You are on page 1of 6

2SWLPL]DWLRQRI%DVH2SHUDWLRQ3RLQWVRI

07'&*ULGIRU,PSURYLQJ7UDQVLWLRQ6PRRWK

Saddam Aziz, Hui Jiang Jian-Chun Peng, Jia-Qi Ruan, Huai-Zhi Wang
College of optoelectronics Engineering College of Mechatronics and Control Engineering
Shenzhen University Shenzhen University
Shenzhen, China Shenzhen, China
saddamaziz@szu.edu.cn wanghz@szu.edu.cn

Abstract—A new approach to determine the operating point the selection of BOP influences not only the precision of power
of multiterminal dc (MTDC) network for improving transition balance but the transition smoothness of operation state of the
smooth, based on an optimal power flow OPF model. Aim of the MTDC grid. However, no report on the selection of BOP has
method is to reduce transition distance of MTDC network. taken into account the transition smoothness issue till now. In
Therefore, in the proposed method, distance between the BOPs of
a system at different load level ratios is deployed as an objective
fact, there are two main methods for determining the BOP of
function. Operating points are being determined by the output of MTDC grids: sensitivity-based method [17] and cost-
the OPF model we proposed in this paper. The proposed method minimization-based OPF methods [18]. They have their own
is programmed in MATLAB and implemented on an eight-bus inherent pros and cons.
test power system. Compared to previous works, results show the The sensitivity-based method for determining BOP of
validity of proposed approach and ability to improve the MTDC grids consists of chiefly four steps. Firstly, one of the
transition smooth in a MTDC network. Voltage Source Converters (VSC) is selected as a voltage
regulator, who’s DC terminal or DC bus is denoted by the
Keywords—MW balance equation; voltage differnce; cost-
biggest bus number n for convenience, is viewed as the slack
minimization-based OPF(CMOPF) ; base operation point (BOP)
bus. And an initial voltage value Vn 0 of the slack bus n is pre-
I. INTRODUCTION specified for power flow and sensitivity calculation. Secondly,
one of the VSC (not including the one used for voltage
Nowadays, development and usage of renewable energy
regulator) is assumed to be lost (N-1 security), its power will
resources to meet the ever-increasing electric power demand
reach zero and will be compensated by the voltage regulator.
and to cut down the consumption of fossil fuels has been
accelerated [1-2]. These reasons are precursor to offshore wind Power flow and the sensitivity of bus voltage to Vn are
farm rapid growth, especially in the Yellow Sea [3], because of calculated. Based on the sensitivity, the allowable range of
vast offshore wind energy feasibility. Grid integration Vn 0 only considering the lost VSC is obtained. Likewise, the
problems are caused by massive wind farms in the Yellow Sea
[4-6] for instance the requirement of long distance sub-sea allowable range of Vn 0 considering one of the rest VSCs is
power transmission and handling the changings of wind power
obtained. Thirdly, the common allowable range of Vn is found.
in the power grid. These issues let AC power transmission
mode very difficult in voltage control. However, they can be The mean value of the lower and upper limits of this common
properly handle by using multi-terminal DC (MTDC) grid [7– allowable range plus Vn 0 produces the slack bus voltage
9]. As a result, the operation technology of MTDC grids has
attracted many researchers [10-14]. reference value Vref . Finally, power flow is run again to give
Different from AC power systems, the power balance all other bus voltages, which define a BOP and will be taken as
control of MTDC systems is a challenge problem since there is the voltage reference values of P-V controllers of individual
no a frequency-like variable can be used to control the system terminals, respectively. Mathematically, sensitivity-based
power balance [15],16]. To achieve grid power balance and at method works accurately only in a small enough area around
the same time guarantee good bus voltage profile, the reference the BOP where the sensitivity is computed. Power system load
voltage values of P-V controllers of individual terminals have level, however, always changes continuously and more often
to be optimized frequently (such as each minute). The set of this change is big enough. As a result, the BOP determined by
reference voltage values of the MTDC grid define an operation sensitivity-based method is inaccurate and the method does not
state of it and thus called a base operation point (BOP). In fact, take into account the transition smoothness issue.
This work was jointly supported by National Science Foundations of
The cost-minimization-based OPF (CMOPF) methods for
China (51707123, 51477104), Natural Science Foundations of Guangdong determining BOP of MTDC grids give different results
Province (2017A030310061, 2016A030313041), and the Foundations of
Shenzhen Science and Technology Committee (JCYJ20170302153607971).

‹,(((
depending on the objective function selected. Usually, grid between different states of the grid will be smoothest. So, we
transmission loss (operation cost) or generation cost is selected should minimize f in determining BOP.
as the objective function to be minimized [19], [20] and [21]. The state of the MTDC grid is governed by circuit laws. So,
According to electric circuit theory, when objective function is the determination of the set of P-V droop control reference
system transmission losses, each attempt to limit the voltage values or the BOP should satisfy all bus MW balance
transmission losses will have impact on operational voltages. equations. In addition, limited to device rated voltage and DC
Consequently, one or more than one operational voltages of the line allowable maximum current, voltage and current related
buses will touch the maximum limit [7]. This is also true for security constraints should be considered in the determination
the generation cost based OPF. As a result, when a bus voltage of the BOP.
of the system reaches its upper or lower limit of the boundary, m 1 n
the operation point can easily violate the boundary if any
disturbance occurs. Therefore, the security of the MTDC
min f ¦¦ (V
k 1 j 1
kj  VK 1 j )2 (2)
system with BOP determined by the CMOPF methods is not n
robust in security. In addition, just like the sensitivity-based s.t.¦Vki (Vki  Vkj ) gij  PGki  PDki 0 (3)
method, the CMOPF methods do not cope with transition j 1
smoothness problem.
In order to compensate the defects of existing methods this  I ij max d gij *(Vi  V j ) d I ijm ax (4)
paper proposes a new model to determine the BOP of a 0.95 d Vki d 1.05 (5)
MTDC. In the model, a transition distance expression is built
and minimized under the constraints of bus power balance
equations and variable limits. The rest of the paper is structured k {1, 2,3,..., m}& i, j {1, 2..., n}
as follows. Section II gives the new mathematical model for Power injection at the load buses is taken as equality
optimizing the BOP of MTDC Grid. Section III presents and constraints. So, equality constraints are in the form of Bus
analyzes the simulation results and of the proposed method, Megawatt (MW) balance equation. In the equation (3), the
including comparison with the CMOPF method. At last, equality constraints of the optimization problem consist of
conclusions are made in Section IV. power injected at each node of the system and the load level of
the system. The equation (4) shows that the power flow at
II. MATHEMATICAL MODELING
each node, where g ij is the conductance of the all branches
The set of P-V droop control reference voltage values of the
MTDC grid define an operation state “a BOP” of the grid. of the network and Vi is voltage at node i and V j is voltage
According to circuit theory, “a BOP” can be described by a bus
voltage vector. When the grid transfers from one BOP to at node j . Likewise, Vi  V j is difference of voltage
another, the distance between the two BOPs reflects the magnitude at node i to node j .
performance of the transition. The bigger the distance the DC overvoltage can influence the security of system.
bigger the step to be transited by all P-V droop controllers, Therefore, bound constraints are considered in order to keep
therefore the worse of smoothness of the grid state transition the bus voltage in limit and secure system BOP. Literature
between the two BOPs. As a result, we take this distance as the
allows r5 % voltage deviation for power transmission, so the
objective function in determining the BOP to let the transition
of the grid states smoother. boundary limit is between 0.95 and 1.05 p.u.
Consider a MTDC grid with n bus numbers (assume the Where I ijm ax and  I ij max in (4) represents the capacity of
DC bus number of the voltage regulator is n ) and m load maximum and maximum current that could pass through each
levers. In order to make the transition of the MTDC grid states branch of the network. And gij* (Vi  V j ) shows in (4) the
smoother throughout the m load levers, we build the following
function. actual current for all the branches and we applied the limits in
m 1 n (4) for the actual current and the limits are I ijm ax and  I ij max
f ¦¦ (V
k 1 j 1
kj  VK 1 j )2 (1)
where I ijm ax and  I ij max represent the maximum and
Notation f denotes voltage difference, and k represents the minimum allowed current.
So, the new model for determining the set of P-V droop
number of load level ratios and Vkj denotes the voltage control reference voltage values or the BOP is based on bus
amplitude at node j with respect to initial load level ratio and MW balance equations. Obviously, model (2)-(4) is a typical
nonlinear optimization model. Such kind of model can be
VK 1 j denotes the voltage amplitude at node j with respect to solved by MATLABs YALMIP Toolbox.
very next load level ratio respectively shown in (1). It is
III. SIMULATIONS AND RESULTS
obvious if f achieves the minimum value, the transition
M-file has been developed in MATLAB® R2015b software
for the proposed and CMOPF method. Both methods are tested
on the DSC3 part of the CIGRE B4 DC grid test system [22]. TABLE 4 LOAD LEVEL RATIOS
Simulations are done using a PC based on AMD A10-7850K
LOAD LEVEL RATIO LOAD LEVEL RATIO
Radeon R7, 12 Compute Cores 4C+8G, 3.7 GHz, 8 GB RAM NO.
with Windows 10 64-bit operating system. The full OPF is 1 1.00
consist of quadratic objective function and nonlinear 2 1.20
constraints. So, optimization problem is solved by using 3 1.05
4 0.80
MATLABs YALMIP Toolbox.
The test system has 8 buses, 8 branch network with DC TABLE 5 LOAD LEVEL RATIO NO. 1
overhead lines (OHL) and DC cables. The entire system
information is given in TABLES. Transmission line data, LOAD LEVEL RATIO =1
Generation data, Load data and Load level ratios are stated in Bus No.
P1 P2 V1 V2
TABLE 1-4, respectively.
1 19.8120 19.8120 1.0097 1.0477
2 06.0000 06.0000 1.0496 1.0496
3 09.9375 09.9375 1.0121 1.0500
4 -03.0000 -03.0000 1.0094 1.0474
5 -06.0000 -06.0000 1.0077 1.0457
6 -09.0000 -09.0000 1.0037 1.0419
7 00.0000 00.0000 1.0010 1.0393
8 -17.4140 -17.4140 0.9963 1.0348

TABLE 6 LOAD LEVEL RATIO NO. 2

LOAD LEVEL RATIO =1.2


Bus No.
P1 P2 V1 V2
1 23.774 23.774 1.0103 1.0472
2 07.200 07.200 1.0127 1.0495
3 11.925 11.925 1.0132 1.0500
Fig. 1 Eight-bus test power system 4 -03.600 -03.600 1.0100 1.0469
5 -07.200 -07.200 1.0079 1.0449
TABLE 1 POWER SYSTEM DATA 6 -10.500 -10.500 1.0032 1.0403
7 00.000 00.000 0.9999 1.0371
TRANSMISSION LINE DATA 8 -20.818 -20.818 0.9943 1.0317

From (i) To (j) I-max (p.u)


rij (p.u)
TABLE 7 LOAD LEVEL RATIO NO.3
1 2 0.00029687 18.12
1 6 0.00035625 28.00 LOAD LEVEL RATIO =1.05
1 7 0.00089063 28.00 Bus No.
2 3 0.00044531 18.12 P1 P2 V1 V2
3 4 0.00029687 18.12 1 20.7742 20.7742 1.0098 1.0475
4 5 0.00029687 18.12 2 06.3000 06.3000 1.0120 1.0496
6 7 0.00035625 28.00 3 10.4334 10.4334 1.0124 1.0500
7 8 0.00026719 28.00 4 -03.1500 -03.1500 1.0096 1.0473
5 -06.3000 -06.3000 1.0077 1.0455
TABLE 2 GENERATION DATA 6 -09.4500 -09.4500 1.0036 1.0415
7 00.0000 00.0000 1.0007 1.0387
Bus No. P(MW) 8 -18.2934 -18.2934 0.9958 1.0340
1 19.8118 TABLE 5-8 are showing the OPF results for all cases
2 06.0000 considering distance between operating points (1) and power
3 09.9375
loss (6) as transmission losses respectively.
4 03.0000
m 1 n

TABLE 3 LOAD DATA min f ¦¦ g


k 1 j 1
ij *(Vi  V j )2 (6)

Bus No. P(MW)


5 6 k {1, 2,3,..., m}& i, j {1, 2..., n}
6 9
Per unit data calculation is based on given below values: To verify the simulation results, the simulation is done at
Sbase = 100MW, Vdcbase = 800 kV. Where DC OHL= +/- different load level ratios. There are four cases based on load
400 kV and DC cable = +/- 400 kV. The MTDC system’s level ratios. In each case, there are two parts. The first part of
branches resistance and maximum current that can flow in all each case is based on the proposed objective function and the
branches of the whole power system is given in TABLE 1. second part is based on transmission losses objective function
(6). The power flow of the system is dictated by equality
constraints in the form of MW balance equation (2) in each case 1,2,3&4 respectively presented on TABLE 5-8. In Fig.2-5
case CMOPF method. The goal here is to examine the the results of Bus voltage and bus number of the proposed and
proposed and CMOPF method for eight-bus test power system. CMOPF method are presented by red color and black colors
It must be noticed, during second part considering CMOPF respectively. Meanwhile, the results of both methods are shown
method one of the bus voltages reached at the upper limit in the curve form.
boundary at different load levels. Bus voltage at branch No.3
can be seen, where its voltage’s values is at the maximum Case:2
1.06
given boundary and all other voltages of the buses that inject Proposed method
the power at branch No.1, No.2, No.4 to branch No.8 are CMOPF method
1.05
within given limits shown in the TABLE 5-8. Here the
security risk in terms of bus voltage at branch 3 can violate in 1.04
any case of disturbance is obvious.

Bus Voltage(pu)
1.03
TABLE 8 LOAD LEVEL RATIO NO 4
1.02
LOAD LEVEL RATIO =0.8
Bus No.
P1 rij P2 V1 V2 1.01
1 15.8494 15.8494 1.0090 1.0481
2 04.8000 04.8000 1.0106 1.0497 1
3 07.9500 07.9500 1.0110 1.0500
4 -02.4000 -02.4000 1.0088 1.048
0.99
5 -04.8000 -04.8000 1.0074 1.0466 1 2 3 4 5 6 7 8
6 -07.2000 -07.2000 1.0043 1.0435 Bus Number
7 00.0000 00.0000 1.0021 1.0414 Fig. 3 Bus voltage curves versus bus number (Case 2)
8 -13.9843 -13.9843 0.9983 1.0378
TABLE 5-8 describes the results obtained when load level
ratios are 1, 1.20, 1.05 and 0.8 for all buses of the system Case:3

respectively. V1 and V2 are the voltages of the buses and P1


1.06
Proposed method
CMOPF method
and P2 are the values of the power injected at each bus used in 1.05

the both (proposed and CMOPF) methods. And the output 1.04
results of the both methods are compared at different load
levels in TABLE 5-8. Then the values of bus voltages are
Bus Voltage(pu)

1.03
shown. However, regarding first case considering the
proposed method all the bus voltages are within the 1.02
recommended boundary. This finding demonstrates the
importance of proposed method in improving the transition 1.01
smooth
1
Case:1
1.06
0.99
Proposed method 1 2 3 4 5 6 7 8
CMOPF method Bus Number
1.05
Fig. 4 Bus voltage curves versus bus number (Case 3)
1.04
Whether a solution is achieving its goal or not, a better way
is to plot bus voltage curve vs bus number of both cases. After
Bus Voltage(pu)

1.03
the results are plotted in Fig.2-5, it can be compared. By
comparison of the results, it can be observed that proposed
1.02
method will keep the better results in terms of no bus voltage
violated the given limits. The transition smooth with respect to
1.01
load level ratios during this simulation are shown in Fig.6 and
1
Fig.7, where proposed method’s results are compared with the
results achieved by CMOPF method. As, it is proven in the
0.99 curve, the transition of proposed method is smoother than
1 2 3 4 5
Bus Number
6 7 8
CMOPF method.
In the proposed method, all the voltages of the buses that
Fig. 2 Bus voltage curves versus bus number (Case 1)
inject the power will keep within the limits and in case any
Fig.2-5 are showing numerical results of the bus voltage on kind of disturbance occurs, the system will be in secure mode
eight bus test power system considering all the cases for the it will also guarantee the transition smooth. On the other hand,
the CMOPF approach has no potential to do so. So, proposed IV. CONCLUSION
method could eliminate the drawbacks associated with the A new methodology allowing the distance between the
CMOPF methods regarding transition smoothness. BOPs as an objective function to be included in the solution is
1.06
Case:4
presented in this Paper. In this method, distance between the
Proposed method BOPs is considered to be minimized. Therefore, we choose
1.05
CMOPF method
this distance as the objective function to determine the BOP.
The mathematical model is proposed for determining the
1.04 BOP. And to determine the set of P-V droop control reference
voltage values or the BOP should satisfy all bus MW balance
Bus Voltage(pu)

1.03 equations. Further, limited to device rated voltage and DC line


allowable maximum current, voltage and current related
1.02 security constraints should be take into account in the
determination of the BOP. Thus, the security of the MTDC
1.01 system is feasible by keeping the bus voltage values in
boundary limits. Consequently, the transition of the grid states
1
will be smoother
The simulation results of the proposed method and CMOPF
0.99
1 2 3 4 5 6 7 8 method, are compared. It is shown that at different load levels,
Bus Number after minimizing the transition distance the system is secure,
Fig. 5 Bus voltage curves versus bus number (Case 4) and transition is smooth

REFERENCES
-3
x 10
5.5
Scenario1 [1] Wang HZ, Wang GB, Li GQ, Peng JC, Liu YT. Deep belief network
5
Scenario2 based deterministic and probabilistic wind speed forecasting approach.
Appl Energy 2016;182(15):80–93.
[2] Wang HZ, Li GQ, Wang GB, Peng JC, Jiang H, Liu YT. Deep learning
4.5 based ensemble approach for probabilistic wind power forecasting. Appl
Energy 2017;188(15):56–70.
4 [3] Trevor M.Letcher, Wind Energy Engineering, Elsevier, May 2017
Load level

[Online]. Available: https://www.elsevier.com/books/wind-energy-


engineering/letcher/978-0-12-809451-8
3.5
[4] L. Xu; B. W. Williams,L.Yao. Multi-terminal DC Transmission Systems
for Connecting Large Offshore Wind Farms . In Proceedings of 2008
3 IEEE PES General Meeting - Conversion and Delivery of Electrical
Energy in the 21st Century, 2008.
2.5 [5] E. Eriksson, P. Halvarsson, D.Wensky, and M. Hausler, “System
Approach on Designing an Offshore Wind power Grid Connection”,
Proc. of the 4th International Workshop on Large-Scale Integration of
2
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 Wind Power and Transmission networks for Offshore Wind Farms,
Transition Distance Sweden, Oct. 2003
[6] S. M. Bolik, “Grid Requirements Challenges for Wind Turbines”, Proc.
Fig. 6 Transition Smoothness curve
of the 4th International Workshop on Large-Scale Integration of Wind
Power and Transmission networks for Offshore Wind Farms, Sweden,
0.96
Scenario1
Oct. 2003.
Scenario2
[7] Catalin Gavrilu, Ignacio Candela, Alvaro Luna, Antonio Gomez-
0.95 Exposito and Pedro Rodriguez “Hierarchical Control of HV-MTDC
Systems With Droop-Based Primary and OPF-Based Secondary,” IEEE
0.94 Trans. On Smard Grid, vol. 6, no. 3, pp.1502-1510, May 2015.
[8] T. M. Haileselassie, “Control, dynamics and operation of multi-terminal
[9] N. R. Chaudhuri and B. Chaudhuri, “Adaptive droop control for
0.93
Load level

effective power sharing in multi-terminal DC (MTDC) grids,” IEEE


Trans. Power Syst., vol. 28, no. 1, pp. 21–29, Feb. 2013.
0.92 [10] M. K. Bucher, R. Wiget, G. Andersson, and C. M. Franck,
“Multiterminal HVDC networks—What is the preferred
0.91
topology?”IEEE Trans. Power Del., vol. 29, no. 1, pp. 406–413, Feb.
2014.
[11] C. Feltes, H. Wrede, F. W. Koch,et al, “Enhanced Fault ride-
0.9 throughmethod for wind farms connected to the grid through VSC-based
HVDCtransmission,”IEEE Trans. on Power Systems, Vol.24, No.3, pp:
0.89 1537-1546, August 2009.
1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 [12] Lie Xu, Liangzhong Yao, Chistian Sasse, “Grid integration of
Transition Distance
largeDFIG-based wind farms using VSC transmission,” IEEE Trans. On
Fig. 7 Transition Smoothness curve Power Systems, Vol.22, No.3, pp:976-984, August 2007.
[13] Junhong Wu, Qian Ai, “Research on multiterminal VSC-HVDC system
for wind-farms,” Power System Technology, Vol.33, No.4, pp: 22-27,
October ,2009.
[14] Xiaodong Zhu, Keliang Zhou, Ming Cheng, et al, “Topologies and
control of VSC-HVDC systems for grid-connection of large-scale
offshore wind farms,” Power System Technology, Vol.33, No,18, pp:17-
24, October 2009.
[15] Chengyong Zhao, Dongliang Hu, Guangkai Li, et al, “Control strategy
for integration of wind farms by multi-terminal VSC-HVDC,” Power
System Technology, Vol.33, No.17, pp:135-140, September 2009.
[16] B. K. Johnson, R. H. Lasseter, F. L. Alvarado, R. Adapa. Expandable
Multiterminal DC Systems Based on Voltage Droop. IEEE Transactions
on Power Delivery, 8(4):1926̢1932, October 1993.
[17] Weixing Lu, and Boon Teck Ooi. “DC voltage limit compliance in
voltage-source converter based multi-terminal HVDC,” in IEEE PES
General Meeting,” vol. 2, pp. 1322-1327, Jun. 12-16, 2005.
[18] Hetzer, J., Yu, D. C. and Bhattarai, K. An economic dispatch model
incorporating wind power. IEEE Trans. On Energy Conversion. 2008,
23(2):603-611.
[19] M.Aragüés, A. Egea- Àlvarez, O.Gomis-Bellmun, and A. Sumper,
"Optimum voltage control for loss minimization in HVDC multi-
terminal transmission systems for large offshore wind farms," Electric
Power Systems Research, vol. 89, pp. 54-63,2012
[20] R. T. Pinto, P. Bauer, S. F. Rodrigues, E. J. Wiggelinkhuizen, J. Pierik,
and B. Ferreira, "A Novel Distributed Direct-Voltage Control Strategy
for Grid Integration of Offshore Wind Energy Systems Through MTDC
Network," Industrial Electronics, IEEE Transactions on, vol. 60, pp.
2429-2441, 2013
[21] Liang Xie, Hsiao-Dong Chiang and Shao-Hua Li, “Optimal Power Flow
Calculation of Power System with Wind Farms,” in Power and Energy
Society General Meeting, 2011 IEEE
[22] (Sep. 19, 2013). CIGRE B4 Working Group. [Online]. Available:
http:/b4.cigre.org/publications

You might also like