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
806 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 24, NO. 2, MAY 2009 A Unified Approach for the Optimal PMU Location for Power System State Estimation Nabil H. Abbasy and Hanafy Mahmoud Ismail Abstract—Phasor measurement units (PMUs) are considered as a promising tool for future monitoring, protection and control of power systems. In this paper, a unified approach is proposed in order to determine the optimal number and locations of PMUs to make the system measurement model observable and thereby can be used for power system state estimation. The PMU placement problem is formulated as a binary integer linear programming (BILP), in which the binary decision variables (0, 1) determine whether to install a PMU at each bus, while preserving the system observability and lowest system metering economy. The proposed approach integrates the impacts of both existing conventional power injection/flow measurements (if any) and the possibility of single or multiple PMU loss into the decision strategy of the optimal PMU allocation. Unlike other available techniques, the network topology remains unaltered for the inclusion of conventional measurements, and therefore the network connectivity matrix is built only once based on the original network topology. The mathematical formulation of the problem maintains the original bus ordering of the system under study, and therefore the solution directly points at the optimal PMU locations. Simulations using Matlab are conducted on a simple testing seven-bus system, as well as on different IEEE systems (14-bus, 30-bus, 57-bus, and 118-bus) to prove the validity of the proposed method. The results obtained in this paper are compared with those published before in literature. Index Terms—Integer programming, optimal location, PMUs, state estimation. I. INTRODUCTION T HE commercialization of the global positioning system with accuracy of timing pulses in the order of 1 microsecond has made possible the commercial production of phasor measurement units (PMUs). The PMU is a power system device capable of measuring voltage and current phasor in a power system. Synchronism among phasor measurements is achieved by same-time sampling of voltage and current waveforms using a common synchronizing signal from the global positioning satellite (GPS) [1]–[3]. The way power systems are controlled is ought to be revolutionized by such a new technology. However, the overall cost of the metering system will limit the number and locations of PMUs. In addition to the cost factor, different criteria are suggested for the proper allocation of PMUs in a given system. Network observability, state estimation accuracy and robustness present samples of such criteria. Manuscript received May 27, 2008; revised September 10, 2008. Current version published April 22, 2009. Paper no. TPWRS-00426-2008. The authors are with the Department of Electrical Engineering, College of Technological Studies, Shuwaikh, Kuwait 70654, Kuwait (e -mail: abbassyna@ hotmail.com; [email protected]). Digital Object Identifier 10.1109/TPWRS.2009.2016596 Network observability determines whether a state estimator will be able to determine a unique state solution for a given set of measurements, their location, and a specified network topology. The problem is independent of the measurement errors (or even the measurement values), branch parameters as well as the operating state of the system. There are three basic approaches to conduct network observability analysis; namely numerical, topological, and hybrid approaches. The numerical observability approach is based on the fact that a unique solution for the state vector can be estimated if the gain matrix is nonsingular or equivalently if the measurement Jacobian matrix has a full column rank and well conditioned. The topological observability approach is based on the fact that a network is fully observable if the set of measurements can form at least one measurement spanning tree of full rank [5]. The placement of a minimal set of phasor measurement units to make the system measurement model observable is the main objective of this paper. Due to the high cost of having a PMU at each node, some of the studies performed in the mid 1980s focused on PMU placement and pseudo measurements for complete or partial observability of the system for static and dynamic state estimators [4]–[7]. One of these studies suggested a gradual placement of PMU and provided a methodology for PMU placement with a selected depth of unobservability [7]. This study defined the depth of unobservability as the distance of an unobserved bus to its observed neighbors and provided a methodology for a phased installation of PMUs. A methodology for PMU placement for voltage stability analysis in power system was developed in [8]. Reference [9] introduced a strategic PMU placement algorithm to improve the bad data processing capability of state estimation by taking advantage of PMU technology. In addition, the optimum number of PMUs placement, which makes the power system observable, has been reached through a genetic-based algorithm [10]. An algorithm based on integer programming was successfully implemented to find the optimum number and the locations of the installed PMUs of the power system [11], [12]. In this paper, a unified approach is developed to find the minimum number (least cost) and locations of PMUs, such that the power network is observable, as a priori step for power system state estimation. The developed method accounts for the existing conventional measurements in the mathematical model of the optimal PMU placement strategy; while considering the chance of single or multiple PMU loss in its decision making. The problem is formulated as a binary integer linear programming (BILP) problem. Only the branch-bus model of the network is needed to obtain the minimum number of the PMUs and their locations. Simulation results conducted on a simple testing seven-bus system, as well as the IEEE 14-bus, 30-bus, 57-bus and 118-bus systems are presented. 0885-8950/$25.00 © 2009 IEEE ABBASY AND ISMAIL: A UNIFIED APPROACH FOR THE OPTIMAL PMU LOCATION FOR POWER SYSTEM STATE ESTIMATION II. PMUS AND POWER SYSTEM STATE ESTIMATION PMUs are now used in power systems for many potential applications. The PMUs receive their synchronized signals form the GPS Satellite and are now being manufactured commercially. Their importance comes from the fact that they can provide synchronized measurements of real-time phasor of voltage and currents to the state estimator [2]. A PMU located at any bus can measure the phasor voltage of that bus (magnitude and angle) and as many as needed phasor currents (magnitude and angle) of branches emanating from that bus. Applications of the phase measuring units include; measuring frequency and magnitude of phasors, state estimation, instability prediction, adaptive relaying, and improved control. State estimators are extensively used in modern electric power system utilities control systems to monitor the state of the power system. Various measurements such as complex powers and voltage and current magnitudes received from different substations are fed into the state estimator. Using an iterative nonlinear estimation procedure, the state estimator calculates the power system state. The state (vector) is a collection of all the positive sequence voltage phasors of the network and, from the time the first measurement is taken to the time when the state estimate is available, several seconds or minutes may have elapsed. Because of the time skew in the data acquisition process, as well as the time it takes to converge to a state estimate, the available state vector is at best an averaged quasi-steady-state description of the power system. Consequently, the state estimators available, in this way, in control centers are restricted to steady state applications only. If voltages at all substations are measured at the same instant by using synchronized phasor measurement units, true simultaneous measurements of the power system state could be obtained. It is also sensible to use the positive sequence currents, which provide data redundancy. This leads to a linear estimation of the power system state, which uses both current and voltage measurements. A dynamic state estimator is also obtained by using synchronized phasor measurements. This can be achieved by maintaining a continuous stream of phasor data (voltage and current) from the substations to the control center. In this case, a state vector that can follow the system dynamics can be constructed [2]. In fact, by using PMUs, the state estimator can play an important role in the security of power system operations. III. STATEMENT OF THE PROBLEM A. Without Considering Conventional Measurements [11], [12] A PMU placed at bus will measure the phasor voltage of bus and a predetermined number of phasor currents of the outgoing branches of that bus. The number of the measured phasor currents depends on the number of PMU channels made available. In this paper, we assume that a PMU placed at bus will measure all phasor currents of the branches connected to that bus, in addition to the phasor voltage of bus . Therefore, with the absence of any conventional measurements in the system, bus will be observable if at least one PMU is placed within the 807 set formed by bus and all buses incident to it. Therefore, the problem of optimal PMU placement is one where the objective is to minimize the number of PMUs utilized while preserving the system observability. This objective can be formulated as (1) where is a binary decision variable vector, defined as (2) is a binary network connectivity matrix defined as in (3) (3) . is the vector of PMU cost coefficients, is a vector whose is a entries are all ones, and is the total number of buses. vector function whose entries are non-zeros if the corresponding bus voltage is observable using the given measurement set and zeros otherwise. B. Considering Conventional Measurements In practice, PMUs need to be installed in real systems which are already monitored by conventional injection and/or power flow measurements, in order to enhance the state estimator performance. Therefore, the model presented in the above section needs to be modified to account for the existence of such conventional measurements in the network under study. Reference [12] introduced a method to include conventional measurements in the optimal PMU placement strategy. This method will be referred to as the Individual Bus Merging (IBM) method. A brief description of this method, along with its merits, is given in the following section. Next an alternative proposed method, which will be referred to as the Augmented Bus Merging (ABM) method, will be introduced. 1) The Individual Bus Merging (IBM) Method: This method proposes an approach for determining the optimal PMU locations for systems equipped with conventional measurements. First, an associate set of buses will be defined for each available zero/nonzero injection measurement in the system. This set will be formed by the injection bus and all its associate (connected) buses. Using network equations, the available injection measurement at a particular bus allows one to calculate the phasor voltage of only one bus among its associate set of buses, providing that the phasor voltages of all the remaining buses in that set are known. Therefore, the IBM method suggests that the injection bus to be merged with any one of its associate buses, and to resolve the BILP problem defined by (3) for finding the optimal PMU locations. With this merging process, the number of system buses will be lowered by 1 for each available injection measurement. In addition, the network topology will be altered and the network connectivity matrix will need to be reestablished accordingly. Similarly, the flow measurement in a particular branch allows one to calculate the phasor voltage of one branch terminal bus, providing that the phasor voltage of the 808 Fig. 1. Seven-bus system. other branch terminal bus is known. Therefore, the IBM method suggests combining the observability constraint equations [included in (3)] of the two terminal buses of the flow branch into one constraint equation. Accordingly, the number of constraint equations will be reduced by one for each existing flow measurement, while the problem variables will be unaltered. We have extensively tested the approach explained above for a number of systems and case studies. It is found that the solution of the IBM is sensitive to the selection of the bus to be merged with the injection bus, especially in the presence of flow measurements. For example, refer to the seven-bus system shown in Fig. 1 with one zero injection measurement (denoted by ) exists at bus 3 and a flow measurement (denoted by X) in branch 1–2. Bus 3 was merged with one of its associated buses at a time and the BILP was solved for each case. Results of this case study (presented in Section IV) show that different solutions may be obtained for different selections of the bus merged with the injection bus. The network will be reconfigured as many times as the number of existing injection measurements. Another limitation of this approach is that if the solution results in a PMU to be placed at the merged bus, it will mean that the PMU should be placed at the original injection bus or at its associate merged bus, or at both. In such a case, a topological observability test must be conducted to determine the final decision regarding that particular PMU allocation. 2) Proposed Augmented Bus Merging (ABM) Method: A newly developed method is proposed in this section that incorporates the effect of existing conventional measurements in the formulation of the optimal PMU selection problem. The proposed method makes use of the relaxation provided by the conventional measurements on the observability constraints, while avoiding the possibility of getting different solutions provided by other available methods. The key fact is that the existence of a conventional measurement at a bus would relax the observability conditions imposed by the PMU placement strategy at that particular bus. First, for each bus having an injection measurement, a set of associate buses will be formed by the injection bus and all its connected buses. Therefore, as explained in Section III-B.1, the phasor voltage of only one bus can be calculated using the known injection belonging to at bus and thus need not be directly observed by a PMU. Similarly, the phasor voltage of one terminal bus of a branch having a flow measurement can be calculated providing that the other bus terminal phasor voltage is known. In order to implement these observations, the network buses are reordered, via a properly constructed permutation matrix . Bus reordering is made such that the buses which are not incident to any conventional measurement come first, followed by a set of augmented buses. Each augmented bus corresponds to an available conventional IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 24, NO. 2, MAY 2009 measurement. An augmented bus corresponds to an injection measurement comprises the injection bus and all its associate buses, whereas an augmented bus corresponds to a flow measurement comprises the two terminal buses of that particular flow measurement. The rows of the permutation matrix represent the new bus-order of the system whereas the columns of the permutation matrix represent the original system bus-order. will then be A new system branch connectivity matrix formed as a linear transformation of the original system convia the developed permutation matrix . nectivity matrix of inAdopting the following abbreviations: jection measurements; of flow measurements; of buses not associated to conventional meanumber of conventional measuresurements; ; and of buses associments ated to conventional measurements, we get (4) where is bus-bus incidence matrix for the is buses not incident to conventional measurements, bus-bus incidence matrix for the augmented buses. is bus-bus connectivity matrix for buses not incident to conventional measurements, is bus-bus connectivity and matrix for the augmented buses. The right-hand side vector must be modified accordingly. The new right-hand side vector will be defined as , where is , whose elements is and is . are all equal to 1, In order to reflect the relaxation provided by the existence of is set to be equal conventional measurements, each entry in to the number of buses connected to the injection bus while the are all equal to 1. Finally, the optimal PMU elements of placement problem, considering the existence of conventional measurements can be stated as (5) To illustrate the above formulation, we refer again to the seven-bus system shown in Fig. 1, with one conventional injection measurement placed at bus 3 (referred to as ) and one flow measurement placed in line 4–5. The original system connectivity matrix is In this case, the injection at bus 3 is incident to buses 2, 4, 6, in addition to bus 3, whereas the flow in branch 4–5 is incident ABBASY AND ISMAIL: A UNIFIED APPROACH FOR THE OPTIMAL PMU LOCATION FOR POWER SYSTEM STATE ESTIMATION to buses 4 and 5. Therefore, the set of buses incident to conwhere the ventional measurements will be set of buses not incident to conventional measurements will be . Next, buses 2, 3, 4, and 6 will be augmented into whereas buses 4, 5, 3, and 7 will be a newly created bus The new bus order augmented into a newly created bus and permutation matrix will thus be The right-hand side vector will be and the new connectivity matrix One of the merits of the proposed approach is that it can be easily programmed in a general format using an effective programming tool such as that of Matlab. Thus, it can be applied for systems of any size and topology. The network connectivity matrix is built only once based on the original network topology and need not be reestablished for the inclusion of the conventional measurements. The mathematical formulation of the problem maintains the original bus ordering of the system under study, and therefore the solution directly points at the optimal PMU locations. C. Modeling Considering PMU Loss The number of PMUs proposed by the integer programming problem (5) represents the critical number required to make the power system observable. So far, it has been assumed that this number with its proposed locations will function perfectly. Like any other measuring device, PMUs are prone to failures although they are highly reliable. Therefore, it is necessary to guard against such unexpected failures of PMUs. Reference [12] proposed a method to account for considering single PMU loss in the PMU placement problem. This method will be referred to as the Primary and Backup (P&B) method. A brief description of that method is given in the following section. Next, an alternative proposed method will be presented. This proposed method will be referred to as the Local Redundancy (LR) method. 1) Primary and Backup (P&B) Method: In this method, two independent PMU sets are determined, a primary set and a backup set, where each of which can make the system observable on its own. The primary set of PMUs is determined by building the constraint functions according to the procedure described in Section III and solving the BILP problem. To find the backup set of PMUs, it is suggested that all the variables to be removed in the constraint (3), where bus belongs to the primary set, in order to avoid picking up the same bus which 809 appears in the primary set. Then the BILP problem is solved to obtain the backup set. When trying to explore this approach, it is found that the method maintains the system observability with a single PMU loss either in the primary set or in the backup set. It is important to note that both the primary and backup sets are independent, and each of which, standing alone, can render the system observable. Therefore, this method is also able to preserve the system observability under multiple PMU loss, provided that these multiple PMU loss occur in either the primary set or the backup set, but not in both. Numerical experimentation conducted on this method shows that the method may fail if multiple PMU loss occurs which combines lost PMUs from both the primary and backup sets simultaneously. 2) Proposed Local Redundancy (LR) Method: An alternative method to account for single or multiple PMU loss is proposed in this section. The proposed method attempts to provide a local PMU redundancy to allow for the loss of PMUs while preserving the global network observability. Recall that each entry in the right-hand side vector of (3) is set to 1 in order to guarantee the observability of each bus via at least one PMU. constraint in (3). If the right-hand side of Now, consider the that constraint is changed to 2, it will practically mean that in order for that bus to be observable, at least two PMUs must be installed in the set of buses formed by all buses incident to bus , including bus itself. In other words, we can say that the complex voltage of bus will be “reached” by at least two PMUs. If this concept is extended for all constraints in (3), all elements of the right-hand side vector will be changed to 2. This directly implies that each bus will be allowed to loose, at most, one PMU in its vicinity (set of buses connected to this particular bus including the bus itself) without sacrificing its observability. When the concept is extended to all buses, the network global observability will be directly maintained, with possible single PMU loss. It should be admitted that when the BILP problem is solved in this way, a higher number of PMUs will be expected, and accordingly the cost of metering system will be significantly affected. However; as it is the case in designing any metering system, a tradeoff must be made between the economic restrictions and the required degree of metering system reliability. From the theoretical point of view, the method explained above can be extended for the consideration of multiple PMU loss as well. For instance, the right-hand side of a particular constraint can be set to 3 in order to account for the loss of, at most, two PMUs among the set of PMUs responsible for observing that particular bus. The method can also be made adaptive. Different numbers can be assigned to the right-hand side vector to account for different levels of PMU loss. These numbers can be assigned according to the heaviness of connectivity of each bus in the system. D. Cost Consideration As mentioned before, the PMU is a power system measurement device capable of measuring the synchronized voltage phasor of the bus where it is installed and the current phasors of some or all branches connected to that particular bus. In of all PMUs are assumed to be preceding section the cost equal (at 1 p.u.). It is necessary to consider the unequal cost of 810 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 24, NO. 2, MAY 2009 the PMUs in the formulation of the problem to demonstrate its effect on the number and location of PMUs required keeping the observability of the system. The cost of a PMU depends on a number of factors, including the number of measuring terminals (channels), CT and PT connections, power connection, station ground connection, and GPS antenna connection. However, what really distinguishes between different PMU costs is the number of channels, since the remaining items are common to all PMU installations. Available literature refer that some of the larger PMUs (measuring up to ten phasors plus frequency) cost approximately $30 to $40 thousands of dollars while the smaller ones (measuring from one to three phasors plus frequency) cost considerably less. To account for the effect of unequal PMU cost, in (1) has to be modified. The procedure suggested here is to assume that the cost of a PMU placed at a particular bus connected via a branch to one bus only is set to 1 p.u.. Since the number of PMU channels is dominant in determining the PMU cost, we may assume that this cost increases by a decimal increment p.u., for each additional channel. Then the PMU problem is formulated to determine the minimum number and locations of PMUs required for the system to be observable. A reasonable selection for may be taken as 0.1. Combining the ideas presented in this section, the steps for the implementation of the proposed unified PMU method can be summarized as follows. 1) Read the network branch/bus data. and PMU cost 2) Form the network connectivity matrix coefficient vector . 3) Define the array of buses comprising zero injection mea, and the array of branches comprising surements ; where flow measurements defines the from-to buses where flow measurements exist. . 4) Establish the array of associated buses ; where 5) Establish the array of nonassociated buses , and is the set of all system buses. ; 6) Establish the new bus-order vector is defined as . where 7) Establish the permutation matrix . . 8) Establish the new connectivity matrix 9) Form the new right-hand side vector , . 10) If a single PMU loss is considered set 11) Solve the BILP problem IV. RESULTS AND DISCUSSION The unified approach presented in Section III is programmed on Matlab and case studies are performed on the test seven-bus system (Fig. 1), IEEE 14-bus system (Fig. 2), IEEE 30-bus, 57-bus, and 118-bus systems. A. Validation of the Augmented Bus Merging (ABM) Method The seven-bus system shown in Fig. 1 is used in order to validate the basic results of the present study. Table I summarizes Fig. 2. IEEE 14-bus system. results for different case studies, without considering PMU loss. A comparison is made in Table I between the IBM method and the ABM method proposed in this paper. When ignoring conventional measurements (Case 1) the two methods intuitively yielded identical solution for the optimal PMU number and location. In case 2 the zero injection bus was considered. According to the IBM method, different bus merging for the injection bus 3 was performed. The IBM method in this case suggested buses 2 and 4 for locating the PMUs, for all different selections of the bus merged with the injection bus. However, it is noted that this solution is the same as that solution of case 1 (when the zero injection was ignored), which means that the obtained solution in this case did not make benefit of the available zero injection at bus 3. On the contrary, the proposed ABM method (case 2, columns 4 and 5) provided different optimal locations for PMUs (buses 1 and 4), although it maintains the same minimum number of PMUs. When exploring this latter solution, it is found that the states of buses 1 and 2 will be observed by a PMU located at bus 1, while the states of buses 3, 4, 5, and 7 will be observed via the PMU located at bus 4. The state of the remaining bus 6 can be calculated using the available injection measurement at bus 3. One more advantage in the solution of the proposed ABM method in this case is that the number of utilized PMU channels is 4 compared to 6 channels if PMUs were located at buses 2 and 4. Next, in addition to the injection measurement at bus 3, a flow measurement was then added to branch 1–2 in case 3. Again, different bus merging for the injection bus 3 was performed and the BILP was solved for each one of these bus merging. As shown in Table I case 3, the IBM method gave different PMU locations (buses 3 and 4 and buses 2 and 4) with different bus merging. The results of this case study indicate that the IBM method is sensitive, in terms of the location of PMUs, to the selection of bus to which the injection bus is merged with. On the contrary, the solution provided by the proposed ABM method in this case (buses 1 and 4) is unique; again with a fewer number of utilized PMU channels. ABBASY AND ISMAIL: A UNIFIED APPROACH FOR THE OPTIMAL PMU LOCATION FOR POWER SYSTEM STATE ESTIMATION TABLE I RESULTS FOR THE SEVEN-BUS SYSTEM WITHOUT CONSIDERING PMU LOSS 811 TABLE III RESULTS FOR THE 14-BUS SYSTEM WITHOUT AND WITH CONSIDERING SINGLE PMU LOSS (IGNORING ZERO INJECTION) TABLE IV RESULTS FOR THE 14-BUS SYSTEM CONSIDERING UNEQUAL COSTS OF PMUS (WITHOUT CONSIDERING SINGLE PMU LOSS AND IGNORING ZERO INJECTION) TABLE II RESULTS FOR THE SEVEN-BUS SYSTEM CONSIDERING SINGLE PMU LOSS (IGNORING ZERO INJECTION) TABLE V RESULTS FOR THE 14-BUS SYSTEM WITH CONVENTIONAL MEASUREMENTS B. Results Without Conventional Measurements-Without and With Considering PMUs Loss As stated before, the PMUs placed by the proposed unified approach are assumed to function perfectly. Sometimes, one or more of these PMUs may fail to operate and therefore, it is necessary to guard against such unexpected failures. A part of this paper is to apply the proposed method for single PMU loss to systems with and without conventional measurements. Table II demonstrates results for the seven-bus system, in case of ignoring zero injections, using both the P&B and the proposed LR methods. Results of Table II indicate that the number of PMUs required to guard against such single PMU loss using the proposed LR method is less by 1 than that number obtained by using the P&B method. The optimal PMU placement algorithm, using both the P&B and the LR method, is applied to the IEEE 14-bus system shown in Fig. 2. Zero injections are ignored in this case, and results are shown in Table III. In this case, a complete agreement between the two approaches, regarding the optimal number and locations of PMUs is achieved. The effect of considering unequal costs of PMUs in the placement strategy problem is conducted in this study. As proposed in Section III, the cost of a PMU may be increased by 0.1 p.u. for each additional branch (channel) emanating from its respective bus. Results of this case study are shown in Table IV, without considering single PMU loss and ignoring zero injections. For each case, the number of utilized channels is shown in the same table. These results indicate that the number of PMUs in case of equal and unequal PMU costs is the same (4 PMUs) but the locations are different. Since the overall cost of PMU metering system increases with increasing number of measuring channels, therefore, results of Table IV indicate that the overall cost of PMU metering system may be substantially affected by considering different (unequal) costs for PMUs. C. Results With Conventional Measurements-Without and With Considering PMUs Loss The application of the proposed unified approach to IEEE 14-bus system with conventional measurements is carried out. Results are shown in Table V, without and with single PMU loss consideration. The P&B and proposed LR methods for single PMU loss are applied for the purpose of comparison. The system has only one injection measurement at bus 7 and one flow measurement in branch 5–6. With no PMU loss considered, the optimal number of PMUs is 3 with their locations as indicated in Table V. In case of considering single PMU loss, both the P&B and the proposed LR methods possess the same optimal number of PMUs (which is 7 in this case). However, a slight difference in their locations is depicted (bus 1 in the P&B method is interchanged with bus 5 in the proposed LR method). In order to check the validity of the proposed unified approach for large systems applications, case studies are applied to the 30-, 57-, and 118-bus IEEE systems with the data shown in Table VI. Each system has a number of injection buses but no 812 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 24, NO. 2, MAY 2009 TABLE VI DATA FOR 30-, 57-, AND 118-BUS SYSTEMS TABLE VII OPTIMAL NUMBER OF PMUS REQUIRED FOR THE 30-, 57-, AND 118-BUS SYSTEMS TABLE VIII SIMULATION RESULTS FOR THE 118-BUS SYSTEM (WITHOUT CONSIDERING SINGLE PMU LOSS) Simulation results for the three cases are shown in Table VIII. It is clear from the table that as the number of flow measurements increases, the number of PMUs required keeping the system observable decreases. It can also be noticed, when comparing these results with those previously published in literature that the required number of PMUs is reduced from 29 when considering no flow measurements to 24 when considering 15 flow measurements. A very good agreement between the results obtained using the proposed unified approach and those published before is achieved. As it is clear from the results and as expected, conventional measurements generally reduces the number of PMUs required to maintain the system observable. V. CONCLUSION In this study, a unified approach is proposed for determining the optimal number and locations of PMUs required making the entire power system observable. The proposed unified approach considers the impacts of both existing conventional measurements and the possibility of single or multiple PMU loss into the decision strategy of the optimal PMU allocation problem. The proposed approach is easy in implementation using MATLAB as an effective programming tool. Considering single PMU loss, a new concept (method) is developed which permits single or multiple PMU loss keeping the entire system observable. The effect of PMU meters cost on their optimal number and locations is simulated in this unified approach through a suggested procedure. The developed approach is applied to different IEEE power systems (14-bus, 30-bus, 57-bus, and 118-bus) and results are compared with those published in literature with very good agreement. REFERENCES flow measurements. Two case studies are considered; without considering single PMU loss and with considering single PMU loss. Results are shown in Table VII. As compared to the P&B method, it is clear from Table VII that the proposed LR method yields a lower number of PMUs to maintain the observability of the system. It is also noticed that the number of PMUs in the backup set is greater than that of the primary set in the three systems when using the P&B method. As expected, results of this case study generally reveal that considering single PMU loss has the effect of increasing the number of required PMUs. Simulations using the proposed unified approach for PMU meter locations are carried out on IEEE 118-bus system for fixed number of injection buses and different number of flow measurements. Three cases are studied. In the first case, five flow measurements are considered. Case 2 contains ten flow measurements (5 more than case 1), while case 3 contains 15 flow measurements (5 more than case 2). The locations of the flow measurements for these three cases are shown in Table VIII. The number of injection buses is taken as 10 for the three cases and their locations are as shown in Table VI. These simulations are carried out without considering single PMU loss. [1] R. F. Nuqui and A. G. Phadke, “Phasor measurement unit placement techniques for complete and incomplete observability,” IEEE Trans. Power Del., vol. 20, no. 4, pp. 2381–2388, Oct. 2005. [2] A. G. Phadke, “Synchronized phasor measurements in power systems,” IEEE Comput. Appl. Power, vol. 6, no. 2, pp. 10–15, Apr. 1993. [3] A. G. Phadke, J. S. Thorp, and K. J. Karimi, “State estimation with phasor measurements,” IEEE Trans. Power Syst., vol. 1, no. 1, pp. 233–241, Feb. 1986. [4] A. G. Phadke, J. S. Thorp, and K. J. Karimi, “Real time voltage phasor measurements for static state estimation,” IEEE Trans. Power App. Syst., vol. PAS-104, no. 11, pp. 3098–3107, Nov. 1985. [5] T. L. Baldwin, L. Mili, M. B. Boisen, and R. Adapa, “Power system observability with minimal phasor measurement placement,” IEEE Trans. Power Syst., vol. 8, no. 2, pp. 701–715, May 1993. [6] D. J. Brueni, “Minimal PMU Placement for Graph Observability,” M.S. thesis, Virginia Polytechnic Institute and State University, Blacksburg, VA, 1993. [7] R. Nuqui and A. G. Phadke, “Phasor measurement unit placement based on incomplete observability,” in Proc. IEEE Power Engineering Society Summer Meeting 2002, Jul. 21–25, 2002, vol. 2, pp. 888–893. [8] L. Mili, T. Baldwin, and R. Adapa, “Phasor measurement placement for voltage stability analysis of power systems,” in Proc. 29th Conf. Decision and Control, Honolulu, HI, Dec. 1990. [9] J. Chen and A. Abur, “Placement of PMUs to enable bad data detection in state estimation,” IEEE Trans. Power Syst., vol. 21, no. 4, pp. 1608–1615, Nov. 2006. [10] B. Milosevic and M. Begovic, “Non-dominated sorting genetic algorithm for optimal phasor measurement placement,” IEEE Trans. Power Syst., vol. 18, no. 1, pp. 69–75, Feb. 2003. [11] B. Xu and A. Abur, “Observability analysis and measurement placement for systems with PMUs,” in Proc. 2004 IEEE PES Conf. and Expo., Oct. 10–13, 2004, vol. 2, pp. 943–946. [12] B. Xu, Y. J. Yoon, and A. Abur, “Optimal placement and utilization of phasor measurements for state estimation,” PSERC Pub. 05–20, 2005. ABBASY AND ISMAIL: A UNIFIED APPROACH FOR THE OPTIMAL PMU LOCATION FOR POWER SYSTEM STATE ESTIMATION Nabil H. Abbasy was born 1956. He received the B.Sc. (Hons.) and M.Sc. degrees from the University of Alexandria, Alexandria, Egypt, in 1979 and 1983, respectively, and the Ph.D. degree from Illinois Institute of Technology, Chicago, in 1988. He was an Assistant Professor at Clarkson University, Potsdam, NY, from 1988 to 1989, and then at the University of Alexandria from 1989 to 1994, where he has been a Full Professor since 2000. He has been on leave of absence with the College of Technological Studies, Kuwait, since 1994. His research interests include power systems operation, dynamics, and transients. 813 Hanafy Mahmoud Ismail was born in Cairo, Egypt, in 1956. He received the B.S. and M.S. degrees in electrical engineering from Electrical Engineering Department, Faculty of Engineering at Ain-Shams University, Cairo, in 1979 and 1984, respectively. He received the Ph.D. degree in 1989 from Electrical Engineering Department at the University of Windsor, Windsor, ON, Canada. Since graduation, he has been working at the Electrical Engineering Department, Faculty of Engineering, Ain-Shams University. He is now a Professor, on leave of absence from Ain-Shams University, joining the Electrical Engineering Department at the College of Technological Studies in Kuwait since 1997. He deals mainly with the high voltage power transmission and their associated electrostatics and electromagnetic fields. He is also working in the area of power systems, under ground cables, and fault detection on transmission lines.