The optimal automation level of medium voltage distribution networks

Electrical Power and Energy Systems 33 (2011) 430–438

Contents lists available at ScienceDirect

Electrical Power and Energy Systems journal homepage: www.elsevier.com/locate/ijepes

The optimal automation level of medium voltage distribution networks Dragan S. Popovic ⇑, Ljiljana R. Glamocic, Miroslav D. Nimrihter DMS Group Ltd., Novi Sad, Serbia

a r t i c l e

i n f o

Article history: Received 21 July 2009 Received in revised form 24 March 2010 Accepted 5 October 2010 Available online 12 November 2010 Keywords: Distribution network automation Decomposition method Heuristic combinatory search Fault management Remote control Local automation

a b s t r a c t This paper proposed a new methodology for determining the optimal level of investments in medium voltage (MV) distribution network (DN) automation. The problem of network automation is complex, non-linear and discrete optimization problem of enormous dimensions and it is not possible directly apply appropriate optimization procedures. The proposed methodology is based on heuristic combinatory search algorithm with simultaneous consideration scenarios with different types of automation equipment: local automation and remote control. The basis of the proposed procedure is real fault management procedure on the base of which the appropriate estimation of benefits for different network automation scenarios is done. The essence of the algorithm for determining the optimal solution is decomposition of an optimal automation problem with different types of automation equipment to sub problems of network automation with one type of equipment. The proposed methodology has been tested on real network of the city of Belgrade. Obtained results have proven that the proposed methodology is a powerful tool for the determination of the optimal level of investments in the automation of MV DN. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction One of the most important reasons for introducing the automation of network is more efficient fault management. This reduces average outage duration per consumer in medium voltage (MV) distribution network (DN) in case of faults. It reduces costs due to unsupplied energy and improves network reliability. This altogether brings a higher quality of customers power supply and increased income to distribution utilities. The following equipment is considered for network automation [1–3]: (1) fault detectors (directional and no directional) with local and/or remote fault indication; (2) local automation: (reclosers, autosectionalisers, changeovers) and (3) remote controlled switches on control centers with Supervisory, Control and Data Acquisition (SCADA) and Distribution Management System (DMS). The development of DN automation has been diverse. Distribution utilities differentiate by area, number of consumers and their significance, load density, climate conditions, type of the network (cable, overhead and mixture), treatment of DN neutral nodes Abbreviations: AS, autosectionaliser; CB, circuit breaker; CO, changeover; DN, distribution network; ENSI, energy not supplied index; LFD, local fault detector; MV, medium voltage; NOS, normally opened switch; RCB, remote circuit breaker; RCL, recloser; RCS, remote controlled switch; RFD, remote fault detector; RNOS, remote normally opened switch; SAIDI, system average interruption duration index. ⇑ Corresponding author. Tel.: +381 214893500; fax: +381 214893540. E-mail addresses: [email protected] (D.S. Popovic), [email protected] (L.R. Glamocic), [email protected] (M.D. Nimrihter). 0142-0615/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.ijepes.2010.10.004

and available budget. The majority of distribution utilities begun their automation of DN with local automation [4,5]. Later on, after a decrease in prices of telecommunications, utilities that have dominant city cable network started to introduce remote control (England [6], Italy [7], France [3]), while utilities with predominant rural overhead network kept local automation (Belgium, Finland, Norway [8]). Further reduction of telecommunication price levels brought even more intensified use of remote control, even in rural networks. In rural networks, remote control is introduced most frequently in specific situations when it is necessary to increase reliability on interconnected feeders, as well as feeders which supply important customers. Moreover, even in city networks, which usually supply important customers, elements of local automation in order to increase reliability (Italy). Finally, most utilities are introducing modern control centers with full DMS functionality and additional power application [9], which is enabling utility optimal capacities management (France [3], Italy [7]). Based on all previously mentioned facts it is obvious that there is no clear concept of automation for specific types of networks, but rather a specific solution in each single case. This imposes demand for development of appropriate algorithms for determining the optimal automation level for each particular DN [1–3]. The determination of the optimal level of network automation is complex, non-linear and discrete optimization problem of enormous dimensions and it is not possible directly apply appropriate optimization procedures. Different methods are used to deal with a growing concern in power utilities [10] regarding quantitative justification of the increase in reliability due to the placement of

431

D.S. Popovic et al. / Electrical Power and Energy Systems 33 (2011) 430–438

switches rather than one simply based on reduced interruption times. Ref. [11] provides four rules for protective devices location in order to improve reliability (used reliability index is SAIDI). Fuzzy logic and genetic algorithms are used in [12,13] to determine switch locations. Techniques proposed in [12,14] use genetic algorithms, simulated annealing [15] and other heuristics based on the Bellman’s principle of optimality [16] to minimize the investment in sectionalisers. A genetic-algorithm-based sectionalizes planning method, which considers the costs of energy losses and investment, is proposed in [13]. A general combinatorial optimization procedure known as simulated annealing is used in the paper [15] as it can be applied to a non-linear objective function to solve sectionaliser placement problem with the consideration of outage, maintenance and costs of investment. The paper [17] proposes using of two-stage decomposition technique and convex analysis to address the problem of remote controlled switches allocation with classical optimization. A heuristic approach in conjunction with simple numerical computation is presented in the paper [14] where some useful strategic formulations for switch relocation are proposed; however, the optimal switch number and tie switch relocation are not included. The same authors in [18] propose an efficient rule-based method for obtaining optimal numbers and locations of automated switches considering reliability worth. However, methodologies mentioned above mostly process the installation of one type of automation equipment. Such access does not lead to the optimal solution because in real DN optimal solution with one type of automation equipment can be improved by introducing different types of automation equipment. This ensue the conclusion that optimal automation level of a real DN cannot be determined by applying appropriate optimization procedures. The number of all possible scenarios with one type of automation equipment in real network is enormous. Thus, automation problem is solved by the mechanism of combinatory search of only reasonable scenarios of network automation. Scenarios are obtained by adding new elements of the same equipment type to the initial scenario until the saturation, after which scenarios are combined. The quality of each generated scenario is made upon cost/benefit analysis. Cost represents investments in network automation, while benefit is the reduction of total expected annual costs (due to unsupplied energy, cut-off power, costs of maintenance, etc.). This paper presents a new methodology for the determination of the optimal automation level of DN [1] with simultaneous consideration of different types of automation equipment which in optimal way solves the problem between speed of calculation and accuracy. Applying real models of network enables high quality and real evaluation of the effects of applying network automation. It is simplified by using heuristic and decomposition problem of finding optimal solution with simultaneous consideration of different types of automation equipment and a set of potential scenarios for network automation necessary to search is significantly reduced. The proposed methodology is verified on a smaller part of DN of the city of Belgrade and results are compared. Obtained results show that the proposed methodology is a powerful tool for the determination of the optimal level of investments in the automation of DN. The paper consists of five parts. The second part includes detailed description of the methodology used for automation of the DN that is proposed in this paper. The third part presents proposed methodology tested on real DN of the city of Belgrade. Conclusion and references are given in the fourth and fifth part of this paper. After that, the list of abbreviations used in this paper is given.

2. Methodology A new methodology for the determination of the optimal level of DN automation with simultaneous consideration of different

types of automation equipment is presented in this paper. The determination of all possible scenarios for network automation generated by numerical combinatory search algorithm of real DN is very wide. A proposed algorithm is verified on a smaller part of the analyzed DN of the city of Belgrade consisting of one supplying substation with four feeders which supplies customers on 12 distribution substations. All possible scenarios of network automation are generated and combined by numerical search algorithm. This calculation took more than 3 months. The number of scenarios with one type of equipment is in proportion of 109, the number of all scenarios is in proportion of 1024. With the methodology proposed in this paper the number of scenarios is drastically reduced to a reasonable number of 1125 scenarios. The solution is almost the same as the optimal solution obtained with the numerical algorithm. The optimal solution obtained by the proposed methodology is slightly different from the optimal solution obtained by the numerical algorithm in position of two reclosers. The number of scenarios considering all types of equipment is increasing with the increase of network dimensions (available positions of automation equipment) and applying numerical algorithm in real DN is not possible. The essence of the proposed algorithm is the decomposition of the optimal network automation problem to sub problems of network automation with one type of equipment. At the end, the sub problems synthesis is done with the purpose of finding an optimal solution with simultaneous consideration of different types of automation equipment. The proposed methodology for optimal scenario of network automation which is based on heuristic combinatory search algorithm consists of five main steps (Fig. 1). 2.1. Analysis of considered DN and generating strategies In the first step of the algorithm, the analyses of considered network (Fig. 1) are performed (type of network, supply concept, per-

START 1 Analysis of considered network and detection of “weak” points Generating set of strategies – NS 2

i = 1, NS Defining initial scenario in strategy i Creating set of scenarios in strategy i – NCRE(i)

3

j = 1, NCRE(i)+1 Reliability assessment of scenario j in strategy i Cost/benefit analysis of scenario j in strategy i

4 Ranking all scenarios (initial and created) of strategy i and selection of suboptimum 5 Synthesis of suboptimal solutions and finding optimal solution Sensitivity analysis STOP Fig. 1. Proposed algorithm for the determination of the optimal DN automation.

432

D.S. Popovic et al. / Electrical Power and Energy Systems 33 (2011) 430–438

centage of involved cable and overhead lines, number and location of important customers, weak network points with analysis of existing equipment for automation and remote control system, etc.). The main part of these analyses is the detection of critical network parts, like critical supplying substation, critical feeders and their points from the aspect of reliability. In addition, the following is taken into consideration: solutions of the automation in similar networks, engineering intuition (heuristics), decisions making according to desired reliability and available budget (‘‘budget constrained optimization”). The experience gained from solutions obtained for DN in urban (city) areas with cables includes remote controlled automation equipment. The reliability of these networks can be improved by adding local automation for important customers. Usually, in DN in rural areas with overhead or mixed lines local automation (later introducing remote controlled switches) is implemented. Based upon above mentioned analyses a set of strategies is defined for the automation of considered network. Each strategy represents the implementation of one or several types of automation equipment. Types of equipment in different strategies must be independent or weakly dependent from the point of view of times relevant for fault management procedure. For example, fault detectors have influence on the duration of fault location, and changeovers on the duration of supply restoration. In thus way, fault detectors and changeovers can be the subjects of two different strategies since they are mutually independent from the aspect of time in fault management procedure. 2.2. Generating scenarios One strategy consists of a set of scenarios. Each scenario presents setting of a certain number and type of automation equipment in network. One scenario that is initially defined and a set of scenarios are created from this, initial one. For example, one possible strategy for network automation is introducing remote control of switches and fault detectors with remote indication, since it is natural that these two mutually dependent types of equipment are combined. Initial scenario for this strategy can be network configuration with remote controlled circuit breakers on feeder heads. In the second step of an algorithm for the optimal network automation (Fig. 1), for each initial scenario a desired number of created scenarios is generated. Scenarios are created by adding new elements of the same type of equipment to the initial scenario, which increases investments

and improves reliability all the way until the saturation – adding new equipment does not significantly contribute to increasing reliability, but significantly increases the costs of investments of the scenario. For example, for initial scenario with remote control, the following created scenarios are generated: remote control on circuit breaker on the feeder head and on each normally opened switch (NOS). Subsequently, fault detectors with remote indication are added, as well as remote controlled switches in substations/ poles located on feeder mid-point, in substations/poles located on thirds of feeders, etc. starting from detected critical parts of network (critical supplying substations, critical feeders . . .). During the generation of scenarios, the following can vary (Fig. 2): (1) Area of network to be automated; (2) objects inside the area (distribution substations, switching substations and/or poles); (3) bays within objects that are being automated (input/ output); (4) special locations of automated objects within the area. 2.3. Reliability assessment and cost/benefit analysis For each of generated scenarios, the reliability profile of network is provided. There are two different approaches of reliability assessment [19,20]. The first approach is based on analysis of simplified and idealized models of network and fault management procedure [19]. Advantage of this approach is that for a very short time, a relatively large number of scenarios for automation can be searched over, while its shortcomings is that it does not give a real evaluation of automation network effects. The second approach, used in the proposed methodology, is based on real models of network and includes exact fault management procedure applied for reliability assessment in the considered network [1,2,20,10] which gives far more realistic evaluation of benefit gained by network automation. Using real models, all elements and equipment are modelled with their exact and real data (i.e. lines are modelled with real parameters and failure rates (using historical data), consumers with real power and importance, switches with real breaking capacities and current ratings). The reliability assessment with the exact fault management procedure is considered on the basis of network topology and a configuration of automation equipment. For fault location procedure fault indicators and/or different types of switches are used. During supply restoration different types of manual and remote controlled switches at normally opened points are considered (changeovers, load break switches, disconnectors). For a certain fault (after it is located) the best variant of supply • •

•

•

•

Distribution substations, Switching substations, Poles.

• •

Area

Remote control, Local automation.

Objects

Entire MV network, One or several supplying substations, • One or several MV feeders. • •

Bays within objects

•

In NOSs, On the feeder’s mid-point, on points of feeder’s thirds, quarters, In each, each second, third, fourth object on the feeder.

Location Type of automation Type of equipment

Transformer bay, Feeder bay (input and/or output), • Lateral bay (important or all laterals). • •

RTU, Reclosers, • Autosectionalisers, • Change-overs. • •

Fig. 2. Generating scenarios for DN automation.

D.S. Popovic et al. / Electrical Power and Energy Systems 33 (2011) 430–438

restoration is chosen, on the basis of switch type and feeder load transfer capability. Power flow is used for checking the breaking capacity and violations on an alternative source of supplying. In the third step of the algorithm (Fig. 1) for each created scenario, reliability indices ENSI and SAIDI are estimated through the simulation procedure. Regarding reliability indices calculation, the main problem is in estimation of outage duration, which is the function of: position of considered consumer node compared to source of supply and faulted section, existence of alternative source of supply, type of automation equipment, number of equipment and a certain network automation scenario. Details about the way of time calculation are given in Appendix A. The quality of each generated scenario is made upon cost/benefit analysis and quantified by C/B ratio for planned period of npl years. The cost represents investments in network automation (Cinv), while the benefit is the reduction of total expected annual costs (due to unsupplied energy, cut-off power, costs of maintenance) defined by [1]

B ¼ C act  C DA ;

ð1Þ

where Cact and CDA are actualized total costs in the network prior and after network automation, defined by

C act ¼

npl X ðC ENSIt þ C Mt Þð1 þ id Þt ;

ð2Þ

t¼1

C DA ¼ C inv þ

npl X ðC ENSIDAt þ C MDAt Þð1 þ id Þt ;

ð3Þ

t¼1

where C ENSIt ; C ENSIDAt is the costs of unsupplied energy in the tth year in actual network prior to and after network automation, calculated by using the function of interruption duration for each customer in analyzed network, like in [21,22], C Mt ; C MDAt is the costs of maintenance in the tth year in actual network prior to and after network automation and id is actualization rate. 2.4. Ranking scenarios and obtaining suboptimal solution In the fourth step of the proposed algorithm from Fig. 1, all scenarios from one strategy are ranked and suboptimal scenarios are obtained. Scenarios can be ranked according to appropriate C/B ratio, benefit (calculated using Eqs. (1)–(3)) or appropriate reliability index. The optimization problem is defined as multi-objective with the optimization criterion:

maxfBj g:

ð4Þ

minfRIj g; RIj ¼ fSAIDIj ; ENSIj g;

ð5Þ

minfðC=BÞj g;

ð6Þ

j

j

j

where RI is one of the reliability indices. The following constraints can be considered for the optimal scenario determination:

433

Under a limited budget (constraint (7)), automation equipment in scenario j is set in such a way that the overall benefit is maximized (criterion (6)) and reliability indices are minimized (criterion (5)), considering constraint (8). Multi solution can be obtained with different budgets. By using multi-objective optimization criteria, the problem can be defined in the following way: how much to invest in network automation to reduce reliability index to a certain level, as one of the most important measures of many distribution utilities. In that case, optimization criterion is (4), with constraint (9)). Or, how much is necessary to invest in network automation to get minimal possible reliability index (compared to optimization criterion (5), without considering the constraints (7)–(9)). 2.5. Synthesis of suboptimal solutions and finding optimal solution The result of previous step is ranking list of scenarios for network automation. Each strategy matches one ranking list and one suboptimum list. Determining the optimum is performed on the basis of all strategies and it comprises the simultaneous implementation of different types of automation equipment. In this way, suboptimal solutions synthesis is done, as well as the search for the optimum. Consequently, all possible combinations of scenarios from different strategies need to be considered, which at the same time satisfy budget constraint (constraint (7)). This provides recommendation on how to optimally spend a certain amount of money for network automation. In the decomposition of the optimal network automation problem, the initial assumption is that certain strategies are independent or weakly dependent from the aspect of fault management procedure. In case when there is a certain dependency between several types of equipment from different strategies, in order to check the optimum, there should be checked as many highly ranked scenarios from the lists as possible. This multi solution approach provides sensitivity analysis of the optimal solutions in comparison with the variation of significant parameters value. Forecasted load, costs due to cut-off kW and undelivered kWh and prices of automation equipment should be considered as significant parameters. Considering that there is a particular uncertainty in value of these parameters, it is necessary to investigate the robustness of the optimal solution in relation to value variation of these parameters. By acknowledging the expected level of uncertainty of these parameters, it is possible to change the ranking list and obtain a set of suboptimal solutions and the optimum. 3. Test results The proposed methodology is tested on the DN of the city of Belgrade (Fig. 3) with 187 secondary substations, and 37,000 consumers. Based on detailed analysis of the considered network, two strategies have been selected: (A) remote control and (B) local automation. The explanation of symbols for type and position of objects being automated with different network automation scenarios is given in Table 1. Symbols used for the automation equipment are described in Abbreviations.

C jinv 6 C b limit ;

ð7Þ

3.1. Remote control strategy

ðC=BÞj 6 1;

ð8Þ

RIj 6 RImin ;

ð9Þ

For the strategy with remote control a scenario with remote controlled circuit breaker (RCB) on feeder head is selected as the initial scenario. The first created scenario (scenario R1) is obtained by adding remote control in NOSs on important feeders (FI), on feeders with high priority customers (scenario R2) and on each (F) feeder (scenario R3) – Table 2. Then, created scenarios are gen-

where Cb limit and RImin are limited budget and minimal allowable reliability index, respectively.

434

D.S. Popovic et al. / Electrical Power and Energy Systems 33 (2011) 430–438

erated by adding remote monitored fault detectors (RFD) and remote controlled switches (RCS) along feeders (in every fourth substation/pole, in every third substation/pole, in every second substation/pole, etc.), until adding new equipment achieved an apparent saturation of benefit. Also, selected feeders/laterals vary where automation is performed (feeders/laterals with high priority costumers, important feeders/laterals, all feeders, etc.). The most interesting scenarios, with calculation results for reliability indices ENSI and SAIDI, and C/B ratio, given in Table 2, are ranked according to Cinv. These scenario results are displayed graphically in Fig. 4. It is obvious that reliability indices ENSI and SAIDI are decreasing with the increase in the cost of investment. For scenario R3, from Table 2 and Fig. 4, comparing with the scenario R2 peak C/B value is obtained while reliability index SAIDI decreases. The reason is higher cost of investment for adding remote controlled switches in NOSs of all feeders in scenario R3 (in scenario R2 they are added only in NOSs of feeders with a high priority customers). Scenarios R21 and R22 follows the creation of scenario R3 decreasing SAIDI (ENSI). With budget constraint of 800,000 € the optimal scenario would be scenario R20 from the Table 2 (row colored in grey) which includes:

Fig. 3. Geographical display of the part of DN of city of Belgrade.

Object with NOS Object on the feeder/lateral head Object with high priority customers Important object Each object Each second object Each third object Each fourth object Object on feeder mid-point Object on feeder thirds Object on feeder fourths

Feeder

Lateral

Substation/pole

– FH FP FI F – – – – – –

LH – LI – – – – – – –

SNO SH – – S S2 S3 S4 S02 S03 S04

5

R20

4

C/B SAIDI

3

SAIDI

Type of object

6

2 1 60000 63000 171000 315000 318000 342000 345000 426000 453000 537000 543000 587400 593400 648000 696000 698400 693000 749400 775200 781200 819000 886200 1131000

Selection of object

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

C/B

Table 1 Explanation of symbols for type/position of objects being automated.

0

Cost of investment Fig. 4. SAIDI and C/B as functions of cost of investments for remote control strategy.

Table 2 Display of calculation results for remote control strategy. Sc.

Type of equipment

Feeder, lateral

Substation/pole

Cinv (€)

ENSI (kW h/yr)

SAIDI (h/cons yr)

C/B

R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 R12 R13 R14 R15 R16 R17 R18 R19 R20 R21 R22 R23

RCB+RCS RCB+RCS RCB+RCS R1+RCS R2+RCS R1+RCS+RFD R2+RCS+RFD R3+RCS R3+RCS+RFD R1+RCS R2+RCS R1+RCS+RFD R2+RCS+RFD R3+RCS R3+RCS+RFD R3+RCS R1+RCS R3+RCS+RFD R1+RCS+RFD R2+RCS+RFD R3+RCS R3+RCS+RFD R1+RCS

FI FP F F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI F, LI

SH, SNO SH, SNO SH, SNO S02, LH S02, LH S02, LH S02, LH S02, LH S02, LH S03, LH+S02 S03, LH+S02 S03, LH+S02 S03, LH+S02 S03, LH+S02 S03, LH+S02 S4, LH+S4 S4, LH+S4 S4, LH+S4 S3, LH+S3 S3, LH+S3 S3, LH+S3 S3, LH+S3 S2, LH+S2

60,000 63,000 171,000 315,000 318,000 342,000 345,000 426,000 453,000 537,000 543,000 587,400 593,400 648,000 696,000 698,400 693,000 749,400 775,200 781,200 819,000 886,200 11,31,000

157.60 153.34 148.51 93.72 92.56 91.73 90.65 90.32 88.42 79.10 77.97 77.40 76.31 76.00 67.40 74.34 73.83 65.36 63.24 62.74 62.98 60.65 56.74

5.68 5.52 5.35 3.37 3.33 3.30 3.26 3.25 3.18 2.85 2.81 2.78 2.75 2.73 2.42 2.67 2.66 2.35 2.27 2.26 2.27 2.18 2.04

0.39 0.33 0.65 0.33 0.33 0.35 0.35 0.42 0.44 0.46 0.47 0.51 0.51 0.54 0.54 0.58 0.56 0.58 0.60 0.61 0.61 0.67 0.79

435

D.S. Popovic et al. / Electrical Power and Energy Systems 33 (2011) 430–438 Table 3 Display of calculation results for strategy with local automation. Type of equipment

Feeder, lateral

Substation/pole

Cinv (€)

ENSI (kW h/yr)

SAIDI (h/cons yr)

C/B

L1 L2 L3 L4 L5 L6 L7 L8

RCL+CO L1+AS L1+AS L1+RCL+AS L1+RCL L1+AS L1+RCL+AS L1+AS

FP F, LI F, LI FI, LI FI, LI F, LI FI, LI F, LI

SH, SNO S03, LH+S02 S02, LH S02, LH+S02 S02, LH S04, LH+S03 S02, LH+S03 S, LH+S04

8000 68,000 113,000 123,000 173,000 243,000 283,000 773,000

166.11 120.95 111.44 103.95 94.55 87.07 100.99 72.59

5.98 4.36 4.01 3.74 3.40 3.13 3.64 2.61

0.83 0.09 0.12 0.12 0.14 0.18 0.25 0.48

0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

7

straint is obtained with scenario L8 (row colored in grey), which includes:

6 SAIDI

L8

5 4 3

SAIDI

C/B

Sc.

2

C/B

1 0

8000

– Reclosers (RCL) on feeder heads and changeovers (CO) on substation with NOS (scenario L1). – Autosectionalisers (AS) on all poles (S) in all feeders (F). – Autosectionalisers (AS) on heads of important laterals (LH) and on fourths of important laterals (S04).

68000 113000 123000 173000 243000 283000 773000

Cost of investment Fig. 5. SAIDI and C/B as functions of cost of investments for strategy with local automation.

– Remote controlled circuit breaker on the feeder head (RCB) and remote controlled NOSs on feeder with priority customers (scenario R2). – Remote controlled switches (RCS) and remote monitored fault detectors (RFD) on every third substation/pole (S3) on all feeders (F). – Remote controlled switches (RCS) and remote monitored fault detectors (RFD) on the heads (LH) of important laterals (LI) and on every third substation/pole of important laterals (S3). 3.2. Local automation strategy For the initial scenario for strategy with local automation scenario with recloser (RCL) on feeder head is chosen. The first created scenario is gained by adding changeovers (CO) on substation with NOS where the biggest effect is achieved (scenario L1 in Table 3). Then, autosectionalisers (AS) and reclosers (RCL) are added along feeders and important laterals (in poles on feeder mid-points, on feeder thirds, on feeder fourths, on each pole, etc.). The selected feeders/laterals that are automated also vary (feeders/laterals with high priority costumers, important feeders/laterals). The most interesting scenarios with local automation equipment, with calculation results for reliability indices SAIDI and ENSI, and C/B ratio, ranked according to Cinv are displayed in Table 3 and graphically in Fig. 5. Suboptimal solution for given budget con-

For obtaining optimum it is necessary to investigate all possible combinations of scenarios from both defined strategies which satisfy budget constraint of 800,000 €. In creating this optimum it is necessary to start from assumption that all scenarios from both strategies are independent or weakly dependent. Most interesting combined scenarios are shown in Table 4. With ‘‘Li” (i = 2,4,5,7,8) are indexed modified scenarios ‘‘Li” from Table 3 without considering CO in NOS. During the combination, scenarios from strategy with remote control (‘‘R” scenarios) and strategy with local automation (‘‘L” scenarios) are overlapping and the priority is given to equipment with remote control. The analysis of all combinations of scenarios that satisfy given budget constraint confirms that the combination of scenarios R20 and L5 (scenario L5, but without CO in NOS) – combined scenario S1 from the Table 4 is the optimum which includes equipment from: 1. Scenario R20 – Remote controlled circuit breaker on the feeder head (RCB) and remote controlled NOSs on feeder with priority customers (scenario R2). – Remote controlled switches (RCS) and remote monitored fault detectors (RFD) on every third substation/pole (S3) on all feeders (F). – Remote controlled switches (RCS) and remote monitored fault detectors (RFD) on the heads (LH) of important laterals (LI) and on every third substation/pole of important laterals (S3). 2. Scenario L5 – Reclosers (RCL) on important feeders heads, on their mid-points and on the beginning of important laterals (changeovers (CO) are not set because priority is given to remotely controlled NOS).

Table 4 Display of calculation results for strategy with combination remote control strategy and local automation. Sc.

Combined scenarios

Cinv (€)

ENSI (MW h/yr)

SAIDI (h/cons yr)

C/B

S1 S2 S3 S4 S5 S6 S7 S8 S9

R20+L5* R21+L5* R22 R22+L5* R22+L4* R21+L2* R18+L7* R23+L7* R20+L8*

769,800 867,000 886,200 934,200 944,900 953,500 973,100 1,000,00 1,306,200

61.80 60.15 60.65 59.14 60.63 60.22 61.70 58.32 59.77

2.22 2.16 2.18 2.13 2.18 2.17 2.22 2.10 2.15

0.57 0.61 0.67 0.67 0.63 0.66 0.66 0.63 0.81

436

D.S. Popovic et al. / Electrical Power and Energy Systems 33 (2011) 430–438

A

That means that optimal solution from strategy with remote control (suboptimal scenario R20) with the same budget can be improved by adding reclosers on important feeders mid-points without changeovers (modified scenario L5). The combination of suboptimal scenarios, obtained from strategy with remote control (scenario R20) and strategy with local automation (scenario L8) there is obtained combined scenario S9, shown in Table 4. With scenario S9 are improved reliability indices of both scenarios, R20 and L8, but for network automation using this combined scenario it is necessary to invest more than 1,300,000 €.

CB1

0

s1 s2

s3 s4

s5 s6

s7 s8

1

2

3

4

A

After a fault on section 3, consumer nodes can be divided in two groups according to outage duration (Fig. 6): nodes ‘‘in front of” fault, which belong to area A (nodes 1 and 2) and nodes ‘‘behind” fault, which belong to area C (nodes 3–5). Outage durations for these nodes are presented in Table 5. After fault on section 3 relay, the protection is activated for the analyzed feeder (from Fig. 6) and circuit breaker CB1 leaves all consumers without supplying. Fault

s5 s6

s7 s8

2

3

4

1

s9 RNOS

5

RCB Remotely controlled circuit breaker RNOS Remotely controlled normally opened switch RFD Remotely monitored fault detector Fig. 7. Feeder with RFD, RCB on the feeder head and remote controlled NOS (RNOS).

A RCB1 s1 s2

s3

C RCB2 s5 s6

s7 s8

s9

2 1

2

3

4

5

Fig. 8. Feeder with RFD and RCBs on the feeder head and on feeder mid-point.

A

C

s3 RCB2 s5 s6

RCB1 s1 s2

 lengths of all sections of analyzed feeder are 2 km,  speed of field crew moving during fault management procedure is 1 km/min,  duration of one manual switch operation is 2 min,  duration of one remote switch operation is 0.1 min,  duration of entry into distribution substation is 10 min,  time of repairing faulted element is 5 h (300 min).

C

s3 s4

1

Appendix A

In this Appendix fault management procedure with calculation of consumer outage duration after the fault on section 3 is given. The analyzed feeder from Fig. 6 is automated with different automation scenarios. Two different strategies are shown here: (A) Remote control (Figs. 7–10, 19 and 20) and (B) local automation (Figs. 11–18). The network automation scenarios with combination of equipment from two strategies are shown in Figs. 13 and 14. Outage durations of consumer nodes for feeder automated by different automation scenarios are presented in Table 5. Due to its simplicity, the calculation is presented with the following assumptions:

5

Fig. 6. Analyzed feeder.

RCB1 s1 s2

A.1. Outage duration for different feeder automation scenarios

s9

CB1 Circuit breaker s1,…,s9 Manual switches NOS Manual normally opened switch

4. Conclusion The paper presents a new methodology for the determination of the optimal level of DN automation. The presented methodology is based on the robust heuristic combinatory search algorithm of real model of DN and with multi solutions. The decomposition of network automation problem into sub problems, with the determination of network automation suboptimum within one type of equipment, is used in this search. Finally, sub optimums are integrated and corrected in the effort to gain optimal solution of DN automation with all types of automation equipment. The optimal solution is confirmed by the extensive analysis of combinations of suboptimal solutions. Moreover, used multi solution approach takes into account the uncertainties and variation of forecasted loads, as well as costs due to unsupplied energy and prices of automation equipment. The presented methodology has been tested on real DN of the city of Belgrade. Obtained results have proven that the presented methodology is a powerful tool for the determination of the optimal level of investments in the automation of DN and its optimal design in open access and competitive electricity market environment.

C

s7 s8

s9 s10

4

5

3

NOS 1

2

3

Fig. 9. Feeder with RFD, RCBs on the feeder head and on feeder mid-point and manual NOS.

A

C

s3 RCB2 s5 s6

RCB1 s1 s2

s7 s8

s9 RNOS

4 1

2

3

4

5

Fig. 10. Feeder with RFD, RCBs on the feeder head and on feeder mid-point and remote controlled NOS (RNOS).

A CB1

C

s1 s2

s3 s4

s5 s6

s7 s8

1

2

3

4

s9

5 CB LFD RCL AS

Circuit breaker Fault detector with local indication Recloser Autosectionaliser

CO

Changeover in normally open point

Fig. 11. Feeder with LFD.

5

437

D.S. Popovic et al. / Electrical Power and Energy Systems 33 (2011) 430–438

A CB1

s1 s2

C

s3 s4

A

s5 s6

s7 s8

6

NOS 1

2

3

4

A

C

s3 AS

s5 s6

s7 s8

2

3

4

1

5

Fig. 19. Feeder with LFD, RCL on the feeder head, AS on feeder mid-point and remote controlled NOS (RNOS).

C

s3 AS

s5 s6

s7 s8

s9

7

A

1

s9 RNOS

13

5

Fig. 12. Feeder with LFD and manual NOS.

RCL1 s1 s2

RCL1 s1 s2

s9 s10

2

3

4

RCL1 s1 s2

5

C

s3 RCL2 s5 s6

s7 s8

s9 RNOS

14 Fig. 13. Feeder with LFD, RCL on the feeder head and AS on feeder mid-point.

1

2

3

4

5

Fig. 20. Feeder with LFD, RCLs on the feeder head and on feeder mid-point and remote controlled NOS (RNOS).

A RCL1 s1 s2

C

s3 AS

s5 s6

s7 s8

s9 s10

8

NOS 1

2

3

4

5

Fig. 14. Feeder with LFD, RCL on the feeder head, AS on feeder mid-point and manual NOS.

A RCL1 s1 s2

C

s3 AS

s5 s6

s7 s8

s9 CO

2

3

4

5

9 1

Fig. 15. Feeder with LFD, RCL on the feeder head, AS on feeder mid-point and CO in NOS.

A

C

s3RCL2 s5 s6

RCL1 s1 s2

s7 s8

s9

10 1

2

3

4

5

Fig. 16. Feeder with LFD, RCLs on the feeder head and on feeder mid-point.

A RCL1

s1 s2

C

s3 RCL2 s5 s6

s7 s8

s9 s10

4

5

11

NOS 1

2

3

Fig. 17. Feeder with LFD, RCLs on the feeder head and on feeder mid-point and manual NOS.

A RCL1 s1 s2

C

s3 RCL2 s5 s6

s7 s8

s9

CO

12 1

2

3

4

5

Fig. 18. Feeder with LFD, RCLs on the feeder head and on feeder mid-point and CO in NOS.

location, using bisectional search method and fault isolation procedure start with moving field crew to distribution substation 2 (the distance from feeder head to this node is 4 km). Time needed for the intervention of the crew to this node is proportional to the sum of lengths of sections to this node (lengths of sections 1 and 2 are 2 km) and the speed of field crew moving 1 km/min, which gives the total time of (2 + 2)/1 = 4 min. Time needed to enter substation 2 is 10 min. Then, manual switch s4 is being opened, in order to determine whether the fault is located on the first or the second feeder mid-point (operation duration s4o is 2 min). Then CB1 is being closed (operation duration CB1z is 2 min). Since the circuit breaker is still on, it is concluded that the fault is on the second feeder mid-point, but in order not to manipulate under voltage, CB1 is being opened (operation duration CB1o is 2 min), and switch s4 is being closed manually (operation duration s4z is 2 min). The field crew then moves from distribution substation 2 to substation 4, with the duration of (2 + 2)/1 = 4 min, enters substation 4 (with the duration of 10 min), opens switch s8 (operation duration s8o is 2 min). The conclusion is that the fault is on feeder part between substation 2 and 4. Switch s8 is being closed (operation duration s8z is 2 min). Then the field crew moves from distribution substation 4 to substation 3, with the duration of 2/1 = 2 min, enters substation 3 (10 min) and opens switch s6 (operation duration s6o is 2 min). CB1 on feeder head is being closed (operation duration CB1z is 2 min). While relay protection affects CB1, it is being reopened. Switch s6 is being opened manually (operation s6z duration is 2 min) and in the same substation 3 switch s5 is being opened manually (operation duration s5o is 2 min). CB1 is being closed (operation duration CB1z is 2 min), and because of relay protection activation, it is being reopened. Conclusion is that the fault is on section 3 which is between substations 2 and 3. Switch s5 stays opened, and field crew moving through section 3 (with duration of 2/1 = 2 min) goes again in distribution substation 2 (10 min) and opens switch s4 (2 min). By closing CB1 (operation duration CB1z is 2 min) the procedure of fault location and isolation and supply restoration of consumers in nodes 1 and 2 is finished. Outage duration of consumer in those nodes (from area A) is described by the equation in column TA and row 0 from Table 5. There is no possibility to restore supplying on the analyzed feeder (no NOS), so consumers in nodes 3–5 have to wait for fault repairing, and outage duration of consumer in those nodes (area C) is described by the equation in column TC and row 0 of Table 5. By analyzing the results from Table 5, for the analyzed fault on section 3, the following conclusion can be drawn:

438

D.S. Popovic et al. / Electrical Power and Energy Systems 33 (2011) 430–438

Table 5 Duration of consumer nodes supply interruptions for different feeder automation scenarios. Sc. 0

TA (min) (consumer nodes 1 and 2)

TC (min) (consumer nodes 3–5)

ð2 þ 2Þ þ 10 þ 2 þ |{z} 2 þ |{z} 2 þ ð2 þ 2Þ þ |{z} 2 þ |{z} 2 |fflfflfflffl{zfflfflfflffl} |{z} |{z} |fflfflfflffl{zfflfflfflffl} o z o z z TS2

0!2

r4

p1

p1

r4

2!4

78

TA þ

300 |{z}

¼ 78 þ 300

378

time for repairing

r4

þ |{z} 10 þ |{z} 2 þ |{z} 2 þ |{z} 2 þ |{z} 10 þ |{z} 2 þ |{z} 2 þ |{z} 2 þ |{z} 2 r8o

TS4

r8z

4!3

TS3

r6o

p1z

r6z

r5o

þ |{z} 2 þ |{z} 2 þ |{z} 10 þ |{z} 2 þ |{z} 2 p1z

1

3!2

r4o

p1z

18

0

0

TS2

0!2

2

TS2

2 ð2 þ 2Þ þ 10 þ 2 þ |{z} |fflfflfflffl{zfflfflfflffl} |{z} |{z} o z r4

T A þ |{z} 2 þ |{z} 10 þ |{z} 2 þ 10 þ 0; 1 |fflfflfflfflfflffl{zfflfflfflfflfflffl}

42,1

ð2 þ 2 þ 2Þ þ 10 þ 2 þ |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} |{z} |{z} o

318

2!3

p1

TS3

TS3

0!3

3

0

0

0

0

10 þ |{z} 2 þ |{z} 2 þ |{z} 10 þ |{z} 2 þ |{z} 2 ð2 þ 2 þ 2Þ þ |{z} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} o o z

34

10 þ |{z} 2 þ |{z} 2 þ |{z} 10 þ |{z} 2 þ |{z} 2 ð2 þ 2 þ 2Þ þ |{z} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} o o z

34

0!3

6

0!3

7

TS3

TS3

r5

r5

3!2

3!2

TS2

TS2

r4

r4

p1

ð2 þ 2 þ 2Þ þ 10 þ 2 þ 10 þ 0; 1 |fflfflfflfflfflffl{zfflfflfflfflfflffl} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} |{z} |{z} o

28,1

TA þ

334

TS3

20½s þ 40½s |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}

1

20½s þ 40½s |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}

1

20½s þ 40½s |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}

1

0

0 20½s þ 40½s |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}

1

14

0

0

References [1] Popovic DS, Glamocic Lj, Nimrihter MD, Tanaskovic M, Vukotic D, Damljanovic D. Optimal automation level of medium voltage distribution network. In: 18th International conference on electricity distribution CIRED, paper 400, June 6–9, Turin; 2005. [2] Final report of study. Design of control concept for 10 kV DN JP Elektrodistribucija – Belgrade, Belgrade; 2004. [3] Final report of working group of fault management WG03. Fault management in electrical distribution systems. In: 15th International conference on electricity distribution CIRED, Nice; 1999. [4] Purucker SL, Thomas RJ, Monteen LD. Feeder automation designs for installing an integrated distribution control system. IEEE PAS 1985(10):2929–34. [5] Roman H, Hylla H. Experiences on location of earth faults and short circuits in rural medium voltage networks. In: 17th International conference on electricity distribution CIRED, Barcelona; 2003, p. 3–9. [6] Jackson RE, Walton CM. A case study of extensive MV automation in London. In: 17th International conference on electricity distribution CIRED, Barcelona; 2003, p. 3–36. [7] Cerretti A, Di Lembo G, Di Primio G, Gallerani A, Valtorta G. Automatic fault clearing on MV networks with neutral point connected to ground through impedance. In: 17th International conference on electricity distribution CIRED, Barcelona; 2003, p. 3–6. [8] Chollot Y, Biasse J, Malkot A. Feeder automation improves medium voltage network operating efficiency. CIRED seminar 2008: SmartGrids for distribution, paper 0031; 2008.

300 |{z}

319

time for repairing

T A þ ð2 þ 2 þ 2Þ þ |{z} 10 þ |{z} 2 þ 0; 1 |{z} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} o

19,1

ð2 þ 2 þ 2Þ þ 10 þ 2 þ |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} |{z} |{z} o

318

TS3

TS3

r5

r5

3!5

r5

TS5

r10

r10z

300 |{z}

time for repairing

10 þ |{z} 2 ð2 þ 2 þ 2Þ þ 10 þ 2 þ ð2 þ 2Þ þ |{z} |fflfflfflffl{zfflfflfflffl} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} |{z} |{z} o z

34

ð2 þ 2 þ 2Þ þ 10 þ 2 þ 0; 1 |{z} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} |{z} |{z} o

18,1

TS3

TS3

r5

r5

TS3

3!5

r10

r10z

T A þ ð2 þ 2 þ 2Þ þ |{z} 10 þ |{z} 2 þ 10 þ 0; 1 |fflfflfflfflfflffl{zfflfflfflfflfflffl} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} o

29,1

ð2 þ 2 þ 2Þ þ 10 þ 2 þ 10 þ 0; 1 |fflfflfflfflfflffl{zfflfflfflfflfflffl} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} |{z} |{z} o

28,1

0!3

 consumers from area A reduce outage duration to 0 [min] by the installation of remote controlled switches (scenarios 2, 3, 4) or reclosers on feeder mid-point (scenarios 10, 11, 12 and 14);  consumers from area C reduce outage duration when supply of restoration with RNOS (scenarios 1, 4, 13 and 14) or changeover (scenarios: 9 and 12) exist.

r5

35

TS3

TS3

0!3

time of automatic sectionalising

52

r10

T A þ ð2 þ 2 þ 2Þ þ |{z} 10 þ |{z} 2 þ þ ð2 þ 2Þ þ |{z} 10 þ |{z} 2 |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} |fflfflfflffl{zfflfflfflffl} o z

0!4

13

TS3

0!3

0

0

r10

r10z

T A þ ð2 þ 2 þ 2Þ þ |{z} 10 þ |{z} 2 þ |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} o

0!3

12

r5

TS5

0!3

0

0

TS5

3!5

T A þ ð2 þ 2 þ 2Þ þ |{z} 10 þ |{z} 2 |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} z

0!3

11

r5

¼ 34 þ 300

0!3

time of automatic sectionalising

10

300 |{z}

2!5

time of automatic sectionalising

9

300 |{z}

time for repairing

time for repairing

p1

time of automatic sectionalising

8

r5

34

TS3

0!3

5

r10z

10 þ |{z} 2 ð2 þ 2 þ 2Þ þ 10 þ 2 þ ð2 þ 2Þ þ |{z} |fflfflfflffl{zfflfflfflffl} |fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl} |{z} |{z} o z 0!3

4

r5o

TS3

r5

r5

r10z

r10z

[9] Popovic´ DS. Power application – a cherry on the top of the DMS cake, DA/DSM DistribuTech Europe, Vienna, Specialist Track 3, session 3, paper 2; 2000. [10] Bibliography of distribution automation. IEEE Trans Power Appar Syst 1984;PAS 103(June):1176–82. [11] Luth J. Four rules to help locate protective devices. Electr World 1991(August): 36–7. [12] Levitin G, Mazal-Tov S, Elmakis D. Optimal sectionalizer allocation in electric distribution systems by genetic algorithm. Electric Power Syst Res 1994(31): 97–102. [13] Miranda V. Using fuzzy reliability in a decision aid environment for establishing interconnection and switching location policies. Proceedings of CIRED; 1991. [14] Teng J-H, Lu C-N. Feeder-switch relocation for customer interruption cost minimization. IEEE Trans Power Deliver 2002;17(1):254–9. [15] Billinton R, Jonnavithula S. Optimal switching device placement in radial distribution systems. IEEE Trans Power Deliver 1996;11(3):1646–51. [16] Celli G, Pilo F. Optimal sectionalizing switches allocation in distribution networks. IEEE Trans Power Deliver 1999;14(3):1167–72. [17] Carvalho P, Ferreira L, Silva A. A decomposition approach to optimal remote controlled switch allocation in distribution systems. IEEE Trans Power Deliver 2005;20(2):1031–6. [18] Teng J-H, Lu C-N. Value based distribution feeder automation planning. Electric Power Energy Syst 2006;28:186–94. [19] Billinton R, Allan R. Reliability evaluation of power systems. Bath: Pitman Publishing; 1984. [20] Nimrihter MD. Reliability indices estimation of distribution circuits by application of distribution automation. DA/DSM DistribuTech Europe 96, Vienna; 1996, p. 547–58. [21] Ghajar RFR, Billinton R. Economic costs of power interruptions: a consistent model and methodology. Electric Power Energy Syst 2006;28:29–35. [22] Su C-L, Teng J-H. Outage costs quantification for benefit – cost analysis of distribution automation systems. Electric Power Energy Syst 2007;29: 767–74.