Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Available online at www.sciencedirect.com Electric Power Systems Research 78 (2008) 372–381 Optimised placement of control and protective devices in electric distribution systems through reactive tabu search algorithm Luis G. Wesz da Silva a , Rodrigo A. Fernandes Pereira a , Juan Rivier Abbad b,c , José R. Sanches Mantovani a,∗ a São Paulo State University-UNESP, Department of Electrical Engineering, Avenida Brasil, 56, Zip Code 15385-000 Ilha Solteira, SP, Brazil b Universidad Pontificia Comillas, ICAI, Madrid, Spain c UNSW, Sydney, Australia Received 27 November 2006; received in revised form 6 March 2007; accepted 10 March 2007 Available online 30 April 2007 Abstract Optimised placement of control and protective devices in distribution networks allows for a better operation and improvement of the reliability indices of the system. Control devices (used to reconfigure the feeders) are placed in distribution networks to obtain an optimal operation strategy to facilitate power supply restoration in the case of a contingency. Protective devices (used to isolate faults) are placed in distribution systems to improve the reliability and continuity of the power supply, significantly reducing the impacts that a fault can have in terms of customer outages, and the time needed for fault location and system restoration. This paper presents a novel technique to optimally place both control and protective devices in the same optimisation process on radial distribution feeders. The problem is modelled through mixed integer non-linear programming (MINLP) with real and binary variables. The reactive tabu search algorithm (RTS) is proposed to solve this problem. Results and optimised strategies for placing control and protective devices considering a practical feeder are presented. © 2007 Elsevier B.V. All rights reserved. Keywords: Network protection systems; Restoration; Reliability; Reactive tabu search 1. Introduction Efficient and safe planning of distribution systems includes, among other investments, the placement of control and protective devices in strategic places of the network to improve power supply quality and reliability indices. Control devices include any device that an operator can use to open or close a circuit (either manually or remotely controlled), in order to control power supply interruptions when a permanent fault occurs as well as for the execution of expansion works and preventive maintenance interventions in network components. In this paper, reclosers and isolators (manual and automatic sectionalizing switches) are considered to be control devices. Protective devices include any device that operates automatically to isolate ∗ Corresponding author. Tel.: +55 18 3743 1150; fax: +55 18 3743 1163. E-mail addresses: [email protected] (L.G.W. da Silva), [email protected] (R.A. Fernandes Pereira), [email protected] (J.R. Abbad), [email protected] (J.R. Sanches Mantovani). 0378-7796/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.epsr.2007.03.005 any permanent and/or temporary fault in a feeder. In this paper, fuses and reclosers are considered to be protective devices. In all of these cases, during the planning stage it ought to be foreseen that the network should have a set of switches for the reconfiguration to reduce the area to be disconnected as much as possible [1,2] and, consequently, the loads that will be disconnected when permanent faults occur. In general, the impact caused to customers, due to control and protective device placement in the system, is related to the interruption frequency and duration. Various papers, which deal separately with the optimised placement of control devices (switches) for network restoration [3–7] and protective devices [8–10], are found in literature. Ref. [3] proposes a mathematical model for the optimal placement of control devices in radial distribution systems. In this model, outage, maintenance and investment costs are considered, and the objective is to optimise the relation between sectionalizing switch placement costs and the corresponding benefits related to the improvement of the reliability indices. The optimisation problem is solved using the simulated annealing technique. In the model proposed in [2] for the placement L.G.W. da Silva et al. / Electric Power Systems Research 78 (2008) 372–381 of automatic sectionalizing switches in radial or weakly meshed distribution networks, reliability of supply is the main concern. In [4], a model for the optimal placement of control devices considering the cost of non-supplied energy and the investment cost of sectionalizing switch placement is presented. Repairing times and failure rates of the network and the sectionalizing switches are taken into account. The problem is solved using genetic algorithms. In [6] the switch placement problem is formulated as a combinatorial optimisation problem with a non-linear and nondifferentiable objective function. So as to solve the problem, an optimisation algorithm based on the philosophy of ant colonies is used. In [8], the problem of protection device placement is formulated as a binary (0/1) integer non-linear programming problem that is solved using genetic algorithms. In [9,10], a binary linear programming model is developed for the placement of protection devices, fault locators and sensors in the network, in order to improve the reliability indices of the system. Both technical and economic constraints related to the coordination and selectivity among protection devices and the number of devices to be placed is considered. For its solution, classic multiobjective programming techniques along with fuzzy logic, among others, are proposed. Ref. [12] presents a mixed integer non-linear programming (MINLP) model, with real and binary variables, for the control and protective device placement problem. The model proposed in [12] is solved in two steps: (1) protective devices are placed considering specific constraints for each feeder; (2) maintaining protective devices, placed during the first step, in their position, control devices are optimally placed. Reclosers are considered to be both protective and control devices. This paper presents a novel technique to carry out optimised placement of control and protective devices in the same optimisation process, simultaneously. The problem is modelled through MINLP with real and binary variables. The reactive tabu search algorithm (RTS) is proposed to solve this problem. In order to illustrate the application of this technique, a practical, 13.8 kV, 134 nodes, overhead, three-wire feeder with a delta-grounded wye connection substation transformer, is used. 373 The mathematical model for control and protective devices placement and replacement for system restoration proposed in this work, has been designed using the concept of interruption costs and the historical database of permanent and temporary fault indices in the system [3,4,6,8–10,12]. It is formulated as a combinatorial optimisation problem with a non-linear objective function, non-differentiable, including real and integer variables and a set of linear and non-linear constraints that is detailed following. 2.1. Objective function The proposed objective function is the sum of the fixed cost due to the placement and replacement of switches and protective devices, and the cost of interruptions. In the mathematical model, the type of customer (residential, commercial and industrial), the connected load (L), the permanent (λ) and temporary (γ) fault rates, the costs associated with individual customer interruptions (long- and short-time interruptions), and the feeder’s length is considered. The average cost of interruptions (both temporary and permanent) is used to compute the total cost of interruptions [6,9,10,12]. These costs can be assessed from surveys or simply be based on any pre-established standards imposed by the regulatory agencies with its associated penalization. These standards usually measure the number and duration of the interruptions in the system through, for example, the system average interruption frequency index (SAIFI) and the system average interruption duration index (SAIDI) of distribution systems. The objective function for each feeder i of the distribution system is composed by the following terms: f = FCi + n seci (CIPj + CITj ) (1) j=1 2. Mathematical model Fixed costs (FCi ) are related to the investment for the purchase and installation of control and/or protective devices. In the case of replacement, involved fixed costs are basically due to the repositioning of control and/or protective devices. These costs are written as: FCi = FCPjk Xjk + FCCj Yj (2) Optimising control and protective devices placement and/or replacement for distribution network restoration means the searching for the best points to place different types of devices to minimize the sum of the cost of investment in the network and reduce the interruptions costs. The mathematical model should consider the search for optimal solutions that is both technically and economically viable. For example, fuses will clear faults very cheaply in the short term, but since the number of faults may be large and most of them temporary, protecting the system against temporary faults using reclosers may be cheaper in the long term [11]. The number and types of protective devices to be installed in a particular feeder will depend on the importance of the system, magnitude and type of load, circuit configuration and exposure to risks of different nature, which can lead the system towards permanent or temporary fault conditions. The interruption costs for each feeder i of a given distribution network is given by annual cost of interruption due to the occurrence of a permanent fault (CIPj ) and interruption costs due to the occurrence of temporary faults (CITj ) in the section j of feeder i. The cost of each interruption is calculated taking into account the location of all control and protective devices on the radial distribution feeder. Fig. 1 illustrates a generic distribution system presenting seven feeders as well as the protective and control devices placed on the feeders. The feeder is divided into sections according to the location of the installed devices. That is, every installed device establishes two sections: the upstream section will see the device at its receiving node (at the end of the section) and the downstream section will see the device at its sending node (at the beginning of the section). A fault in any section will j∈Q j ∈ CH 374 L.G.W. da Silva et al. / Electric Power Systems Research 78 (2008) 372–381 Fig. 1. Scheme for section establishments by means of protective and control devices for a generic distribution system. affect the rest of the sections differently, depending on the type of the installed devices. Therefore, the costs of an interruption will depend on the type of the devices installed. This is because the load restoration plan for permanent fault occurrence at each section depends on these devices. Sections of the distribution network are classified according to the device installed at its beginning: • Sections with a recloser (or circuit breaker) at the beginning (dsj = 1): Reclosers are used for switching and protection against permanent and temporary faults. When a permanent fault occurs at this kind of section, the fault is only isolated in this section, whereas customers from downstream sections can be restored (supplied by neighbouring feeders) and customers from upstream sections do not suffer power interruption. Costs of interruptions include the costs of repair and switching (load re-managing for the neighbouring feeders). When a temporary fault occurs at this kind of section, all customers located in-series and downstream to the recloser will experience a temporary outage. The costs of interruptions in these sections are those associated to short interruptions. Customers from upstream sections will not suffer power interruption. In Fig. 1, sections i, iii and viii illustrate this type of section. • Sections with a fuse at the beginning (dsj = 2): Model developed in this work considers fuse cutouts to be placed only for protection purposes. Fuse cutouts have only a protection function and do not present switching capability. In order to isolate a fault that occurs in this kind of section, two sectionalizing switches, one placed upstream and another placed downstream to the fuse cutout, must be opened. Customers from adjacent sections to this switching device can be restored, i.e., other neighbouring feeders can supply them. Costs of interruptions are the costs of repair and switching (load remanaging). When a temporary fault occurs at this kind of section, the closest upstream in-series protective device with automatic reclosing capability will be tripped. All customers located in-series and downstream to this device will experience a temporary outage. Cost of interruptions in these sections are those associated to short interruptions. Customers L.G.W. da Silva et al. / Electric Power Systems Research 78 (2008) 372–381 from sections upstream to the automatic reclosing device will not suffer power interruption. In Fig. 1, sections iv, v and vii illustrate this type of section. • Sections with an isolator at the beginning (dsj = 3): Isolators do not have protective function. When a permanent or temporary fault occurs at this kind of section, an upstream in-series recloser will be tripped, if the fault is within its protection zone, otherwise the closest upstream in-series fuse will blow. If the tripped protective device is a fuse, sections defined by the isolator and defined by the fuse, are isolated. In the case of the protective device being a recloser, if the only section defined by the isolator is isolated, customers from downstream sections can be restored (switched for other neighbouring feeder). Costs of interruptions are the costs of repair and switching (load re-managing). When a temporary fault occurs at this kind of section, the closest upstream inseries protective device with automatic reclosing capability will be tripped. All customers located in-series and downstream of this device and within will experience temporary outage. Cost of interruptions of these sections are those costs associated to short interruptions. Customers in upstream sections will not suffer power interruption. In Fig. 1, sections ii, vi and ix illustrate this type of section. Mathematically the permanent and temporary interruptions costs can be established through the equations: CIPj = n seci 375 of sections upstream to the permanent-faulted section and the cost of interruptions to customers of downstream sections. In this way, the CIP for any section j of a generic feeder can be interpreted as following: • For permanent faults at section j, loads from this section should be isolated from the system to repair the faulted element. This presents a cost associated to long time interruptions. • For permanent faults at section j, in order to analysis the effects of the fault on the loads that not belongs of this section, the type of device that defines section j has to be verified to assign the cost of interruption: (a) For sections defined by reclosers (dsj = 1) or fuses (dsj = 2), loads located upstream to the faulted section do not suffer power supply interruptions since both reclosers and fuses are actuated by permanent faults. To assign the cost of interruption time for loads located downstream to the faulted section the type of device that defines section j has to be verified according the following criteria: if the type of device that defines section j is fuses (dsj = 2), the interruption time for loads located downstream to the faulted section is greater in order to the repairs to be performed. In this model, the possibility of having a fuse λk lk (LRk CRR + LCk CRC + LIk CRI) k=1 ⎧ 0, if dsk = 3 ⎪ ⎪ ⎨ M λk lk + ⎪ (LRm CCR + LCm CCC + LIm CCI), if dsk = 3 ⎪ k=1 ⎩ n seci m=1 ⎧ L ⎪ ⎪ ⎪ (LRm CRR + LCm CRC + LIm CRI), if dsk = 2 ⎪ ⎪ ⎪ ⎪ m=k+1 ⎪ ⎪ ⎧ ⎪ ⎪ ⎪ RS ≥ LTm ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ L ⎨ RS ∈ Ωi ⎪ n seci ⎨ (LRm CCR + LCm CCC + LIm CCI), if dsk = 2 + λk lk ⎪ ⎪ ⎪ m=k+1 ⎪ ⎪ Vi > Vlower ⎪ k=1 ⎪ ⎪ ⎪ ⎪ ⎪ ⎩ ⎪ ⎪ Vi ∈ Ωi ⎪ ⎪ ⎧ ⎪ ⎪ ⎪ ⎪ ⎪ L ⎨ RS < LTm ⎪ ⎪ ⎪ ⎪ RS ∈ Ωi (LRm CRR + LCm CRC + LIm CRI), if ⎪ ⎪ ⎪ ⎩ m=k+1 ⎩ dsk = 2 CITj = γj lj (LRj CTR + LCj CTC + LIj CTI) + γj lj L (3) (LRm CTR + LCm CTC + LIm CTI) + γj lj m=j+1 ⎧ 0, if dsj = 1 ⎪ ⎪ ⎨ MT × ⎪ (LRm CTR + LCm CTC + LIm CTI), if dsj = 1 ⎪ ⎩ m=1 CIPj is the sum of the cost of interruptions to customers of the permanent-faulted section, the cost of interruptions to customers (4) with switching operation capacity is not being considered. The cost is, therefore, associated to the interruption repair time; and, if the type of device that defines section 376 L.G.W. da Silva et al. / Electric Power Systems Research 78 (2008) 372–381 j is reclosers (dsj = 1) or isolators (dsj = 3). Loads located downstream to the faulted section can be re-managed to a neighbouring feeder, if there is enough reserve capacity and the minimal operating voltage magnitude constraint is not violated. In this case, a short-term interruption occurs due to the load re-managing (switching) of the faulted section to neighbouring feeders. The cost is, therefore, associated to an interruption switching time. If load re-managing is not possible, a greater interruption time will be necessary to repair and restore the service at section j. The cost is, therefore, associated to the interruption repair time. (b) Sections defined by isolators (dsj = 3): Some devices (a recloser or a fuse) placed upstream to the faulted section will actuate. However, some time (shorter than the repair time) will be needed to manually open the isolator and to restore the power to upstream sections. The cost is, therefore, associated to the interruption switching time. CITj is the interruption costs due to the occurrence of temporary faults in the feeder and reflects the effects of short power interruption due to the actuation of protective devices with automatic reclosing. This cost is incorporated in the model aimed at minimizing the drawbacks of temporary faults by the adequate placement of these devices. The different terms of CIT for any generic section j of a given distribution system can be interpreted as follows: • For a temporary fault at section j, the closest in-series downstream recloser or substation circuit breaker with reclosing relay is tripped. Therefore, in-series and downstream loads of this protective device are disconnected temporarily due to occurrence of that fault. The cost is, therefore, associated with a short-time interruption. This way for temporary faults at section j, firstly the dsj of the section is verified: if the section is defined by a recloser (dsj = 1) loads from the upstream section j do not suffer alterations in their operation due to the fact that the recloser actuates for temporary faults. In other words, only loads from downstream sections experience an interruption; and, if the section is defined by a isolator (dsj = 3) or fuse (dsj = 2), a protective device with automatic reclosing upstream from this section and in-series with the faulted point should actuate. The cost is, therefore, associated with a short-time interruption. are considered through a fast and efficient distribution network power flow algorithm [13]. Constraints of economical nature are related to the installation and operation cost of devices, as well as the type and importance of loads. 3. Technical solution This section presents the technique employed to solve the optimisation model proposed. 3.1. Selection of candidate points for the placement of control and protective devices The candidate points for the placement of control and protective devices are selected based on rules established from technicians’ and engineers’ knowledge on network operation and planning. Recloser and fuse placement candidate points are the following: • Recloser or fuse placement at the beginning of long lateral branches in addition to the main line reclosers or circuit breakers. • Recloser or fuse placement at the ending node of sections containing important loads that can not suffer reclosing effects and require high service continuity. • Recloser placement at the beginning of laterals that feed loads that can suffer reclosing effects and classified as special or of great importance. • Recloser placement at the beginning of laterals that supply important loads located in areas subjected to a high number of temporary faults and that can suffer reclosing effects. For placing isolators, candidate points are defined considering: • total amount of loads or customers that are part of a given section; • nature and importance of load; • socio-economic characteristics of each type of customer—residential, commercial, industrial, public services, schools and hospitals; • permanent and temporary fault rates for the set of customers from each section of the distribution network. 2.2. Constraints 3.2. Reactive tabu search Constraints that should be incorporated into the optimised placement problem of switch and protective device present technical and economic nature. Constraints of technical nature considered in the proposed model include coordination and selectivity of the protection along with system topology (number of in-series devices), power supply capacity of substations and feeders, feeder capacity limitations, fulfilment of real and reactive power demands, as well as quality of the power supply service, such as voltage profile. In this work, technical constraints related to the operation of the distribution system The mathematical model for protective devices and switches placement and replacement is a MINLP problem and, for its solution, a specialized RTS algorithm [14] is proposed. 3.2.1. Codification A codification in decimal integer base is used for mapping possible solutions for the problem of optimised placement of protective and control devices, as illustrated in Fig. 2. Each section (or branch) of the feeder that is a candidate is assigned L.G.W. da Silva et al. / Electric Power Systems Research 78 (2008) 372–381 377 Fig. 2. Implemented codification scheme—RTS. a value depending on which device has been placed: 0 for no device, 1 for a recloser, 2 for a fuse and 3 for an isolator. The feeder is divided in a main section and lateral branches. It is possible to install any type of device, considered in this paper, in the main section. The lateral branches are subdivided into two categories: first lateral category, where several types of protective devices can be installed, and second lateral category, where only fuses can be installed. Each section has been previously defined as belonging to a certain type that depends on which devices are allowed to be placed in it. Seven types of sections are defined, and the devices allowed in each case are the following: Type 1: reclosers, fuses, and isolators; Type 2: reclosers and isolators; Type 3: reclosers and fuses; Type 4: fuses and isolators; Type 5: fuses; Type 6: reclosers; Type 7: isolators. 3.2.2. Initial configuration The RTS starts from an initial configuration where a certain number of devices of each type is placed in the feeder. A good-quality initial configuration is used as a starting solution to obtain a good algorithm performance. In order to generate this configuration, an exploration of the search space of the feeder under analysis is carried out, so a suitable quantity of control and protective devices can be determined. Maximum quantity of control and protective devices for placement and replacement is fixed and established by technical and economic criteria. Devices are placed randomly considering a placement order according to their importance for the system’s reliability. That is, first reclosers, then isolators and finally fuses are placed at different buses j, satisfying the definition of each type of section (type = 1, 2, . . ., 7), for each one of the i feeders of the system. 3.2.3. Fitness function The RTS algorithm uses a fitness function, that includes the value of the objective function for each set of placed control and protective devices, for which additional terms are added in order to represent penalties assigned to the violation of the power flow constraints in the feeder. 3.2.4. Neighbourhood structure For the RTS algorithm to search for new feasible solutions, a neighbourhood structure has to be defined. The RTS algorithm will look for a better solution than the current one in a set of neighbouring solutions. In this work, two sets of neighbours have been defined. The first set of neighbours N(x) contains all the possible solutions where a single device has been moved from its current location to any other allowed location within the feeder that does not already have another device installed. The second set of neighbours is defined by increasing in one unit, each type of device independently. The search mechanism within this definition of neighbourhood is structured as follows: (i) Random placement of one more recloser in the system. If the number of placed reclosers is greater than the allowed maximum, go to step (ii). Otherwise perform a search in the first set of neighbours N(x). If a better solution is found, go to step (i). (ii) Random placement of one more isolator in the system. If the number of placed isolators is greater than the allowed maximum, go to step (iii). Otherwise perform a search in the first set of neighbours N(x). If a better solution is found, go to step (i). (iii) Random placement of one more fuse in the system. If the number of placed fuses is greater than the allowed maximum, stop the search. Otherwise perform a search in the first set of neighbours N(x). If a better solution is found, go to step (i), if not stop the search. 3.2.5. Tabu list During implementation of the RTS algorithm, two lists are used to store the attributes (number and location of the different devices) of solutions already evaluated: a tabu list (TL) of variable size and a list with long-term memory that stores the attributes of those solutions that are selected more than once throughout the search process. The TL keeps the evaluated solutions of a pre-established number of iterations, considering also a maximum number of devices that can change their positions simultaneously. Solutions evaluated during the search along with the number of times they have been evaluated are stored in the long-term memory. In this way, after the last movement is performed, it can be verified if there are repeated solutions and if so, the number of iterations between the two visits can be calculated. The reactive mechanism increases rapidly the size of the TL when configurations are being repeated. On the other hand, this list can be slowly reduced if, after a specified number of iterations, there are no repeated solutions. In addition to these two mechanisms to make the list grow or shrink, there is another situation where the size of the list might be altered. If the list grows too much, making all movements prohibited (and no aspiration criterion is satisfied), the escape mechanism is used. 378 L.G.W. da Silva et al. / Electric Power Systems Research 78 (2008) 372–381 Fig. 3. A practical, 13.8 kV, 134-node overhead three-wire distribution feeder. 3.2.6. Escape mechanism The escape mechanism is implemented as a random generation of a new initial solution. When several repeated solutions appear, a new random placement of the devices is carried out in order to obtain a current configuration far away from the local optimum solutions that cause the cycling phenomenon. When the escape mechanism is applied the tabu search process is re-initialized. 3.2.7. Stop criterion The stop criterion adopted is a maximum number of allowed iterations. 4. Tests and results The actual development and tests are focused on the placement of control and protective devices in overhead and radial distribution feeders, which present a three-phase main line with lateral branches, some of which may be single or two-phase circuits. The proposed technique has been tested on a 13.8 kV, real-life 134-bus feeder [15]. This feeder has eight interconnection lines with seven neighbouring feeders (two interconnections belong to the neighbour feeder 3), as shown in Fig. 3 and Table 1. Taking into account the criteria presented in Subsection 3.1, thirty-six (36) possible locations were selected for the placement of isolators, thirty-eight (38) for the placement of fuses and eleven (11) for placement of reclosers. Permanent and temporary fault rates used in this test are λ = 0.072 and γ = 0.98 faults/(km year−1 ). Table 1 Capacity of neighbouring feeders (including reserve capacity) Feeder Capacity, kW 2 3 (two connections) 4 5 6 7 8 11,910.00 12,170.00 12,750.00 13,190.00 21,603.00 17,302.00 21,800.00 Table 2 shows the repair and switching costs used in this work, adapted from [16]. The effect of the protective devices due to reclosing sequences are considered by the model through the term called CIT. Considered values for CTR, CTC and CTI are equal to US$ 1.00. Table 3 shows the control and protective devices fixed costs. These costs are composed by costs of devices purchase, material for installation and labour costs practiced in the Brazilian market. A fixed cost of US$ 5,000.00 year−1 for reclosers is considered as able to simulate a life cycle of 5 years. Fixed costs Table 2 Repair cost (RC) and switching cost (SC), by category of customer Category RC, US$/kW (4 h) SC, US$/kW (1.5 h) Residential Commercial Industrial 6 120 18.544 0.4 60 9.62 L.G.W. da Silva et al. / Electric Power Systems Research 78 (2008) 372–381 Table 3 Control and protection equipment fixed costs Device Cost, US$ Recloser Fuse Sectionalizing switch 25,000.00 1,000.00 2,730.00 corresponding to fuses and isolators in Table 3 are not modified to consider the life cycle of 5 years. The considered constraints for the 134-bus feeder are the following: • the maximum number of reclosers is four (4); • there is no limitation for the number of fuses to be placed; • fuses cannot be installed in the main section of the feeder nor in connections with future load currents above 50 A; • there is one circuit breaker between buses 1 and 2; Table 4 Best solutions found 379 • the maximum number of in-series fuses is equal to three (3); • fuses cannot be placed upstream to a recloser; • flow capacity of the feeders and reserve capacity of substations should not be violated. In order to validate the efficiency of the model and the proposed technique, four tests were carried out on the practical feeder. The different tests considered different maximum numbers of reclosers: 0, 1, 3 and 4. The best solution found for the placement of reclosers, isolators and fuses for each test are shown in Table 4. The corresponding temporary interruption costs (CIT), permanent interruption costs (CIP), fixed costs (FC) and total cost of interruption (CI) are also shown in Table 4. 4.1. Comments about performed tests Test 1 shows the important role that the recloser plays in the system’s reliability. In this configuration, the circuit breaker with 380 L.G.W. da Silva et al. / Electric Power Systems Research 78 (2008) 372–381 reclosing relay, placed between buses 1 and 2, is the only device with the capacity to eliminate temporary faults. The sensitivity of this device is limited to its protection zone. Temporary faults outside its protection zone are treated through the actuation of fuses, unnecessarily disconnecting the set of customers within where the fault took place, requiring therefore a network restoration process. In the proposed model, this aspect is considered in the CIT, which presents a high value, compared to the other configurations shown in Table 4. 5. Conclusions In this paper, a novel approach for an integrated placement and replacement of control and protective devices in distribution network feeders was presented. The problem was modelled as a MINLP. This model considers the main actual physical aspects of the problem that directly affect the investment costs and improvement of reliability indices. The reactive tabu search algorithm proposed for solving the MINLP was developed through a codification system and neighbourhood structure that considers the actual physical structure of problem under analysis. The obtained results showed the excellent performance of the algorithm. Of course, as RTS is a metaheuristic algorithm, global optimum cannot be guaranteed. The proposed technique found good-quality solutions, which fulfilled imposed physical and operational constraints, for tests carried out with a practical feeder. For each test, a proposed set of good-quality solutions, including the best solution found and shown in Table 4, was obtained. Taking into account the utility’s operating and maintenance procedures, engineers and technicians can analyse this set of solutions, and they can adopt the most suitable configuration according to their belief. Acknowledgements Authors gratefully acknowledge the financial support from CNPq and Fapesp (Grant no. 02/04811-9), as well as Prof. Dionı́zio Paschoareli Jr. for his support in the revision of this paper. Appendix A. List of symbols CITj CRC CRI CRR CTC CTI CTR dsj FCi FCCj FCPjk lj L LC(·) LI(·) LR(·) LTm M MT n seci Q CCC CCI CCR CH CIPj cost of interruptions to commercial customers within a determined interval time for performing load remanaging into other section; cost of interruptions to industrial customers within a determined interval time for performing load remanaging into other sections; cost of interruptions to residential customers within a determined interval time for performing load remanaging into other section; set of points where the placement of sectionalizing switches is allowed; interruption annual costs due to permanent faults at section j; RS Vi Vlower Xjk Yj Ωi interruption annual costs due to temporary faults at section j; cost of interruptions to commercial customers within a determined interval time for performing repairs in section (·); cost of interruptions to industrial customers within a determined interval time for performing repairs in section (·); cost of interruptions to residential customers within a determined interval time for performing repairs in section (·); cost of interruptions to commercial customers due to occurrence of a temporary fault at section j; cost of interruptions to industrial customers due to occurrence of a temporary fault at section j; cost of interruptions to residential customers due to occurrence of a temporary fault at section j; variable that determines time and cause of the power supplying interruption of section j, taking into account the kind of control and protective device that is placed at the sending node of section j, and ds = 1 for recloser, ds = 2 for fuse and ds = 3 for isolator; fixed costs are related to the investment for the acquisition and installation of control and/or protective devices; installation fixed cost of sectionalizing switch at point j; installation fixed cost of protective device type k at point j; length of section j (km). set of downstream sections of the section j; commercial loads connected at section (·); industrial loads connected at section (·); residential loads connected at section (·); total load that should be transferred to neighbouring feeders in order to isolate a fault in section m; set of upstream sections of the j section for the next section with ds = 1 or 2; set of sections upstream of the section j, up to some section with ds = 1; total number of sections of feeder i. Defined as a function of the quantity of placed devices; set of point where the placement of protective devices is allowed; reserve capacity of feeders in the neighbourhood of feeder i (Ωi ); buses voltage magnitude of the neighbouring feeders of i (Ω); limits of minimum operation allowed for the voltage magnitude on the system buses; decision binary variable for placement of protective devices type k at point j; decision binary variables for placement of sectionalizing switches at point j; neighbouring feeders of feeder i that present interconnection switches placed for operation planning and allow load re-managing of feeder i; L.G.W. da Silva et al. / Electric Power Systems Research 78 (2008) 372–381 λj γi permanent fault rate of section j (fault/(km year−1 )); temporary fault rate of section j (fault/(km year−1 )); References [1] L.-H. Tsai, Networks reconfiguration to enhance reliability of electric distribution systems, Electr. Power Syst. Res. 27 (2) (1993) 135–140. [2] S. Toune, H. Fudo, T. Genji, Y. Fukuyama, Y. Nakanishi, Comparative study of modern heuristic algorithms to service restoration in distribution systems, IEEE Trans. Power Deliv. 17 (1) (2002) 173–181. [3] R. Billinton, S. Jonnavithula, Optimal switching device placement in radial distribution systems, IEEE Trans. Power Syst. 11 (3) (1996) 1646– 1651. [4] G. Celli, F. Pilo, Optimal sectionalizing switches allocation in distribution networks, IEEE Trans. Power Syst. 14 (3) (1999) 1167– 1172. [5] G. Levitin, S. Mazal-Tov, D. Elmakis, Optimal sectionalizer allocation in electric distribution systems by genetic algorithm, Electr. Power Syst. Res. 31 (1994) 97–102. [6] J.-H. Teng, Y.-H. Liu, A novel ACS-based optimum switch relocation method, IEEE Trans. Power Syst. 18 (1) (2003) 113–120. [7] J.-H. Teng, C.-N. Lu, Feeder-switch relocation for customer interruption cost minimization, IEEE Trans. Power Deliv. 17 (1) (2002) 254–259. 381 [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, Electr. Power Syst. Res. 69 (1) (2004) 77–84. [9] F. Soudi, K. Tomsovic, Optimal distribution protection design: quality of solution and computational analysis, Int. J. Electr. Power Energy Syst. 21 (1999) 327–335. [10] F. Soudi, K. Tomsovic, Optimal trade-offs in distribution protection design, IEEE Trans. Power Deliv. (2001) 292–296. [11] P.M. Anderson, Power System Protection, McGraw-Hill and IEEE Press, 1999. [12] L.G.W. Silva, R.A.F. Pereira, J.R.S. Mantovani, Optimized allocation of sectionalizing switches and control and protection devices for reliability indices improvement in distribution systems, in: Proceedings of the Transmission and Distribution Conference and Exposition: Latin America, Sao Paulo, Brazil, October, 2004, pp. 51–56. [13] M.E. Baran, F.F. Wu, Network reconfiguration in distribution systems for loss reduction and load balancing, IEEE Trans. Power Deliv. 4 (2) (1989) 1401–1407. [14] R. Battiti, The reactive tabu search, ORSA J. Comput. 6 (2) (1994) 126–140. [15] Electrical Energy Systems Planning Laboratory Practical 135 Bus Feeder Data. Available: http://www.dee.feis.unesp.br/lapsee/TestSystems/135 Bus Feeder.pdf. [16] A.A. Chowdhury, D.O. Koval, Value-based distribution system reliability planning, IEEE Trans. Ind. Appl. 34 (1) (1998) 23–29.