Sunteți pe pagina 1din 5

Voltage Control of a Wind Power Plant Using the

Adaptive Q-V Characteristic of DFIGs


Jinho Kim, Geon Park, and Byongjun Lee Edward Muljadi
Yong Cheol Kang* Korea University National Renewable Energy
Chonbuk national university WeGAT Research Center, Laboratory, US
WeGAT Research Center Korea eduard.muljadi@nrel.gov
Smart Grid Research Center, Korea leeb@korea.ac.kr
jkim@jbnu.ac.kr
powerupme@jbnu.ac.kr
*Corresponding Author;
E-mail: yckang@jbnu.ac.kr

Abstract For the stable operation of a grid, the voltages of all disturbance occurs. To achieve this function, a WPP should
buses in the grid should be maintained within an acceptable range. not only ride-through the fault, but also supply the reactive
For a grid with high wind penetration, wind power plants (WPPs) power after the fault clearance as well as during the fault [4,
are required to have the capability to maintain the voltage at the 5]. Therefore, a WPP controller is essential to manage the
point of common coupling (PCC). Wind generators in a WPP collective response of the WGs in a WPP.
produce the different active power due to the wake effects and The researches on the voltage control of a WPP using a
consequently contain different reactive power availability. This plant controller have been reported [6, 7]. A centralized
paper proposes a voltage control scheme of a doubly-fed induction
voltage control of a WPP was proposed that allocates the
generator (DFIG)-based WPP using the adaptive reactive power-
required reactive power to a WG with a weighting factor,
voltage (Q-V) characteristic of DFIGs. The WPP controller
calculates a voltage error signal using a PI controller and sends it
which is inversely proportional to the active power
to each DFIG. From the voltage error signal, the DFIG injects the production of each WG [6]. This control scheme allows the
reactive power based on its adaptive Q-V characteristic, which WPP to supply a large amount of reactive power and thus
depends on the available reactive power. The proposed scheme recovers the voltage of the PCC fast. However, the scheme
adjusts the Q-V characteristic depending on the operating might supply the excessive reactive power due to a high
condition of a DFIG. The proposed scheme enables the WPP to weighting factor and a late voltage recovery with a large
recover the PCC voltage to the nominal value within a short time overshoot is inevitable.
after a disturbance by utilizing the reactive power capability of a A hierarchical voltage control scheme for a WPP using
WPP. The performance of the scheme was investigated for a 100 the WG controller and WPP controller was suggested [7]. To
MW WPP consisting of 20 units of a 5 MW DFIG for a grid fault achieve slow requirements, the reactive power-voltage (Q-V)
condition. The results show the proposed scheme successfully characteristic in the UK grid code is implemented in the
recovers the PCC voltage within a short time after a disturbance. WPP controller, while to achieve fast requirements, a turbine
level VAR control is used using the voltage reference. It
I. INTRODUCTION defines the reactive power demand as a function of a
Wind generation has become one of the most competitive specified voltage range around the rated voltage. The scheme
renewable energy sources. The global installed capacity of showed a good performance in terms of the fast control when
wind generators (WGs) is expected to increase to 832 GW a fault occurs. However, some errors around the rated
by 2020 [1]. The variable-speed wind generators (VSWGs), voltage are inevitable in the steady state after a disturbance
which employ power conversion devices, have been widely because the WPP controller does not fully utilize the reactive
used [2]. Among then, a doubly-fed induction generator power capability of the WPP. In addition, a WG controller
(DFIG) shares about 50 % of the wind energy market [3]. has the same gain of the reactive current response to the
To achieve the stable operation of a grid, the voltages at voltage reference. Thus, the available reactive power of WGs
all buses should be maintained within a specified operational is not fully utilized.
range. The voltage at a node, which is known as a local This paper proposes a voltage control scheme of a WPP
variable, should be controlled by a generator or a reactive using the adaptive Q-V characteristic of a DFIG. The Q-V
power compensating unit closest to the applicable node. In characteristic of a DFIG is determined by its active power
the case of a grid with high wind penetration, WGs and wind output, which depends on the wind speed arriving at the
power plants (WPPs) should be capable of supporting the DFIG. The Q-V characteristic of each DFIG has different the
voltage at the point of common coupling (PCC) when a slope and width of the linear region depending on available

978-1-4799-5138-3/14/$31.00
978-1-4799-5776-7/14/$31.00 2014
2014 IEEE
IEEE
reactive power capability. Thus, the DFIGs with larger voltage reference for WGi (uWGiref), measured voltage of the
capability of reactive power have larger slopes and thus give WG terminal (uWGi) and uWGiref. kQi and uWGiref are set to 2
more contribution on supporting reactive power when a and 1 pu, respectively.
disturbance occurs. The performance of the proposed scheme
was investigated for a 100 MW WPP consisting of 20 units
of 5 MW DFIGs under various disturbances including
reactive load variation and a grid fault using an EMTP-RV
simulator.
II. VOLTAGE CONTROL OF A WPP BASED ON THE
ADAPTIVE Q-V CHARACTERISTIC OF A DFIG

A. Control strategies of a DFIG


Fig. 1 shows a typical configuration of a DFIG. The
rotor-side converter (RSC) controls the active and reactive
powers of a stator winding. In order to maximize the stator Figure 1. Typical configuration of a DFIG
active power, a maximum power point tracking control is
implemented in the RSC controller. In addition, a voltage
control function is implemented to keep the stator terminal
voltage as a nominal value or to inject the reactive power
into the grid. Moreover, a crowbar is used to protect the
DFIG from the grid fault. This device shorts the rotor
(a) WPP controller
windings through the resistors during the fault.
On the other hand, the grid-side converter (GSC) controls
the dc-link voltage as constant and injects the reactive power
from the GSC to support the grid voltage. Alternatively, the
GSC only maintains the dc-link voltage as a constant without
a reactive power injection.
(b) WG controller
B. Conventional voltage control schemes [6, 7]
Fig. 2 shows the WPP and WG controllers of a Figure 2. Conventional voltage control scheme [6]
conventional voltage control scheme [6]. The WPP
controller calculates the required reactive power (Qref) from
the voltage error through a PI controller and Qref is allocated
to a WGi with a weighting factor, which is inversely
proportional to the active power output of the WG. The
reactive power set-point (QWGiref) for WGi is determined by
(a) WPP controller

Pavg
WGi
Qref u Qref    
PWGi

where Pavg is the average active power for all WGs and PWGi
is the active power output of WGi. This reactive power
allocation is based on the principle that a generator supplying
less active power can supply a greater share of the reactive (b) WG controller
power.
Fig. 3 shows the WPP and WG controllers of a Figure 3. Conventional voltage control scheme [7]
conventional voltage control scheme [7]. The WPP
controller generates the reference voltage signal (uref),
which is obtained using the required Q-V characteristic of a
WPP specified in a grid code (see Fig. 4) and a PI controller. Q max  Q min
kQ
The gain (kQ) in Fig. 3 is set to 6.6. Then, Qref is calculated u max  u min

and compared with the reactive power measured at the PCC.


From the reactive power error, uref is obtained using a PI
controller and sent to WGi.
The WG controller operates at a voltage control mode
and generates the reactive current reference (Idr_ref) using the
characteristic of the reactive current to the voltage as shown
in Fig. 5. An input signal of the WG controller consists of the Figure 4. Q-V characteristic of the UK grid code [5]
respectively, while the linear region of the conventional
scheme [7] is (-0.05, 0.05). This means that WG1 will inject
the maximum reactive power of 0.51pu at the uWGiref of
0.036pu, whereas WG2 will inject the maximum reactive
power of 0.43pu at the uWGiref of 0.034pu. Thus, WG1 gives
more contribution than that of WG2.
In addition, WGs modify the Q-V characteristic
depending on the reactive power capability. Therefore, the
Figure 5. Characteristic of the reactive current to the voltage proposed control scheme is able to adjust the reactive power
injection based on the reactive power capability if the
C. Proposed voltage control scheme operating conditions are changed due to the change of the
wind speed.
Fig. 6 shows the WPP and WG controllers of the
proposed voltage control scheme. The WPP controller
simply calculates uref using a PI controller from the voltage
error signal, which is sent to WGi.
In the WG controller, Qref is generated using the adaptive
Q-V characteristic of a WG. Then, Idr_ref is obtained through
a PI controller, which has input with an error between Qref (a) WPP controller
and QWGi.
We will explain how the adaptive Q-V characteristic of a
WG is obtained. The active power of a WG in a WPP
depends on the wind speed arriving at the WG.
Consequently, WGs have different active power outputs and
thus different reactive power capability. This can be seen in
the typical PQ diagram of a DFIG as shown in Fig. 7, which
was derived by considering the limitations caused by the (b) WG controller
rotor current, the rotor voltage, and the stator current [8]. In
this diagram, the rotor current is the limiting factor for Figure 6. Proposed voltage control scheme
reactive power production, while the stator current is the
limiting factor for reactive power absorption.
Fig. 8 shows some examples of the Q-V characteristic of
a DFIG. The red line represents the Q-V characteristic,
which is set by the reactive power capability (0.33 pu) at the
maximum active power and the range around the rated
voltage is s0.05 pu. The Q-V gain of the WPP controller in
Fig. 3 is set to 6.6 for the voltage control scheme of [7].
However, the proposed scheme uses the Q-V characteristic
for the WG controller, which is determined by the available
maximum and minimum reactive power at a certain active
power output. The Q-V gain of a DFIG (kQi) depends on the
reactive power capability range of the DFIG, and thus the
gains are varied depending on the operating condition of the
DFIG. The adaptive gain kQi is calculated by
Figure 7. Typical PQ diagram of a DFIG

lim it  Qlower lim it 


WGi WGi
Qupper     
kQi
0.1

where QWGiupper limit and QWGilower limit are available maximum


and minimum reactive powers of a WGi at a certain active
power output.
As an example, the two black lines in Fig. 8 represent the
Q-V characteristics of WG1 and WG2, which are operating at
the different wind speeds. The active powers of WG1 and
WG2 are 0.4 pu and 0.6 pu, respectively. The ranges of the
reactive power of WG1 and WG2 are (-0.89pu, 0.51pu) and (-
0.82pu, 0.43pu), respectively. Thus, kQ1 and kQ2 are set to 14
and 12.5, respectively. In addition, the linear slope region of
are (-0.036pu, 0.064pu) and (-0.034pu, 0.066pu), Figure 8. Adaptive Q-V characteristic of a WG controller
III. WAKE SPEED CALCULATION [9] communication time between WPP controller and WG
WGs in a WPP generate electricity by extracting the controller, the reactive power support for restoration of the
kinetic energy in the wind. Thus, the upstream WGs will PCC voltages depends on the DFIG controllers. Immediately
impact the wind speed at other WGs further downstream. after the fault occurrence, the proposed and conventional [7]
This shadowing effect is called as the wake effect. schemes supplied higher reactive powers than conventional
In order to obtain the wake speeds in a WPP, the scheme [6]. This is because both schemes employ the
cumulative impact of multiple shadowing and the effect of voltage control mode for a WG, which reacts to the large
the wind direction are considered in this paper. The resultant voltage variations, while the reactive power control mode is
wind speed of a WGj, vj, can be obtained by used in the conventional scheme [6].
After the fault clearance, the voltage recovery to the
nominal value depends on the WPP controller. For the
  
 u v x  v
n
0 2 conventional scheme [6], it takes longer time to recover the
vj v 0j  i
2
wk ij j
i 1
PCC voltage to the nominal value than the other two
iz j schemes. This is because the weighting factors depend on
active power of a WG and the active power is fluctuated
where vj0 is the incoming wind speed at WGj without any after the fault clearance. In addition, the proposed scheme
shadowing, xkj is the radial distance between WGk and WGj, successfully restores the voltage to the nominal value within
vwk(xkj) is the speed of the wind approaching WGj from the a short time, while the conventional scheme [7] does not.
shadowing WGk, Ek is the ratio of an area of WGj under the The WPP controller makes uref directly through PI
shadow of WGk to its total area, and n is the total number of controller, while WPP controller in the conventional scheme
WGs [9]. [7] makes uref through PI controller and Q-V characteristic.
IV. CASE STUDY
In order to investigate the performance of the proposed
scheme, a model system of Fig. 9, which includes a WPP
and the grid, was chosen. The WPP consists of 20 units of a
5 MW DFIG. Four DFIGs are connected to each feeder
through 2.3kV/33kV transformers. Five collector feeders are
connected to the 33/154 kV substation transformer through
the 33 kV submarine cables and then to the PCC through a
10 km long 154 kV submarine cable. The distance between (a) PCC voltages
two neighboring DFIGs is set to 1 km. The short circuit ratio
of the WPP is set to 6.
The proposed scheme was tested and compared with the
conventional voltage schemes [6] and [7] for three cases. The
WPP controller sends a reference signal to the WG controller
at a rate of 0.1 s. In order to take the wake effect into
account, the free wind speed and direction are set to 12 m/s
and 0 deg, respectively. The PCC voltages and reactive
powers are obtained for the verification.

(b) PCC reactive powers


Figure 10. Results for case 1

B. Case 2: 60 MVAr reactive load connection at 7 s


Fig. 11 shows the PCC voltages and reactive powers for
case 2. At 7 s, the reactive load is connected to a grid. The
PCC voltage is successfully recovered to the nominal value
faster than other algorithms when the proposed algorithm is
used. Because the reactive power injection of a DFIG in
proposed scheme is based on adaptive Q-V characteristic of
each DFIG, which enable the DFIGs to inject reactive power
proportional to the slopes of the Q-V characteristic as shown
Figure 9. Model system in Fig. 8 and the narrow slope control ranges enable the
DFIG to give more contribution to the voltage control. On
A. Case 1: Grid fault at 7 s for 300 ms the other hand, the conventional scheme [6] allocates the
Fig. 10 presents the results for case 1. At 7 s, a grid fault reactive set-points for DFIGs inversely proportional to the
occurs and lasts for 300 ms. At the fault instance, the PCC active power outputs of each DFIG. This allocation
voltages are dropped to 0.5 pu. During the fault, due to the algorithm causes the overshoot in the voltage because of
assignment of excessive reactive power set-points to DFIGs. V. CONCLUSION
In the conventional scheme [7], DFIG controllers use the This paper proposes a voltage control of a DFIG-based
fixed characteristic gain of the reactive current to the voltage WPP using the adaptive Q-V characteristic of a DFIG. In the
as 2, and thus the DFIGs in the WPP inject the same amount proposed scheme, the WPP controller calculates voltage
of reactive power. error signal using a PI controller and sends it to each DFIG.
C. Case 3: 35 MVAr motor load tripping at 7 s From the voltage error signal, the DFIG injects the reactive
Fig. 12 shows the results for case 3. At 7 s, the motor power using its adaptive Q-V characteristic, which depends
load, which consumes 35 MVAr, is disconnected to a grid as on the available reactive power.
a disturbance. Note that the PCC voltage did not converge to The simulation results demonstrate that the DFIG-based
the nominal value before the disturbance occurs in the case WPP using the adaptive Q-V characteristic gains injects
of the conventional scheme [7] is used. The proposed more reactive power for the large and small disturbances.
scheme recovers the voltage faster than [6]. This is because The adaptive Q-V characteristic enables DFIGs to react to
the proposed scheme uses voltage control mode for a DFIG the voltage deviation fast with the increased control gain and
controller and the adaptive Q-V characteristic of a DFIG. narrowed linear control range.
ACKNOWLEDGMENT
This work was supported by the National Research
Foundation of Korea (NRF) grant funded by the Korea
government (MSIP) (2010-0028509).
REFERENCES
[1] Global Wind Energy Council, Global wind energy outlook 2010.
[2] F. Blaabjerg and Z. Chen, Power electronics for modern wind
turbines, 1st ed. Seattle, WA: Morgan & Claypool, 2006.
(a) PCC voltages [3] M. Liserre, R. Cardenas, M. Molinas, and J. Rodriguez, Overview of
multi-MW wind turbines and wind parks, IEEE Trans. Industrial
Electronics, vol. 58, no. 4, April 2011, pp. 10811095.
[4] E.ON Netz GmbH, Grid Code: High and Extra High Voltage,
August 2006.
[5] National Grid, The Grid CodeIssue 4, Revision 5, 31st December,
2010.
[6] M. E. Moursi, G. Joos, and C. Abbey, A secondary voltage control
strategy for transmission level interconnection of wind generation,
IEEE Trans. Power Electron., vol. 23, no. 3, pp. 11781190, May
2008.
[7] J. Fortmann and I. Erlich, Use of a deadband in reactive power
control requirements for wind turbines in European grid code, 11th
(b) PCC reactive powers
international workshop on large-scale integration of wind power into
Figure 11. Results for case 2 power systems, Lisbon, Portugal, 1315 November 2012.
[8] T. Lund, P. Sorensen, and J. Eek, Reactive power capability of a
wind turbine with doubly fed induction generator, Wind energy, vol.
10, pp. 379394, 2007.
[9] F. Koch, M. Gresch, F. Shewarega, I. Erlich, and U. Bachmann
Consideration of wind farm wake effect in power system dynamic
simulation, Power Tech, Russia, June 2730, 2005.

(a) PCC voltages

(b) PCC reactive powers


Figure 12. Results for case 3

S-ar putea să vă placă și