be
necessary to support coordinated control of inverters on the
sending and receiving sides; in addition, when a fault occurs
in an AC system, the self-commutated inverters would stop
because of AC and DC overcurrents, thus causing a temporary interruption. Here we propose a new control method
to solve these problems.
The control method proposed in this study utilizes
self-commutated inverter characteristics along with induction machine characteristics, so that the generated power is
transmitted to the receiving system via a DC system, and
the frequency on the windfarm side is controlled by the DC
system. All generated power can be received steadily, even
though the receiving system has a relatively small capacity.

SUMMARY
HVDC consisting of self-commutated inverters is
able to be applied for power transmission connecting from
a remote large windfarm to a weak AC system. Most
self-commutated HVDC is applied for connection between
two AC systems that have synchronous power sources, but
it is not suitable for a windfarm that consists of induction
generators without synchronous power source. This paper
presents new control for the self-commutated HVDC system connected to a large windfarm of induction generators.
The effect of the proposed control is shown by EMTP
simulation. 2008 Wiley Periodicals, Inc. Electr Eng Jpn,
166(4): 3139, 2009; Published online in Wiley InterScience (www.interscience.wil e y. c o m ) .
DOI
10.1002/eej.20539

Key words: self-commutated inverter; induction


machine; wind power; inverter control; EMTP.

1. Introduction
The construction of windfarms is progressing in the
context of demand for sources of abundant renewable energy. Existing land-based wind power facilities have been
built in areas having very large AC power systems (thus
providing a large short-circuit capacity) [1]. In the future,
however, we may expect large-scale sea-based windfarms
provided with long-distance transmission cables. The receiving AC systems will be small and weak in comparison,
which suggests the possibility of using self-commutated
DC transmission. Existing self-commutated DC transmission schemes assume AC systems on both ends, with both
inverters operated in coordinated fashion depending on the
power to be transmitted. In this case, the AC frequency of
the AC system is maintained, and the inverters are controlled so as to adjust the inverter-generated voltage to the AC

2. Assumed System
Here we assume power transmission from a windfarm located on a remote island or at sea, with the system
configuration shown in Fig. 1. Below we refer to the windfarm side as the sending side, and to the AC power system
2008 Wiley Periodicals, Inc.
31

An ordinary AC power system with a voltage of 33


kV and a short-current capacity of 20 MVA is assumed on
the receiving side. This is a rather small system compared
with the DC transmission capacity (20/12.5 = 1.6). In many
cases, the AC power systems to which windfarms are connected are likewise weak systems with small short-circuit
capacity, and hence we set the short-circuit capacity sufficiently low. The ratings of the whole system are given in
Table 1.

3. Proposed Control
Below we survey problems that can arise when selfcommutated DC transmission is applied to windfarms,
explain our approach to solving these problems, and propose a specific method of controlling the sending and
receiving sides. We assume that there are no interconnections with other AC systems on the sending side, thus
considering an autonomous windfarm.

Fig. 1. System under consideration.

side as the receiving side. We assume that ten 1-MW wind


power generators are installed at the sending side, with a
total rated output of 10 MW. There are various types of wind
power generators, but here we assume the simplest stalltype induction generators. This design allows the use of
fixed blades without pitch control, with as few moving parts
as possible, resulting in low price, ruggedness, and high
reliability. We assume that no synchronous power source is
installed on the sending side. The DC transmission system
has a monopolar configuration using a 50-km cable. Cable
is also laid on the land side, with a single-point ground. A
self-commutated inverter with a bridge circuit using highvoltage self-extinction devices (IEGT [8]) is employed for
AC-DC conversion. The DC transmission specifications
are: 12.5 MVA, 25 kV, 0.5 kA.

3.1 Approach to control


In a windfarm connected directly to a self-commutated DC transmission system, the power supply from the
windfarm can be utilized in full, and transmission can be
continued even in the case of a system fault. When designing control so as to implement these features, the following
problems must be considered.
(1) Establishment of DC voltage and AC voltage on
the sending side at start-up.
(2) Coordination between inverters on the sending
and receiving sides during power generation.
(3) Inverter shut-off on the sending side in the case of
a sending fault.
(4) DC overvoltages caused by an AC voltage drop
on the receiving side in the case of a receiving fault.

Table 1. System ratings

Now let us consider how the above problems may be


solved.
(1) Start-up At start-up, the receiving-side inverter is
first operated to establish DC voltage. We inject an AC
resistance on the receiving side for a short time; when the
DC voltage is set up due to the rectifying action of the
inverters diodes, gate pulses are fed to the inverter to
establish the DC voltage. The AC voltage is maintained at
a specified level, just as in STATCOM control. Once the DC
voltage is established, the inverter on the windfarm side is
operated, and an AC voltage of a specified frequency and
amplitude is established. After that, the circuit breaker is
activated on the AC side, and the windfarm service load is

32

supplied. At this moment, power is sent from the receiving


side to the windfarm side.
(2) Power generation When an AC voltage is established on the windfarm side, wind power generators are
successively connected to produce power. The generators
start operation with a leading (negative) slip with respect to
the frequency generated by the inverter. The power thus
generated is sent to the DC circuit via the inverter. Since
there are no AC power sources other than the wind power
generators on the windfarm side, and the sending-side
inverter is controlled to maintain a certain AC voltage, the
inverter acts as a slack power source. As a result, the
generated power flows into the DC circuit, even without
conventional power control. On the other hand, the DC
voltage increases with the generated power, but a fixed DC
voltage is maintained by the receiving-side inverter, and
therefore the power fed to the DC circuit from the sending
side is transmitted unaltered to the AC system on the
receiving side. Thus, even though power control is not
applied to both inverters, the power generated by the windfarm passes through the DC circuit and is then transmitted
to the AC system on the receiving side. The amplitude and
frequency of the AC system voltage on the windfarm side
are governed by the inverter, and the induction generators
produce power depending on that voltage. The operation of
the inverter on the receiving side depends on the frequency
of the AC system. Thus, stable coordinated DC transmission between the sending and receiving sides becomes
possible, without relying on fast communications between
the inverters.
(3) Fault on sending side When a fault occurs in the
AC system on the sending side, the AC voltage drops and
the inverter on the sending side produces a fault current. An
additional circuit is provided so that the voltage produced
by the inverter exceeds the windfarm AC voltage by an
amount corresponding to the impedance of the inverters
transformer. This circuit is called a limiter compensation
circuit. In this case, no overcurrent occurs because a current
of more than 1 p.u. does not flow, even though the AC
voltage drops to zero. Therefore, system operation can be
continued without inverter shutoff caused by overcurrent,
and the normal voltage recovers once the fault is removed.
(4) Fault on receiving side When a fault occurs in the
AC system on the receiving side, the AC voltage drops. In
this case, the fault current is suppressed by inverter current
control, and the normal operation state is restored once the
fault is removed. The problem here is that the DC voltage
rises. Since the AC voltage drops and power cannot be
transferred, the power generated by the windfarm accumulates inside the DC circuit, resulting in an overvoltage.
There is no way to transfer the power, and the only solution
is to suppress its generation. For this purpose, we suggest
increasing the frequency on the windfarm side. Since induction generators are assumed to be installed in the wind-

farm, the internal voltage has a leading slip. If the frequency


of the AC voltage is increased, then the slip of the internal
voltage frequency becomes relatively small for a short time,
and hence power generation decreases.
Here we use the DC voltage as the signal to increase
the frequency. When the DC voltage exceeds some overvoltage level, say, 1.1 p.u., an additional circuit is actuated to
slightly increase the frequency of the sending-side inverter.
In this case, the frequency of the AC voltage on the windfarm side exceeds 50 Hz, which provides a short-term
adjustment of the generated power, because induction machines can operate smoothly in a certain speed range.
The faults that can occur in the DC system are either
cable faults or inverter faults, and therefore power transmission must be interrupted in this case.
Based on the above approach described in items (1)
to (4), we propose control on the sending and receiving
sides of DC transmission from a windfarm as described
below.
3.2 Control on sending side
The control on the sending side includes frequency
control and AC voltage control as shown in Fig. 2.
(1) Frequency control: application of frequency compensation circuit In frequency control, the fundamental
frequency F0 of the voltage generated by the inverter is set;
here F0 = 50 Hz. In addition, a circuit for frequency compensation by DC voltage is added. This compensation
circuit operates in case of a DC overvoltage so as to increase
the frequency. The frequency signal produces the voltage
phase by means of an integrator. Frequency compensation
by the DC voltage is used to decrease wind power generation when the DC voltage tends to increase (for example,
in case of a fault on the receiving side), and the generated

Fig. 2. Sending-side control.

33

power is not transmitted. The input signal is fed to the


first-order delay low limiter, so that the control signal
becomes fast when the frequency is increased, and slow
when the frequency is decreased.
(2) AC voltage control: application of limiter compensation circuit In AC voltage control, the amplitude of
the generated voltage is set, and its difference from the
detected bus voltage is used for control. In addition, the
detected AC voltage with a bias is used as the high limiter
value at the output of the AC control circuit. The limiter
compensation (bias Vb) in AC voltage control is provided
in order to suppress the inverter overcurrent that occurs at
the standard inverter voltage when the voltage drops on the
sending side (e.g., because of a fault). The bias is set
according to the impedance of the inverters transformer. In
the system shown in Fig. 1, the transformer impedance is
0.2 p.u. (self-capacity base), and therefore the bias Vb is set
to 0.2 p.u. The inverter voltage (generated behind the transformer impedance of 0.2 p.u.) is 0.2 p.u., even though the
AC bus voltage drops to 0.0 p.u.; therefore, the current is
1.0 p.u. (overcurrent is suppressed).

This is a form of conventional self-commutated inverter control that includes DC voltage control and AC
voltage control. The phase on the AC side of the interconnecting transformer is detected from the AC voltage on the
receiving side and used for internal voltage phase control.
As in conventional control, the three-phase voltage is transformed into two-axis d-q parameters, and a PLL phase
detector is provided. The inverter current is transformed
into d-q parameters, and current control is performed with
regard for the DC/AC control outputs. The current control
produces the phase Vdr and amplitude Vqr of the inverter
voltage. These signals are used to derive the three-phase
voltage generation signal for the PWM control input. Here
XC is the impedance of the inverters transformer (in our
case, XC = 2 p.u.). Thus, the d-q parameters are decoupled.
As a result, the inverter can be operated while controlling
the DC voltage and AC voltage. In addition, voltage control
suppresses overcurrents that occur when the AC voltage
drops (e.g., in case of an AC system fault).
Previously, the authors applied the conventional control used for DC transmission (that is, the aforementioned
receiving-side control) on both sides, and examined feedback frequency control in the initial stage of active power
control on the sending side [4]. However, it required switching of the frequency control on the sending side in the
transition from start-up to power generation. On the other
hand, the control method proposed in this study allows
stable operation from start-up to power generation, as well
as continuous operation in case of AC faults, without relying on fast communications between the sending and receiving sides.

Based on the two control outputs, the generated voltage phase and generated voltage amplitude, the three-phase
voltage transform is calculated and sent to a conventional
PWM control circuit. In PWM control, firing and extinction
pulses are sent to each inverter arm. By using such control,
the AC voltage of the windfarm can be maintained at the
specified frequency and amplitude.
3.3 Control on receiving side
The control on the receiving side is illustrated in Fig.
3.

Fig. 3. Receiving-side control.

34

4. Verification by Simulations
We performed simulations on the system shown in
Fig. 1 using EMTP (ElectroMagnetic Transients Program)
[10]. We used the Universal Machine model [2] included in
EMTP for the induction generators, with the parameters
given in Table 2.
The control system was simulated by TACS, with the
inverters being represented by a TACS control switch
model. The settings were Kv1 = 10, Tv1 = 0.1, Kf = 20, Tf =
2.0, Kv2 = 0.5, Tv2 = 0.02, Ke = 0.3, Te = 0.0033.
Since the objective was to confirm the basic control
functions and their effectiveness, we disregarded dead-time
delays of the control systems, while considering voltage
and current detection delays.
4.1 Operation of generators
The results calculated for cut-in of the first generator
after the establishment of the AC voltage on the windfarm
side are shown in Fig. 4. Panel (a) presents the AC voltage
waveforms on the sending side; here the inverter is operated
at fixed frequency (50 Hz) and voltage (33 kV). At instant
t1, the induction generator cuts in, causing some distortion
of the waveform. Panel (b) shows the inverter current. The
waveform indicates an in-rush current in the induction
generator. Here we assume that the pre-load is a service load
of 100 kW, and that the appropriate current flows before
generator cut-in. As regards the inverter, the operation of
one generator with relatively small capacity poses no problems such as overcurrent.
Figure 5 illustrates the case in which 9 generators are
in operation, and the 10th cuts in. As a result of connecting
the 10th generator, the peak current increases by up to 0.4
p.u. The inrush current is distributed between the inverter
and the 9 generators, which results in a weaker effect than
that due to the cut-in of the first generators. On the other
hand, as can be seen from Fig. 5(b), the inverter current
includes the currents of the 9 generators plus the inrush
current of the 10th generator, and therefore inverter overcurrent protection must be taken into account.

Fig. 4. Simulation results for cut-in of induction


generator (first unit).

4.2 Generator output growth


Here we simulate the case in which the wind becomes
stronger during the rated operation of 10 generators, thus

Table 2. Parameters of induction generator

Fig. 5. Simulation results for cut-in of induction


generator (10th unit).

35

in AC voltage control (Fig. 2), while Fig. 8 assumes the


application of this compensation circuit.
In both Figs. 7 and 8, the DC voltage drops because
of the sending-side fault, and power is temporarily injected
from the receiving side to maintain the DC voltage. The
original operating state recovers once the fault is removed.
The system operates steadily, although the generated output
is halved because only five generators are in service. Comparing the AC currents, the peak value is about 7.5 p.u. in
Fig. 7, but is reduced to about 1.8. p.u. in Fig. 8. This
indicates that adding the limiter compensation circuit to the

increasing the generator output. Figure 6 presents the results obtained by assuming a stepwise torque increase of
10%.
Figure 6(a) shows the active and reactive power on
the sending and receiving side (Ps, Qs, Pr, Qr); as is evident
from the diagram, the generated power increases smoothly
from moment t2. Figures 6(b) and 6(c) show the AC voltage
and DC voltage on both sides. The voltages remain stable
and almost unchanged in either case. This indicates stable
transmission, with the transmitted DC power increasing
with generation.
4.3 AC fault on windfarm side
Consider the case in which a fault occurs on a windfarm feeder, and 5 of the 10 generators trip out. Figures 7
and 8 show the simulation results when the circuit breaker
in Fig. 1 opens because of a fault lasting for 0.1 s at a
resistance of 1 . In the diagrams, (a) shows the active and
reactive power on the sending and receiving side, and the
AC voltages, DC voltages, and AC currents are shown in
panels (b), (c), and (d), respectively. Figure 7 shows results
obtained without the use of the limiter compensation circuit

Fig. 7. Simulation results for fault on sending side


(without limiter compensation circuit).

Fig. 6. Simulation results for generator torque increase.

36

Fig. 9. Simulation results for fault on receiving side


(without frequency compensation circuit).

the frequency control on the sending side, while Fig. 10


assumes the application of this compensation circuit.
In Fig. 9, inverter frequency control on the sending
side is not provided with the frequency compensation circuit. Since the fundamental frequency is maintained on the
sending side, the generated power remains almost unchanged when the fault occurs. As a result, the DC voltage
increases to more than 1.4 p.u. After the fault is removed,
the AC voltage increases sharply to about 1.8 p.u. due to
inverter AC voltage control on the receiving side. In addition, power is released by DC voltage control, restraining
the increase in the DC voltage. After that, the inverter
control on the receiving side becomes unstable so that
operation cannot be continued. One of the reasons for this
instability is that the short-current capacity of the AC system on the receiving side is only 20 MVA. We confirmed
that instability could be eliminated by increasing the shortcircuit capacity to 25 MVA.
On the other hand, in Fig. 10, the frequency compensation circuit acts to increase the inverter frequency on the
sending side and to advance the bus voltage phase when the

Fig. 8. Simulation results for fault on sending side


(with limiter compensation circuit).

AC voltage control on the sending side is helpful in reducing the inverter current.
4.4 System fault on receiving side
The simulation results for a fault on the receiving side
are presented in Figs. 9 and 10. Here the fault lasts for 0.1
s at a resistance of 1 . In the diagrams, panel (a) shows the
active and reactive power on the sending and receiving side,
and the AC voltages and DC voltages are shown in panels
(b) and (c), respectively. Figure 9 presents the results obtained without the use of frequency compensation circuit in

37

reactive power are needed in order to suppress the overvoltage.


5. Discussion
In the proposed control, an AC voltage limiter compensation circuit is added to the conventional self-commutated inverter control in order to suppress the AC
overcurrent caused by a sending-side fault, and a frequency
compensation circuit is added in order to suppress the DC
overvoltage caused by a side fault. The proposed control
scheme is quite simple and hence it can be applied not only
to the system considered in this study but also to other DC
transmission systems involving windfarms. When applying
the proposed method to a specific system, the control
parameters must be adjusted to the system capacity, the
transformer impedance, and the short-circuit capacity of the
AC circuit. In addition, control stability in the working
frequency range must be estimated.
6. Conclusions
In this paper, we have proposed a control method for
use when a windfarm is connected to a weak AC power
system with a small short-current capacity via a long-distance DC transmission system. The basic functions and
effectiveness of the proposed control were verified by digital simulations. The proposed control method utilizes the
features of self-commutated inverters and solves the problems of conventional control. The application of the proposed method promises to allow the implementation of
remote large-capacity power plants such as sea-based windfarms.

Fig. 10. Simulation results for fault on receiving side


(with frequency compensation circuit).

DC voltage exceeds 1.10 p.u. As a result, the generated


output is decreased and the increase of the DC voltage is
kept below 1.4 p.u.
On the receiving side, an AC overvoltage (about 1.7
p.u.) appears after the fault is removed, for the following
reason. Since a 4.8-MVar filter is connected because of the
small short-circuit capacity (20 MVA), transient phenomena related to fault removal and the Ferranti effect occur. In
addition, the control signal on the receiving side increases
so as to maintain the AC voltage during the fault. As a result,
reactive power is supplied temporarily to the system. After
that, the reactive power should be absorbed by the receiving-side inverter, but the inverter has its rated capacity, and
priority is given to the release of active power so as to
decrease the DC voltage. Thus, there is not much margin
for dealing with reactive power, which is also a factor
contributing to the overvoltage. The main reason is that the
short-circuit capacity of the receiving system is small compared to the DC transmission capacity. Measures such as
increasing the inverter capacity and the installation of SVCs
or other equipment with sufficient capability to absorb the

REFERENCES
1. Yokoyama R (editor). Deregulation of power market
and technological development. Tokyo Denki University Press; 2001.
2. Sood VK. HVDC and FACTS controllers. Kluwer
Academic Publishers; 2004.
3. Applied technologies for power control of self-commutated converters. IEEJ Technical Report, No. 919,
2003.
4. Noro Y, Arai J, Takagi K, Nagaoka N. System study
of direct current power transmission system connected to wind farm. IEEE T&D; 2002.
5. Arai J, Noro Y, Takagi K, Hondo K. Analysis by
EMTP for direct current power transmission system
connected to wind farm. ICEE 2001, Xian, China.
6. Nishi K, Nakajima T, Yokoyama A. Steady-state stability analysis of shaft torsional oscillation in AC-DC
interconnected system with self-commutated converters. Trans IEE Japan 1997;117-B:905914. (in
Japanese)

38

7. Tokiwa Y, Ichikawa F, Suzuki K, Inokuchi H, Hirose


S, Kimura K. Novel control strategies of HVDC
system with self-commutated converter. Trans IEE
Japan 1992;112-B:1926. (in Japanese)
8. Yoshino T, Kanai T, Tobita M. Application of the
high-power insulated-gate semiconductor device to

the static Var compensator and trends of the converter


technology. 2002 National Convention Record IEE
Japan, Vol. 4, 4-S14-6, p 427430. (in Japanese)
9. Power electronics circuits. Ohm Press; 2000.
10. Ametani A (editor). Computer simulations of power
systems. OHM, Supplementary Volume, Sept. 1998.

AUTHORS (from left to right)

Kenichi Tanomura (member) completed the M.E. program at Oita University in 1988 and joined Toshiba Corp. His
research interests are analysis and operation of interconnected dispersed generators, planning and control of next-generation
energy systems.
Junichi Arai (member) completed the M.E. program at Waseda University in 1972 and joined Toshiba Corp. His research
interests are power system analysis, control, and operation. He received an IEEJ Paper Award in 1987. He holds a D.Eng. degree,
is a senior member of IEEE and a member of IEEE and IEIEJ.
Yasuhiro Noro (member) completed the M.E. program at Tohoku University in 1982 and joined Toshiba Corp. His research
interests are analysis of power systems and converter systems. He received an IEEJ Paper Award in 2003. He holds a D.Eng.
degree.
Kikuo Takagi (member) joined Toshiba Corp. in 1979 and graduated from Toshiba Gakuen in 1982. He is now involved
in power electronics for electric power systems.
Masakazu Kato (senior member) completed the doctoral program at the University of Tokyo in 1982. After serving as a
research associate at Hiroshima University, he joined Toshiba Corp. in 1984. His research interests are analysis, control, and
operation of energy systems. In 2005 he joined Tokyo Denki University. C.Eng. (UK) and a member of IEEE (Fellow) and
CIGRE.

39

You might also like