You are on page 1of 4

1

Forward/Backward Sweep Method Based on Map Structure for Power Flow Calculation of Distribution System
Xu Jingzhou, and Chen Xiao

Abstract Power flow calculation is very important to distribution system management, and more and more attention are paid to calculation speed and accuracy due to the booming smart grids which allow interconnection of distributed generation of different kinds. A new forward/backward sweep method based on map structure is presented, which can calculate power flow of all complex radial distribution systems. Special coding of nodes, such as hierarchical coding, is not necessary, and results can be obtained directly through a global search of line map. Test shows that the method is timesaving and insensitive to the system scale, which helps the popularity of smart grids.
IndexTerm Distribution system; forward/backward sweep method; power flow calculation

coding of forward/backward sweep method. Tests show that this method is timesaving and insensitive to the system scale, which can get satisfying results and can be easily applied to practice. II. Problem Description As shown in Figure 1, most distribution systems are radial, which leads to the difference of load flow calculation from the other parts of power system.
i 1 k

Figure 1 Radial distribution system connected with distributed generation

I. Introduction Power flow calculation is very important to distribution system analysis including its planning, operational analysis, state estimation and network reconfiguration, etc. Because of different characteristic features of distribution networks, such as radial structure and high R/X ratio, distribution system load flow analysis differs significantly from transmission systems. Accordingly, a number of methods are specially designed for the solution of the power flow problem on radial distribution systems, most of which are based on the forward/backward sweep method [1]-[10]. Although these methods are reliably convergent, they must comply with some special rules like hierarchical coding, and sometimes Jacobian matrix or additional zero impedance branches are needed, which makes programming and applying very difficult. In this paper, a new system expression based on map structure is provided, which simplifies the calculation process and nodal
Xu Jingzhou works at Nanjing Power Engineering Design Company, Ltd, Nanjing, Jiangsu 210009 China (e-mail: chefxu@hotmail.com). Chen xiao works at Nanjing Power Supply Company, Nanjing, Jiangsu Nanjing, Jiangsu, 210008, (email:lettuce81@163.com).

Firstly, distribution network matrices are generally ill-conditioned, that is, bus admittance matrix is very sparse because branches are always less than nodes in radial distribution system. Secondly, due to high R/X ratio, fast decoupling method can not be applied. Thirdly, since power of the line connected with the endpoint node is just the nodal power, power loss and voltage loss can be resolved directly if the endpoint node voltage is given, so sweep-based method can be adopted to calculate system state variables in sequence. Branch current and branch power can be both used as iterative variables, according to which forward/backward sweep method is a little deferent [10], and branch power is preferred in this paper. III. Proposed Methodology Conventional forward/backward sweep methods are not easily applied and popularized due to weak correlation of data which leads to complex computational logic. Aiming at this problem, map structure is adopted in this paper, based on which coding and calculating becomes very easy for each iterative step just needs to traverse all lines once. A Data processing

Map container allocates each node a unique index having no use for hierarchical classification of network, which endows this storage measure with high searching efficiency. Firstly, let us consider the distribution system as
Items BB System reference voltage Lines front-end node index Access Point of load or DG Parameters included System reference rate Lines tailend node index Load real power Is parameter normalized

U li _ f = 1.0 U li _ b = 1.0 Pli _ f = 0 Qli _ b = 0


1 Step 2. Equation 2 shows the forward sweep for front-end power flow of lines, which is based on line map.

LI

Resistance of line

Impedance of line

LP

Load reactive power

Pli _ f = Pli _ b + Pli Qli _ f = Qli _ b + Qli


2 Where li _ f and li _ f stand for real and reactive power from the front-end of line i respectively, and as the same, li _ b and li _ b stand for real power and reactive power integrated to the tail-end of the line, summing up front-end real and reactive power of all lines connected with
li and end-point of line i . Moreover, stand for real and reactive line loss. So

given in Figure 1, and input system elements data as shown in Table 1.


Table 1 Data inputting formula

Where BB, LI and LP stand for data of bus, line and load respectively. Since conventional calculation methods are required to distinguish between main lines and branch lines, nodal data need to be pre-processed specially in order to get hierachical data structure. As shown in Table 1, method proposed in this paper indexes all nodes without taking consideration of the deference of lines. Secondly, component connectivity is created according to inputting data, forming component map including bus map, line map, load map and node map. Each element map records all the homogeneous elements with a unique index and data information, and besides, node map is adopted to stores the connectivity information of all nodes. In this case, the whole system can be configured through traversing the line map. During backward sweep, tail-end power of a line, i.e. line i, can be figured up through searching front-end node of the line, power of tail-end node can be figured up just through traversing all lines front-end node . B Calculation procedure The method proposed in this paper neednt distinguish main lines from branch lines, which can be demonstrated by the following calculation procedure. Step 1. Parameters need initializing if elements data havent been referred, and at the same time, bus voltage with voltage and power of lines both ends is assumed as follow.

Qli

Pli _ b Qli _ b Pli Qli can be written as and , ,


Pli _ b = Plj _ f + Pldi Qli _ b = Qlj _ f + Qldi
ip j ip j

ip j

3 Where i p j stands for all lines connected with the endpoint of line i.

Pli = Qli =
4

( Pli _ b + Qli _ b ) * R
2 2

U li _ b
2

2 2

( Pli _ b + Qli _ b ) * X U li _ b
2

Step 3. Backward sweep for end-point voltage of line is shown as equation 5. Because the quadrature component of voltage is too little to reckon in, direct-axis component is considered only.

li _ b

=U

li _ f

P R+Q X li _ f li _ f U li _ f

5 The end-point voltage obtained is assigned to the front-end voltage of lines which are connected with the end-point of line i.

U lj _ f = U li _ b

Figure 3 69-Bus system

ip j

Parts of the calculation results are shown in Table 2.


Table 2 Calculation results of voltage and line loss Nodal Branch Voltage Line loss number number
( k +1)

Step 4. Convergence criterion is written as equation 6.


Max( Pli _ f
( k +1)

Pli _ f

(k )

, Qli _ f

( k +1)

Qli _ f

(k )

, U li _ b

U li _ b

(k )

)<

1 2

0.99997 0.99993 0.99984 0.97857 0.97744 0.97134 0.96818 0.95632 0.99993 0.99895 0.99979 0.99415 0.97853 0.97465 0.90915 0.97128 0.96785 0.99992 0.9984

1 2 4 8 9 11 12 27 28 35 36 39 41 42 54 55 57 59 69

0.074878 0.074878 0.194833 6.895699 3.375672 1.015048 2.192752 0.000349 0.000349 0.000479 0.023283 0.115903 0.000044 5.781278 0.041217 0.002622 0.023324 0.001409 0.000012

6 Step 5. Calculation procedure concludes if convergence criterion is satisfied, otherwise, turn to step 2 and keep iterating. The programming logic is very simple which can be seen from the calculation procedure above, because a whole procedure of sweep just traverse line map once, and during the course of each step the calculation is the same for all the lines without consideration of the deference between branch lines and main lines. C Calculation Process Flow Chart The calculation process flow chart is shown as figure 2 according to the calculation procedure listed above.
gd e n z il i t t o u pr m n o a t d a n a D I N C y t i v i t s cn t e n e n n oo p m eo h t ef t o a e r C C Data Initialized Iteration Count k=0 Backward Sweep for End-point Power Flow of Lines Forward Sweep for Front-end Power Flow of Lines Y

4 8 9 11 12 27 28 35 36 39 41 42 54 55 57 59 69

Convergence Criterion Is Satisfied? Y N

The results listed in Table 2 are the same as that in document [11]. The calculation procedure takes 0.028 seconds with 4 times iteration when calculation convergent precision is set at 1e-10. Nodal voltage results are shown in figure 4.
1.02 1 0.98 0.96 0.94 0.92 0.9 1 11 21

Figure 2 calculation process flow chart

IV. Case Study A real 69-bus system in PG&E Power Supply Company is quoted here, which is shown in figure 3. And the test condition for this system is the same as that in document [11].

1 + k = k N

d n E

s t l u s e R t n i r P

Maximun Iteration Count?

t n e g r e v n o C t o N

Figure 4 Power flow calculation result - nodal voltage

V. Conclusion The method presented in this paper gets an advantage of conventional ones with simple logic and friendly programming for iteration just traverses line map once, which is timesaving and insensitive to the system scale and can be easily put into practice. Based on this model, further study is easily carried on, such as taking into consideration of deferent load models and impact assessment of distributed generation (DG) on distribution system, etc. And in sequence, DG optimization can be obtained through choosing proper DG models, which helps the popularity of smart grids. References
[1]

Biographies
Xu Jing-zhou 1981.1- received B.S. degree in electrical engineering from Shanghai Institute of Electrical Power, M.S. degree and Ph.D. degree from Southeast University. He is currently working at Nanjing Power Supply Company. His work mainly includes power engineering design and power system plan. Chen Xiao 1981.5- received B.S. degree and M.S. degree from Southeast University. She is currently working at Nanjing Power Supply Company. Her work mainly includes power system operation and its dispatch.

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

Berg R, Hawkins E S, Pleines W W. Mechanized calculation of unbalanced load flow on radial distribution circuits[J]. IEEE Trans. On Power Apparatus and Systems, 1967, 86(4)415-421. Chen T H, Chen M S, Hwang K J, et al. Distribution system power flow analysis-a rigid approach[J]. IEEE Trans. on Power Delivery, 1991,6(3)1146-1152. Teng Jenhao, Lin Wheimin. Current-based power flow solutions for distribution systems, ICPST 94 Beijing, China, 1994414-418. Van Amerongen R A M, A General Purpose Version of the Fast Decoupled Load Flow, IEEE Trans. on Power Systems, 1989, 4(2)760-770. Fang Zhang, Carol S. Cheng. A modified newton method for radial distribution system power flow analysis[J]. IEEE Trans. on Power Systems,February 1997, 12(1)389-397. Cai Zhongqin, Guo Zhizhong. Newton load flow for radial distribution network based on upstream labeling technique [J]. Proceedings of the CSEE, 2002, 20(6)13-16. Ray D. Zimmerman, Hsiao-Dong Chiang, Fast decoupled power flow for unbalanced radial distribution systems, IEEE Trans. on Power Systems,1995, 10(4)2045-2051. Shirmohammadi D, Hong HW, Semlyen A,et al. A Compensation-based power flow method for weakly meshed distribution and transmission networks [J]. IEEE Trans. on Power Systems, 1988,3(2)753-761. Goswami S K, Basu S K, Direct solution of distribution systems [J]. Proceedings of IEE, Part C, 1991, 138(1)78-88. Sun jian,Jiang Daozhuo,Liu Zhihua. Modified power flow algorithm of distribution network based on forward/backward sweep method [J]. Electric Power Automation Equipment.2004 24 3:81-84 Mesut E.Baran, Felix F.Wu. Optimal Capacitor Placement on Radial Distribution System. IEEE Transactions on Power Delivery, Vol.4, No.1, January 1989:725-734

You might also like