2010 IEEE International Conference on Power and Energy (PECon2010), Nov 29 - Dec 1, 2010, Kuala Lumpur, Malaysia
Optimized Protective Devices Allocation in Electric Power Distribution Systems Based on the Current Conditions of the Devices H. Hashemi. Dezaki, Student member, IEEE, H. Askaraian. Abyaneh, Senior, IEEE, Y. Kabiri, Student member, IEEE, H. Nafisi, K. Mazlumi, H. Akbar Fakhrabadi Amirkabir University of Technology/Department of electrical engineering, Tehran, Iran - Email:
[email protected]
+
Abstract— This paper proposed the new method to carry out the optimized allocation of the protection devices to improve SAIFI based on the current placement of devices. Genetic algorithm (GA) is used to optimize the objective function. Advantages of the method used in this paper are minimal risk to change the protection devices, cost minimization of change the location of the devices and system reliability improvement. In order to illustrate the application of this method, an actual electric power distribution feeder, 20 kV, overhead lines, threewire with a delta-grounded wye connection substation transformer is used. Keyword- distribution systems, protection, reliability, optimization
I. INTRODUCTION
T
HE fundamental purpose of the electric utilities is to service their customers with a reliable and appropriate cost power supply. In the recent studies, the most common indices used about the power quality and reliability of the distribution systems are customer oriented indices. These customer oriented indices are system average interruption frequency index (SAIFI), system average interruption duration index (SAIDI), average system interruption frequency index (ASIFI), average system interruption duration index (ASIDI), customer average interruption duration index (CAIDI) and average system availability index (ASAI). These indices have been detailed in several relevant papers with the reliability of the distribution systems [1,2]. Optimized and safe designing of distribution systems concludes the protective and switching devices in strategic places of the distribution networks to improve the reliability and power quality indices. Several papers are found in literature dealt with the optimized placement of protective and switching devices. References [3-7] survey separately the optimal switch allocation for network energy restoration. Other references have analyzed the optimized allocation of the protective devices in distribution systems [1,8-10]. Also several papers proposed different methods to optimally place both switches and protective devices [11-14]. In the model proposed in [3], the objective function includes outage, capital cost and maintenance to find the optimal placement of the switches. In other reference [4], the proposed method has dealt with the comparison between
978-1-4244-8946-6/10/$26.00 ©2010 IEEE
577
cost of the non- supplied energy and cost of the sectionalizing switches. The algorithm used to solve the optimization problem is genetic algorithm. [8]. Binary linear programming model is used for the placement of protection devices, fault locators and other sensors in the distribution networks [1,9]. In References [11,13], the problem of the optimized switches and protective devices is modeled through non- linear programming (MINLP) with real and binary variables. The reactive tabu search algorithm (RTS) is used to solve the optimized problem. Ref [14] has presented a multi objective optimization methodology to optimally place of the switches and protective devices to improve the system reliability indices. The multi objective ant colony optimization (MACO) has been applied to solve the problem. The proposed placement of switches and protective devices leads to minimize the total cost while simultaneously minimize two distribution network reliability indices including SAIFI and SAIDI. This paper proposed the new method to carry out the optimized allocation of the protection devices to improve SAIFI based on the current placement of devices. Genetic algorithm (GA) is used to optimize the objective function. Advantages of the method used in this paper are minimal risk to change the protection devices, cost minimization of change the location of the devices and system reliability improvement. In order to illustrate the application of this method, an actual electric power distribution feeder, 20 kV, overhead lines, three-wire with a delta-grounded wye connection substation transformer is used. II. PROPOSED ALGORITHM In the proposed method, in process of optimization, the algorithm checks the suggested places for install the protective devices and compares the proposed placement of the devices by the current conditions of the feeder. Each proposed device placement that is different from the placement of the devices imposes a penalty factor to value of the objective. Although in the end of the each step of the optimization, the algorithm compares the total number of the protective devices with the total number of the protective devices. Other value is added to objective in order to the additional devices needed. In this method, SAIFI is determined to minimize. According to the above explanation, it is reasonable to write the objective function in expression as same as the available
equation in the (1):
(λ , X ) ⎡ ⎤ + w × ⎢C × ∑ ( X ) + C × ∑ ( X × I ) ⎥ ⎣ ⎦
O.F = w1 × SAIFI 1
1
j∈β
j
j
j
2
j∈β
j
(1)
j
The objective function is composed by the value of the SAIFI and the cost of the additional devices needed to optimize the system. Where W 1, W 2 are respectively weights for SAIFI and the needed cost to apply the proposed strategies to the distribution feeder. Other parameters used in (1) are λ j , X j , I j C1 , C 2 . X j and I j indicate respectively binary variable demonstrates the section (j) is candidate to allocate the protection device or not and the binary variable to show the existence of the protection device in the section (j). λ j is the failure rate of the section (j). Although the cost needed to additional protective device and the cost of the replacement of the devices are respectively shown in (1) by C1 and C 2 . The flow chart to present the proposed algorithm to optimize the placement of the protective devices is shown as Fig. 1. New method introduced needs to the data of placement of the protective devices in addition to the network topology and information about the loads and customers. Genetic algorithm is used to solve the optimization problem. The main difference among proposed algorithm to available methods to find the optimized protection device allocation is the evaluation of objective function. In this method, the position of the devices is considered. By considering the placement of the protective devices, it is possible to minimize the investment. Although when the needed change in protection system is minimized, it is easier to satisfy the system managers to apply the results. The definition of the parameter I j is demonstrated in (2) as
Δn = ?
ΔX = ?
below:
I
⎧0 ⎪ = j ⎨ ⎪⎩1
if exist device in section j otherwise
(2)
New technique is introduced to calculate SAIFI in [15]. The new model used to calculate the SAIFI of system is briefly shown as (3-5). This model simplifies the calculation of system reliability. All of the symbols used in (3-5) are defined in appendix A. The coefficients used in objective function are the cost of the additional protection device ( C1 ), device replacement cost ( C 2 ), the weights of SAIFI and investment cost ( W 1, W 2 ).
The all values of them is demonstrated in Table 1. The information about switch and protection equipment fixed costs is shown in Table 1 too. These coefficient is similar to the value assumed in [11,14].
578
Fig. 1. Flowchart of the proposed algorithm
2
SAIFI =
∑ factor (i)
(2)
i =1
factor (1) =
T
i −1,i ≠1 ⎛ mb ⎞ + ⎜ ∑ N mj ∑ N mk × A(k + 1, i ) ⎟ k =1 ⎜ j =i ⎟ mb ⎜ flb ( i ) ts ( s ) ⎟ ⎟ ∑ λ mi ⎜ + ∑ ∑ N (s, p) + i =1 ⎜ s =1 p =1 ⎟ ⎜ blb ( i ) ts ( q ) ⎟ ⎜ ∑ ∑ N ( q, r ) × A( fdmb( q ), i ) ⎟ ⎝ q =1 r =1 ⎠
electric power distribution feeder is selected to illustrate the performance of the proposed algorithm [16]. Fig. 2 describes the single line view of the feeder with 33 load points and 32 sections. Failure rate of each section, the number of customers and average load of each section are listed in Table 4.
(3)
factor (2) = flb ( i ) ts ( s )
∑ ∑ λ ( s, p ) × s =1
p =1
flb ( fdmb ( s ) ts ( t ) ⎛ ⎞ × A (1, p ) + ∑ ∑ ∑ N (t , p) × A(1, w) + ⎟ ⎜ N mj t =1 p =1 ⎜ j =1 ⎟ blb ( fdmb ( s ) ts ( v ) ⎜ ⎟ ⎜ ∑ ∑ N (v, y ) × A( fdmb( s ), fumb ( s )) × A(l , w) ⎟ ⎜ v =1 y =1 ⎟ ⎟ ⎜ fumb ( s ) −1 ⎟ ⎜ + ∑ N ml × A(l , p ) × A(l , fdmb( s )) l =1 ⎝ ⎠ mb
(4)
TABLE I SWITCH AND PROTECTION EQUIPMENT FIXED COSTS
Device
Cost (US$/year)
Rcloser
6000
Fuse
1200
Switch
3000
Fig. 2. A typical 11kV, 32 section overhead three-wire distribution feeder TABLE IV DATA OF THE AVAILABLE SWITCHES AND PROTECTIVE DEVICES OF ANALYZED FEEDER
TABLE II COEFFICIENT USED IN PROPOSED ALGORITHM
Parameter
Value
1
2,7,11,13,23,27
Placement of the available fuses
C1
1200
2
11,12,21,23
Placement of the available switches
C2
250
3
1
Placement of the available reclosers
W1
1
W2
10^-2
III. NUMERICAL RESULTS The improved GA algorithm is implemented using Matlab on a Pentium-IV personal computer. In this paper, a typical
579
In the selected distribution feeder, there are 6 fuses, 4 switches and one recloser. There are 20 possible locations for the placement of fuses or sectionalizers. To illustrate the effeteness of improved technique to optimize the allocation of the protective devices, the results of the values of the SAIFI and the objective function in current condition of the system protection and the proposed condition achieved from the object optimization are compared. In Table 5, there are four states explained.
End node
TABLE III DATA OF THE ANALYZED FEEDER Customer Installed Permanent numbers load annual failure rate( f/year) 5 100 0.00954
Branch Number
Initial node
1
1
2
2
2
19
6
90
0.0313
3
19
20
4
90
0.27108
4
20
21
8
90
0.09568
5
21
22
5
90
0.18746
6
2
3
4
90
0.05022
7
3
23
2
90
0.06166
8
23
24
21
420
0.14182
9
24
25
15
420
0.14022
10
3
4
7
120
0.03728
11
4
5
1
60
0.03882
12
5
6
4
60
0.1414
13
6
26
2
60
0.02068 0.02894
14
26
27
3
60
15
27
28
5
60
0.18674
16
28
29
8
120
0.14012
17
29
30
12
200
0.0517
18
30
31
12
150
0.1926
19
31
32
5
210
0.07238
20
32
33
4
60
0.10604
21
6
7
7
200
0.12376
22
7
8
8
200
0.24702
23
8
9
2
60
0.148
24
9
10
2
60
0.148
25
10
11
3
45
0.013
26
11
12
1
60
0.02476
27
12
13
4
60
0.231
28
13
14
6
120
0.14258
29
14
15
2
60
0.1052
30
15
16
3
60
0.109
31
16
17
4
60
0.3442
32
17
18
6
90
0.1148
As shown in Table. 5, if it is not possible to invest in order to increase the system reliability and improve the protection design, the value of SAIFI is 0.6457. Continue with existing protective system needs no investment. But it is desirable to increase the system reliability, there are several scenarios. One scenario is the condition with minimum value of SAIFI. In this condition, the better possible value of SAIFI is accessible. The problem of the scenario discussed is the value of the needed money. By considering the limitation in investment, the scenario discussed is not feasible. The feasible scenario is the scenario satisfies the economical constraints and financial conditions. According to the objective function discussed in this paper, it is possible to find the optimized allocation of the protective devices by considering the economical constraints. To illustrate the proposed algorithm, the value of the objective function is compared to the value of the existing system and the objective function that considers only the cost of the additional needed devices such as the penalty factor. The objective function that discussed is shown as below:
(λ , X ) ⎡ ⎤ + w × ⎢C × ∑ ( X ) ⎥ ⎣ ⎦
O.F = w1 × SAIFI 1
1
j∈β
j
j
(5)
j
The calculated results of the optimized value of the objective function shown in (1) and (5) are respectively in the third and fourth rows of Table. 5. By comparing the value of optimization the objective functions shown in (1) and (5), it is obvious that the needed value of the proposed algorithm in (1) is less than proposed algorithm based on (5). The positions of the protective devices achieved from each discussed scenario are shown in Table. 6. Also the needed money and cost in each scenario is separately discussed.
TABLE V
TABLE VI
SUMMARY VALUES OF THE RESULTS
POSITIONS OF THE PROTECTIVE DEVISED IN THE DIFFERENT SCENARIOS
SAIFI
O.F
FC
Existing system
0.6457
------------
0
Min SAIFI
0.46977
------------
19500
Min objective function(1)
0.4867
0.5133
4000
Min objective function(5)
0.51636
0.53636
2000
580
Total cost to improve ($)
Cost of the additional devices ($)
Cost of the change the position of the device ($)
Proposed positions of the devices
Existing system
0
0
0
2, 7, 11, 13, 23, 27
Min SAIFI
19500
19500
0
All sections expect section (1)
Min objective function(1)
4000
3000
1000
10, 21, 23, 27, 30, 2, 7, 13
Min objective function(5)
2000
1500
500
11, 21, 23, 27, 2, 7, 13
IV. CONCLUSION In this paper, a new algorithm is proposed to allocate the protective devices optimally. In the proposed algorithm, SAIFI is selected to decrease. Important advantage of the discussed algorithm is considering the position of the available protective devices. This algorithm finds the optimized allocation of the protective devices in order to decrease SAIFI and improve the system reliability. When the optimized allocation of the protective devices is found according to the positions of the available protective devices, the needed cost to change the devices is decreased. Although the risk of change the protective devices is decreased effectively. According to the discussed advantages, it is feasible and practical to apply the results of the proposed algorithm to the electric distribution systems. Results of the new method used in this paper conclude the appropriate situations. The proposed algorithm carries out the optimized position to allocate the devices to improve the system reliability while simultaneously minimize the cost of the system and the risk of the changes in system. In order to illustrate the application of this method, an actual electric power distribution feeder, 20 kV, overhead lines, three-wire with a delta-grounded wye connection substation transformer is used. The numerical results illustrate the effectiveness of the new method introduced in this paper.
V. APPENDIX A T: total number of customer in feeder mb: number of sections of the main branch
λ
mi
: Failure rate of the section (i) from the main branch
λ ( s , p ) : Failure rate of the section (p) from sth lateral
branch
N
mi
: Number of customers for section (j) from the
main branch flb (i ) : First downstream lateral branch of section (i) from the main branch N ( s, p ) : Number of customers for section (p) from the th s lateral branch ts( s) : Number of the sections from the sth lateral branch
blb (i ) : First upstream lateral branch of section (i) from the main branch fdmb(i ) : First downstream main branch for ith lateral branch fumb(i ) : First upstream main branch for ith lateral branch VI. REFERENCES [1] F. Soudi, K. Tomsovic, "Optimal distribution protection design: quality of solution and computational analysis," Electric Power and Energy System, vol. 21, pp. 327-335, Apr. 1999.
581
[2] IEEE Working Group on System Design, "Trial use guide for electric power distribution reliability indices," report. 1366, Draft no. 14. [3] R. Billinton, S. Jonnavithula, "Optimal switching decice placement in radial distribution systems," IEEE Trans.on Power System Reasearch, vol. 11, no. 3, pp. 1646-1651, Jan. 1996. [4] G. Celli and F. Pilo, "Optimal sectionalizing switches allocation in distribution networks," IEEE Trans.on Power Delivery, vol. 14, no. 3, pp. 1167-1172, Jul. 1999. [5]G. Levitin, S. Mazal-Tov, D. Elmakis, "Optimal sectionalizer allocation in electric distribution systems by genetic algorithm," Electric Power System Research, vol. 31, no. 1, pp. 97-102, April. 1994. [6] J. Teng, Y. Liu, "A novel ACS-based optimum switch relocation method," IEEE Trans.on Power Systems, vol. 18, no. 1, pp. 113-120, Feb. 2003. [7] J. Teng, C. N. Lu, "Feeder- switch relocation for customer interruption cost minimization," IEEE Trans.on Power Delivery, vol. 17, no. 1, pp. 254-259, Jan. 2002. [8] L. G. W. Silva, R. A. F. Pereira, J. R. S. Mantovani, "Allocation of protective devices in distribution circuits using nonlinear programming models and genetic algorithms," Electric Power System Research, vol. 69,no. 1, pp. 77-84, jul. 2004. [9] F. Soudi, K. Tomasovic, "Optimal trade-off in distribution protection design," IEEE Trans.on Power Delivery, vol. 16, no. 2, pp. 292-296, Apr. 2001. [10] F. Soudi, K. Tomasovic, "Optimized distribution protection using binary programming," IEEE Trans.on Power Delivery, vol. 13, no. 1, pp. 218-224, Jan. 1998. [11] L. G. W. Silva, R. A. F. Pereira, J. R. Abbad, J. R. S. Mantovani, "Optimised placement of control and protective devices in electric distribution systems through reactive tabu search algorithm," Electric Power System Research, vol. 78, pp. 372-381, Apr. 2008. [12] L. G. W. Silva, R. A. F. Pereira, J. R. S. Mantovani, "Optimised allocation of sectionalizing switches and protection devices in distribution networks by using reactive tabu search algorithm," in 18th International Conference of Electric Distribution, Turin, 2005. [13] L. G. W. Silva, R. A. F. Pereira, J. R. S. Mantovani, "Otimized allocation of sectionalizing switches and control and protection devices for reliability indices improvement in distribution systems," in 2004 IEEE/PES Transmission & Distribution Conference & Exposition, Latin America, 2004. [14] W. Tippachon, D. Rerkpreedapong, "Multiobjective optimal placement of switches and protective devices in electric power distribution systems using ant colony optimization," Electric Power System Research, vol. 79, pp. 1171-1178, Mar. 2009. [15] H. Hashemi. Dezaki, H.Askarian. Abyaneh, A. Agheli, S. H. Hosseinian, K. Mazlumi, H. Nafisi, " Optimized investment to decrease the failure rate of distribution lines in order to improveSAIFI," The 4th International Power Engineering and Optimization Conf. (PEOCO2010), IEEE Conference, Shah Alam, Selangor, Malysia: 23-24 June 2010. [16] Mohamed M. Hamada, Mohamed. A.A. Wahab, Nasser. G.A. Hemdan, "Simple and efficient method for steady-state voltage stability assessment of radial distribution systems," Electric Power Systems Research, vol. 80, pp. 152-160, Sep. 2010.
VII. BIOGRAPHIES
Hamed Hashemi Dezzaki was born in 1986, Borujen, Iran. He reeceived B.S. degree in electrical engineering frrom Amirkabir University of Technology, Tehran, Iran I in 2009. Presently, he is a M.Sc. student at thee department of electrical engineering of Amiirkabir University of Technology, Tehran, Irran. His main fields of research are power system m transient and protection of power systems, power quality, high voltaage, reliability, electrical insulation and using artificial intelligence in pow wer system and distributed generation (DG). Hossein Askarian Abyaneh was born in Abyaneh, Isfahan on March 20, 1953. He received the B.S S. and M.S. degree both in Iran in 1976 and 1982 respectively. He also received another M.S. degree and Ph.D. from UMIST, Manchester, U.K. U in 1985 and 1988 respectively, all in electrical e power system engineering. He publishhed over 140 scientific papers in international jouurnals and conferences. Currently, he is a Professor with the Department of Electrical Engineering and Head of the Electrical Engineering Department, AUT, Iran, working in the area of the relay protecttion and power quality. Yahya Kabiri was born in Isfahan, Iran in 1986. He received the B.S. in electrical engineering from Isfahan University of Techhnology, Isfahan, Iran in 2008. Currently he is M.Sc. student at Amirkabir University of Technology, Tehran, T Iran. His main fields of reseearch are fuel cell and photovoltaic hybrid system m, distributed generation (DG) and DG connection to distribution system.
Hamed Nafisi was bornn in Tehran in the Iran, on April 31, 1984. He receeived his B.Sc. and M.Sc. in electrical engineeringg in 2006 and 2008 from the Amirkabir Universitty of Technology (AUT), Tehran, Iran. Now, he is the Ph.D. student in the Amirkabir University of Technology (AUT), Tehran, Iran. His researcch interests include Smart Grid, Power System m Protection, Power Electronic, Using Artificial Intelligence in Power System and Distributed Generation (DG).
Kazem Mazlumi was born in Tehran, Iran in 1976. He received the B.S degree in electrical mirkabir University of engineering from Am Technology, Tehran, Iran, I in 2000, the M.S. degree from Sharif Unniversity, Tehran, Iran, in 2003, the Ph.D. from Amirkabir University of Technology, Tehran, Iran, in 2009. He is currently an assistant professor with Zanjan University, Zanjan, Iran.
582