Escolar Documentos
Profissional Documentos
Cultura Documentos
1, MARCH 2016
AbstractA hierarchical control scheme is proposed for op- reasons [7], including the fact that the coordinated control
timal power flow control to minimize loss in a hybrid multi- mechanism relies heavily on HVDCs communication system,
terminal HVDC (hybrid-MTDC) transmission system. In this thus making it difficult to apply decentralized autonomous
scheme, the lower level is the droop control, which enables
fast response to power fluctuation and ensures a stable DC control (DAC). MTDC development along with HVDC tech-
voltage, and the upper level is power flow optimization control, nology has seen increased importance in China. For example,
which minimizes the losses during the operation of hybrid-MTDC China has built its first two VSC-based MTDC projects in
and solves the contradiction between minimizing losses and Nanao and Zhoushan [8].
preventing commutation failure. A 6-terminal hybrid-MTDC is Since DC systems have a rapid transit process, application
also designed and simulated in PSCAD according to the potential
demand of power transmission and wind farms integration in of a central controller in MTDC is challenging. As a result,
China to verify the proposed control strategy. First, the steady application of DAC becomes an inevitable choice for MTDCs.
state analysis is conducted and then compared with simulation Droop control, which is one type of DAC, is currently a
results. The analysis shows that the proposed control scheme popular technique. The parameter setting method of droop
achieves the desired minimum losses while at the same time control for VSC-based MTDC is studied in [9], [10], and
satisfying system constraints. The proposed control scheme also
guarantees that the hybrid-MTDC not only has a good dynamic the power flow of VSC-based MTDC under droop control
response, but also remains stable during communication failure. is analyzed in [11], [12]. Although the MTDC can remain
stable using droop control, it does not guarantee optimal power
flow (OPF) for loss minimization under arbitrary power inputs.
Index TermsDroop control, hierarchical control, hybrid This causes notable DC voltage deviations particularly during
multi-terminal HVDC, loss minimization, optimal power flow.
marked changes in injected power. This problem has been
addressed in [13][16] in which the optimal power flow control
for loss minimization by regulating reference value of droop
control during VSC-based MTDC operation is studied. It is fair
I. I NTRODUCTION
to say though that existing control strategies cannot be applied
needs, wind farm integration, and use of the existing LCC- Global information
HVDC in China. First, the proposed hierarchical control strat-
egy is described in Section II; then, a steady state analysis is k p ( Pref P ) +
carried out in Section III and compared with simulation results. (U dcref U dc ) = 0
Optimization
This is followed by testing of dynamic response in situations algorithm
under conditions such as wind fluctuation and communication
failure in Section IV.
Loss minimization
Droop reference of
II. P ROPOSED H IERARCHICAL C ONTROL S CHEME
each terminal
A. Hybrid-MTDC System Description
The hybrid-MTDC consists of both VSC and LCC, thus
presenting some new characteristics. VSC rectifiers in an Optimization control Droop control
Thus the objective function, namely, the total losses of hybrid- Injected power;
MTDC, can be expressed as constraints
LCC6 LCC3 G12 G12 0 0 0 I12,max
Line 63
G12 G12 0 0 0 I12,max
G14
0 0 G14 0
I14,max
G14 0 0 G14 0 I14,max
Cable 34
G25
0 G25 0 0 I25,max
0 G25 0 0 G25
I25,max
A =
0 0 G34 G34 0 , b = I34,max ,
VSC1 VSC4
G34
0 0 G34 0
I
Cable 14 34,max
0
0 G34 G34 0
0
0 0 G34 G34 0 2 I63
WF1 0 0 0 G45 G45 I45,max
Cable 45
Cable 12
TABLE II
R ATED PARAMETERS OF H YBRID -MTDC participate in optimization control.
When considering the constant control of LCC3 , it is clear
Category Parameters Values Unit that the DC voltage must decrease when the injected power is
Power 1,000 MW
DC voltage 500 kV
too large. The two sets of droop control references for lower
DC current 2 kA power and larger power are compared in Table V. All the three
LCC
AC voltage
345 (sending ends)
kV converters participate in the optimization control. In situation
230 (receiving ends)
Short circuit ratio 2.5
1, the power of two wind farms is 150 MW and the current of
WF/VSC power 300 MW LCC6 is 1 kA. In situation 2, the power values of two wind
WF voltage 6 kV farms increase to 200 MW and 250 MW, respectively, and the
VSC AC voltage 220 kV current of LCC6 increases to 1.6 kA. As can be seen from
Transformer ratio 230/300
Shunt DC capacitor 150 F Table V, the DC voltage is lowered when the injected power
is larger. Although the power loss rate is relatively high, it is
TABLE III more important to prevent commutation failure by lowering
T RANSMISSION L INE PARAMETERS OF H YBRID -MTDC
DC voltage.
Lines Resistance () TABLE V
Cable 12 1 C OMPARISON B ETWEEN VOLTAGES WITH D IFFERENT P OWER
Cable 14 2
Cable 25 1.6
Cable 34 1.6 Situation 1 Situation 2
Terminal
Cable 45 0.8 Udc (kV) Pdc (MW) Udc (kV) Pdc (MW)
Line 63 5 1 524.991 150 501.731 200
2 525.000 150 501.762 250
3 523.657 767.38 500.275 987.13
4 524.402 12.51 500.872 116.77
in constant power/current control mode and the current order 5 524.557 43.09 501.014 145.58
6 531.657 531.66 508.275 813.24
of LCC3 is the same with LCC6 . In the second situation, Loss Rate (%) 1.33 1.51
LCC3 and VSC5 participate in optimization control; VSC4 is
in constant power control mode. In the third situation, LCC3 ,
VSC4 and VSC5 all participate in optimization control. Above
all the three situations, the power values of WF1 and WF2 IV. S IMULATION R ESULTS
are 150 MW and 200 MW respectively; the current order for Although we are able to obtain the steady state results,
LCC6 is 1.6 kA, the power factor of VSC4 and VSC5 is 1. simulation is still needed to verify these results and compare
them with conventional droop control strategies. The dynamic
TABLE IV
C OMPARISON A MONG O PTIMIZATION R ESULTS IN D IFFERENT
behaviors of the hybrid-MTDC under the proposed hierarchi-
S ITUATIONS cal control also need to be verified in situations such as wind
power fluctuation and communication failure.
Situation 1 Situation 2 Situation 3 In this paper, we present the results of a most complicated
Terminal Udc Pdc Udc Pdc Udc Pdc situation where all three inverter stations are involved in
(kV) (MW) (kV) (MW) (kV) (MW)
1 518.111 150 511.452 150 503.676 150 optimization control and simulations. The V-I characteristic
2 518.147 200 511.511 200 503.711 200 of both two VSC inverters is
3 517.460 827.94 510.507 893.89 502.490 967.33
4 517.460 250 510.748 250 503.010 80.63 kp (Pref P ) + (Udcref Udc ) = 0 (20)
5 517.586 99.57 510.979 22.39 503.132 105.41
6 525.460 840.74 518.507 829.61 510.490 816.78 where, kp = 0.1667, Pref represents power reference whose
Total
power loss 1.66 1.62 1.49 limit is 0300 MW. Udcref represents DC voltage reference
rate (%) whose limit is 475525 kV.
The V-I characteristic of LCC3 is
As can be seen from Table IV, hybrid-MTDC losses will
I = Iref . (21)
decrease as more converters take part in the optimization
control. In situation 1, only VSC5 can be operated by the The upper optimization control computes droop references
optimization control. There is no current in cable 34 and the of each terminal only when the DC voltage deviation exceeds
current in LCC3 is the smallest among the three situations, so a certain value. The events and the corresponding time are
the DC voltage is the largest. In situation 2, although the DC shown in Table VI.
voltage becomes lower, the total power loss rate still decreases The simulation results are shown in Fig. 46. The base
because the power of wind farms partly flows through LCC3 power for LCC and VSC is 1000 MW and 300 MW, respec-
whose loss rate is much less than VSC. In situation 3, all tively, and the base DC voltage is 500 kV. The direction of
three converters participate in the optimization control, so the power flowing into the grid is positive. The hybrid-MTDC
power is distributed more reasonably. As a consequence, the system operates stably at t = 2 s. Power at both wind farms
total power loss rate continues to decrease even if the DC is 0.5 p.u., and the LCC6 current is 0.5 p.u. At this moment,
voltage becomes smaller. By analyzing the data in Table IV, the power of LCC3 is mainly delivered out through LCC3 , and
we see that the power loss rate will decrease as more inverters only a small amount of power is delivered out through VSC5 .
HAN et al.: HIERARCHICAL OPTIMAL POWER FLOW CONTROL FOR LOSS MINIMIZATION IN HYBRID MULTI-TERMINAL HVDC TRANSMISSION SYSTEM 45
TABLE VI
S IMULATION E VENTS AND THE C ORRESPONDING M OMENTS remains high, and the power loss rate is relatively low.
At t = 5 s, the current at LCC6 increased to 0.8 p.u.
Events Time (s) The upper optimization control decreased the DC voltage
Stably operating 2
WF1 s power increased to 200 MW 3 before the current from LCC3 current approached 1.0 p.u.
WF2 s power increased to 250 MW 4 in order to prevent commutation failure. The VSC inverters
LCC6 s current increased to 1.6 kA 5 thus undertake more power, but the power loss rate goes up
Communication failure 6
Communication restoration 7 to 1.51%.
The communication failure occurs at t = 6 s, at which
point the inverters involved in optimization control switch their
0.8 WF1
WF2
droop references to the predetermined values at a reasonable
0.6
LCC3 ramping rate to ensure DC voltage and power within the limits.
0.4 VSC4
The predetermined values for VSC are set to 300 kV and 500
DC Power (p.u.)
0.2 VSC5
LCC6 MW in order to prevent commutation failure and exceeding
0
system constraints. The current order for LCC3 no longer
0.2
contains the part from optimization control, becoming the
0.4
same with LCC6 . At this moment, the DC voltage decreases
0.6
further, and the power of wind farms is delivered out through
0.8
VSCs, so the power loss rate increases to 1.64%. At t = 7
1
2 3 4 5 6 7 8 9 s, the communication is resumed, the converters can receive
Time (s) the optimized droop references again and participate in the
optimization control; as a result, the power loss rate drops to
Fig. 4. Power flow of each converter.
1.51%.
1.07
As can be seen from Fig. 6, the practice optimized power
VSC1
1.06 VSC2 loss rate is not only consistent with the steady state result, but
1.05
LCC3 also much lower than the power loss rate without optimization.
VSC4
DC Voltage (p.u.)
1.04 VSC5 It is shown that the optimization control is accurate and plays a
1.03
LCC6 crucial role in loss minimization, achieving the desired effect.
1.02 As can be seen from Fig. 4 and Fig. 5, the proposed hierarchi-
1.01 cal control exhibits good dynamic behavior for hybrid-MTDC
1 in situations that include power fluctuation and communication
0.99 failure. During the operation of hybrid-MTDC, the system
0.98 constraints are satisfied and no commutation failure occurs.
2 3 4 5 6 7 8 9 Therefore, the proposed hierarchical control in this paper is
Time (s)
suitable for the hybrid-MTDC.
Fig. 5. DC voltage of each converter.
2
Optimized value V. C ONCLUSION
1.9 Theoretical value
Un-optimized value
This paper presents a hierarchical control scheme to operate
Power Loss Rate (%)
1.8
a hybrid-MTDC system with wind farms based on an opti-
1.7
mization algorithm combined with droop control. The upper
1.6 optimization control computes the appropriate references for
1.5
droop control in order to ensure optimal power flow for loss
minimization. The lower droop control guarantees a stable DC
1.4 voltage during frequent but small power fluctuation. By ana-
1.3 lyzing steady state results, power loss rate is seen as decreasing
2 3 4 5 6 7 8 9
Time (s) as more inverters participate in optimization control. When
results from high power and low power are compared, we
Fig. 6. Loss rate of the system. see that the proposed hierarchical control scheme is capable
of lowering the DC voltage of hybrid-MTDC in a timely
manner to prevent commutation failure. The simulation results
The power in WF1 increased to 200 MW at t = 3 s and in PSCAD show that the proposed hierarchical control scheme
that of WF2 increased to 250 MW at t = 4 s. The increased can coordinate inverters in hybrid-MTDC to achieve minimum
power is then allocated among three inverters according to the loss, which is consistent with steady state results. Dynamic
references generated by the optimization control. At t = 25 s, behaviors in power changes and communication failure are
the current flowing through LCC3 is relatively small, so the also favorable, and the system constraints are within limits, as
constant control is not enabled. As a result, the DC voltage shown in simulation results.
46 CSEE JOURNAL OF POWER AND ENERGY SYSTEMS, VOL. 2, NO. 1, MARCH 2016