Proceedings of the 9th WSEAS/IASME International Conference on ELECTRIC POWER SYSTEMS, HIGH VOLTAGES, ELECTRIC MACHINES

A Network Reduction Method with Synchronized Phasor Measurements

MIHAI GAVRILAS1, GILDA GAVRILAS2, OVIDIU IVANOV1 1 Electrical Engineering Faculty, 2Hidrotechnics Faculty Gheorghe Asachi Technical University of Iasi 53 D. Mangeron Blvd., 700050, Iasi ROMANIA mgavril@ee.tuiasi.ro gildagavrilas@yahoo.com ovidiuivanov@ee.tuiasi.ro
Abstract: - This paper presents a new static network equivalencing technique that take into consideration the influence of changes occurring in the reduced system based on voltage phasors measurements from this system. The method was developed around the REI equivalent, but in principle it can be used with any other equivalencing technique. Numerical simulations have shown that real-time measurements provided by PMUs installed on the buses of the external system can considerably improve simulation results produced by traditional REI equivalents for the operating conditions in the internal, observed power system. Key-Words: - Interconnected power systems, network reduction, network equivalents, REI equivalents, synchronized phasor measurement, phasor measurement unit.

1 Introduction
Present day power systems have complex structures and a high interconnection degree with other neighboring power systems. Their complexity and operating conditions are stressed by the installation of new traditional or dispersed generation sources, with high volatility of the output generation. This trend is also strengthened by the globalization of electricity markets, where electricity is traded at national or regional levels. On the other hand, actual power systems are large or very large systems, known as Wide Area Power Systems (WAPS), whose on-line or off-line analysis and control can be considerably complicated. The most widespread approach to this problem uses network or system equivalents, and the associated methods are known as network equivalencing or network reduction techniques. Such models are widely used, especially when the analysis is focused on the operation of a local power system in interaction with other neighboring systems, for which there is little or none information concerning their structure and operating conditions. Basically, the process of building a network equivalent starts by splitting the original system into an internal subsystem and an external subsystem. The final, reduced system will consist of the internal subsystem, represented in full detail, and the equivalent representation of the external subsystem. This system is to be used to analyze the state of the internal subsystem in different topological and loading conditions. After reduction, the operating conditions in the internal subsystem are exact only for the reference regime, for which the equivalent network was computed.

For other regimes, the operating conditions in the internal subsystem are inherently inaccurate, the approximation error depending on the type and level of contingency considered in the internal subsystem. Besides, if the contingency occurs in the external subsystem, it can be taken into consideration only after recalculating the external subsystem equivalent. Thus, traditional equivalencing techniques do not provide efficient methods to assess the influence of changes occurring in the external subsystem over the operating conditions of the internal one. On the other hand, the advent of Synchronized Phasor Measurement (SPM) technology opens large perspectives to the monitoring and control of WAPS. Due to the high synchronization accuracy of 1 sec or better achieved by the most recent Phasor Measurement Unit (PMU) technologies, the PMU data measurements are far better than traditional SCADA ones. In fact, as suggested in [5], if SCADA represents the old X Ray technology, than PMU is the basic tool of the modern MRI (Magnetic Resonance Imaging) technology. This paper proposes a new static network equivalencing technique that takes into consideration the influence of changes occurring in the external network over the internal system. This is possible using PMUs installed on the buses of the external system. The method was developed around the REI equivalent, but it can be used with any other equivalencing technique. As the voltage phasors of the PMU buses are continuously monitored, this information can be used for a better representation of the external subsystem without the need to recalculate the REI equivalent.

ISSN: 1790-5117

92

ISBN: 978-960-474-130-4

Proceedings of the 9th WSEAS/IASME International Conference on ELECTRIC POWER SYSTEMS, HIGH VOLTAGES, ELECTRIC MACHINES

Satellite

SPM systems are currently used primarily for system monitoring and analysis, in applications like: phase angle, voltage stability and overload monitoring and control, state estimation, congestion management, system restoration and adaptive protection. PMUs are growing in number, and it is anticipated that over the next five years, up to 5,000 PMUs will be deployed at key substations from the power system worldwide [5].

PMU Local PDC Central PDC Fig. 1 Structure of a monitoring system based on SPM (PDC Phasor Data Concentrator).

3 Static Network Equivalents

The widely used static network reduction techniques are based on system splitting into three parts: - Internal system (ISys), which is the system under observations. This is the part of the original system on which the most information is available. This is also that part of system for which detailed operating conditions must be computed. Minimal simplifying assumptions are introduced for the ISys. - External system (ESys) is the part of the original system that is located in the far neighborhood of the ISys. Static network equivalents are used just to represent in a simplified form this part of the system. Simplifying assumptions introduced to represent this equivalent are substantial, but keeping a balance is essential for a good accuracy of operating conditions in the ISys. - Boundary system (BSys) is the part of the original system in the close neighborhood of the ISys. Actually the BSys consists of real or virtual buses on the interconnection lines between ISys and ESys. In most cases the BSys is included in the ISys and the boundary is considered only by the nodes in BSys. J. B. Ward has proposed the static network equivalent named after him in the mid of the 20-th century [13]. Later, in the 1970s, P. Dimo introduced the REI equivalent [2], and a group of researchers from EPRI define the ETI equivalent [9]. After those moments, numerous papers and reports aiming to support the development and enhancement of traditional equivalencing techniques were published in the literature, such as: network equivalent definition based on sensitivity analysis [12]; expansion of REI equivalent model for the analyze and planning of power flow interchanges [8]; use of system equivalents in electricity market applications [10]; adaptation of traditional static equivalents to dynamic system analysis [1]; application of artificial neural networks model to contingency studies based on network equivalents [11]. All types of equivalents are generated based on the nodal equation that describes the relationship between complex voltages and current injections through the nodal-admittance matrix: (1) [Y ] [U ] [I ]

2 Synchronized Phasor Measurement Technology

The technology of SPMs is one of the most interesting present day developments used in the field of real-time monitoring and control of electric power systems [4, 7]. PMUs provide real-time measurement of voltage and current phasors at power system substations that are synchronized using GPS time synchronization signals, as shown in Fig. 1. PMUs are electronic devices that use digital signalprocessing components to measure AC waveforms and convert them into phasors, according to the system frequency, and synchronize these measurements under the control of GPS reference sources. The analog signals are sampled and processed by a recursive phasor algorithm to generate voltage and current phasors. Algorithms to compute phasors from measured signals use a time window of data samples to estimate the phasor parameters. Simple algorithms assume a fixed nominal frequency value and compute only the magnitude and the angle of the phasor. More elaborated algorithms estimate all of the three parameters yielding, in general, more accurate results [3]. A set of PMUs measure the time of zero-crossing of the same phase voltage at different system nodes, using as reference the coordinated universal time (UTC) and GPS synchronization.This time intervals are converted into angles with respect to the nominal frequency (50 or 60 Hz). The differences between these angles offer a system-wide view of the state and the operating conditions of the power system. This technology provides real-time snapshots of the state of the power system, which enables system operators to make decisions to increase system loading while maintaining adequate stability margins, to improve operator response to system contingencies, to reduce congestion costs and enhance state estimation results and to advance system knowledge with correlated event reporting and real-time system visualization.

ISSN: 1790-5117

93

ISBN: 978-960-474-130-4

Proceedings of the 9th WSEAS/IASME International Conference on ELECTRIC POWER SYSTEMS, HIGH VOLTAGES, ELECTRIC MACHINES

ESys
ISys REI bus

ZPBN

Si

S REI

Ui j Ii,G

yi,G UG = 0 yj,G yR,G IR,G G yk,G R Ij,G Ik,G UR SR

BSys

Sj Uj k

Fig. 2 Static network equivalents: Ward (a) and REI (b) equivalents.

If this equation is rewritten applying a partition on the external (E), boundary (B) and internal (I) buses, it becomes:
E B I YEE YFE 0 E YEB YBB YIB B 0 YFI YII I UE UB UI = JE JB JI

Sk Uk

Fig. 3 Definition of the Zero Power Balance Network. U 2 PMU 2bus S2

ISys

(2)

The REI equivalent (REI-Eq) is defined based on eq. (1) and a Gauss elimination schema applied to eq. (2) with a modified form of the matrix [ Y ], partitioned on two more fictitious buses (a ground bus G, and an equivalencing bus R), as in eq. (3):
E G R B I YEE YGE 0 YFE 0 E YEG 0 YEB 0 0 YBB YIB B 0 0 0 YFI YII I UE 0 UR UB UI = 0 0 JR JB JI

S REI REIbus S1
BSys

U 1 PMU 1bus

Fig. 4 Static REI equivalent network with two independent PMU-buses.

YGG -yGR -yRG 0 0 G YRR 0 0 R

(3)

- load delinearization insert fictitious bus R and define the equivalent apparent power SR to replace currents IR,G that sum up ground currents Ip,G, then compute the voltage of the REI bus UR and fictitious admittance yR,G:
I R,G I
p p ,G

S R S p
p

(5)

This time the Gauss elimination procedure is applied to produce a partially triangular matrix [ Y ] on columns E, G, and R. The two fictitious buses G and R are introduced by the so called Zero Power Balance Network (ZPBN), defined as a fictitious, temporary and linear lossless network which links the buses from the ESys that are to be eliminated to the fictitious REI bus(es). Due to the lossless character of the ZPBN, the REI-Eq preserves power losses in the initial and equivalent networks. The ZPBN for a single REI node is built using the following procedure (see also Fig. 3): load linearization nodal apparent powers Sp are replaced by shunt admittances connected to the fictitious ground bus G, yp,G:
y
p ,G

UR

SR I* R,G

S* R R,G 2 UR

(6)

S* p U2 p

I p ,G

S* p U* p

, ( p i , j ,k )

(4)

After applying the Gauss elimination procedure to eq. (3), the final mathematical model can be associated to a network representation like the one in Fig. 2. On the other hand, as one can see from eqs. (4) to (6), if the group of buses that defines an equivalent REI bus contains only one bus (p = 1 in eqs. (4) and (5)), the equivalent REI bus will be represented in the final REIEq by the original bus itself. This property was used in the algorithmic procedure to represent buses in the ESys where PMUs were placed. In this way the nature of PMU-buses is kept unchanged by the Gauss elimination procedure. For instance, Fig. 4 shows the REI equivalent network that uses two PMU-buses in the external system, denoted by PMU 1 and PMU 2.

ISSN: 1790-5117

94

ISBN: 978-960-474-130-4

Proceedings of the 9th WSEAS/IASME International Conference on ELECTRIC POWER SYSTEMS, HIGH VOLTAGES, ELECTRIC MACHINES

3 REI Equivalents with PMU-buses

For the traditional model of a REI equivalent, contingencies can be considered only in the ISys. ESys contingencies cannot be represented unless a complete recalculation of the REI equivalent is run to include the contingency in the mathematical model. The new approach that uses REI equivalents with PMU-buses can cope with ISys contingencies like the traditional one. For ESys contingencies modeling, the basic assumption considers that by installing one or more PMUs in certain buses from the ESys and by continuously monitoring voltage phasors in this buses it is possible to model the external system reaction to different operating conditions for both ISys and ESys. To assess the performances of the proposed equivalencing technique the authors have considered only important contingencies in the ESys, such as loss of a large generating unit or switching off of a highly loaded transmission line. Since the contingency analysis refers exclusively to the effect of different contingencies over the operating conditions from the ISys, the performances of the proposed method were assessed using as performance indices errors for apparent power flows and voltage phasors only in the ISys. Performance indices are computed as:
dS
NC NB 1 ref k Sb Sb NC NB k 1 b 1

[ MVA]

(7) (8)

dU

ref k NC NN U n U n 1 100 NC NN k 1 n 1 U ref n

[%]

where: NC number of contingencies; NN number of node; NB number of branches; Unref voltage phasor in node n, for the reference conditions; Unk voltage phasor in node n, for contingency k; Sbref apparent complex power flow on branch b for the reference conditions; Sbk apparent complex power on branch b for contingency k.

ISys, buses #4 and #5 for the BSys, and buses #6 to #14 for the ESys. Since all buses in the original ESys are PQ buses with small values of the load, the ESys contingency was simulated in bus #13, which became a PU bus with active and reactive generation powers of 100 MW and 30 MVAr respectively. The results of this numerical simulation are presented in Tables 1 to 3, and Fig. 5 and 6. As one can see from Tables 1 and 2, when only one PMU-bus is considered in the ESys, the best results correspond to the case when the PMU is located in the same bus where the contingency is considered (bus #13). However, good approximation accuracy was obtained for all cases when one PMU is installed in the ESys. The conclusions are the same, either the voltage dU or the apparent power dS performance index from eqs. (7) or (8) are used. In fact, figures 5 and 6 show that the shape of the curves dU and dS are similar, hence using one or another of these performance indices is optional. If two or more PMU are installed on the buses of the ESys, their effect over the results produced by the REI equivalent are greatly influenced by their relative position in the network. Moreover, the results are better, compared with the reference conditions only when one PMU is located in the bus where the contingency is considered. For other locations, the results get worse, as shown in Table 3.a for the voltage errors. This behavior could be explained by the fact that the original REI equivalent was computed using completely different operating conditions in the ESys, which can no more be reproduced in real-time conditions when voltages from the PMU-buses are fixed. This behavior will determine worse results as more PMUs are added to the ESys.

4.2 The case of the IEEE 57-bus test system

The one line diagram of the IEEE 57-bus test system can be found in [6]. In this case, the splitting of the original systems was done as follows: ISys (1, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39,40, 42, 43, 44, 45, 47, 48, 50, 52, 53, 54, 56, 57), BSys (7, 15, 18, 41, 46, 49, 51, 55) and ESys (2, 3, 4, 5, 6, 8, 9, 10, 11, 12, 13, 14, 16, 17, 58). Moreover, the original bus #1 was renamed as #58, and a new bus #1 (the slack bus) was added. Contingencies were simulated in bus #58, with maximum active and reactive generated powers of 424 MW and 112 MVAr. Partial results of numerical simulations are presented in Tables 3.b and 4. When using only 1 PMU in the ESys, the conclusion of this analysis is similar to the case of the IEEE 14-bus system. Best results are obtained when the PMU is located in the same bus where the contingency occurs (bus #58). At the same time, choosing as performance index the absolute error of apparent power dS or voltage dU is optional.

4 Numerical Results
The validity of the proposed method for REI equivalent definition using PMU-buses in the ESys was tested on the IEEE 14-bus and IEEE 57-bus test systems. For both systems modified versions of the original on-line diagrams or operating conditions were considered. These changes were imposed by the need for a convenient definition of ISys, BSys and ESys, on one hand, and contingencies in the ESys, on another hand.

4.1 The case of the IEEE 14-bus test system

The one line diagram of the IEEE 14-bus test system can be found in [6]. For this case, the ISys, BSys and ESys were chosen as follows: buses #1, #2 and #3 for the

ISSN: 1790-5117

95

ISBN: 978-960-474-130-4

Proceedings of the 9th WSEAS/IASME International Conference on ELECTRIC POWER SYSTEMS, HIGH VOLTAGES, ELECTRIC MACHINES

Fig. 5 - Power flow errors in the ISys, for different levels of Fig. 6 - Voltage errors in the ISys, for different levels of nodal nodal contingency and different location of 1 PMU (IEEE-14). contingency and different location of 1 PMU (IEEE-14). Table 1 IEEE 14-bus. Power flow errors [MVA] in the ISys, computed with eq. (8), for different levels of nodal contingency (loss of generation unit in bus #13) and different location of 1 PMU, compared with the reference case (w/o PMU). 1 PMU in bus: Contingency Reference level (w/o PMU) 13 12 14 10 11 6 9 -100% 29.110 2.199 6.921 4.208 4.635 4.175 4.277 2.237 -90% 26.000 2.048 6.283 4.020 2.377 2.982 3.683 2.261 -80% 22.939 1.884 6.136 3.944 1.387 2.445 3.171 2.446 -70% 19.928 1.617 5.713 3.836 1.148 2.135 2.892 2.485 -60% 16.968 1.434 5.251 3.708 1.383 2.182 2.561 2.502 -50% 14.053 1.236 4.680 3.461 1.620 2.246 2.205 2.419 -40% 11.182 1.021 4.000 3.088 1.763 2.188 1.823 2.230 -30% 8.355 0.790 3.204 2.580 1.743 1.980 1.412 1.920 -20% 5.570 0.544 2.284 1.923 1.508 1.591 0.973 1.474 -10% 2.831 0.282 1.226 1.090 0.996 0.976 0.500 0.866 Table 2 IEEE 14-bus. Voltage errors [%] in the ISys, computed with eq. (9), for different levels of nodal contingency (loss of generation unit in bus #13) and different location of 1 PMU, compared with the reference case (w/o PMU). 1 PMU in bus: Contingency Reference level (w/o PMU) 13 12 14 10 11 6 9 -100% 0.417 0.225 0.627 0.320 0.462 0.191 0.352 0.170 -90% 0.363 0.208 0.606 0.355 0.190 0.157 0.322 0.199 -80% 0.313 0.189 0.579 0.360 0.067 0.129 0.289 0.218 -70% 0.256 0.161 0.528 0.346 0.047 0.157 0.259 0.221 -60% 0.215 0.142 0.477 0.331 0.063 0.176 0.224 0.222 -50% 0.176 0.122 0.417 0.305 0.097 0.182 0.188 0.214 -40% 0.140 0.100 0.350 0.268 0.116 0.175 0.151 0.196 -30% 0.108 0.077 0.275 0.221 0.119 0.155 0.114 0.167 -20% 0.078 0.053 0.193 0.162 0.104 0.121 0.076 0.127 -10% 0.050 0.027 0.101 0.091 0.069 0.073 0.038 0.074 Table 3.a IEEE 14-bus. Voltage errors [%] in the ISys, computed with eq. (9), for different different location of 2 PMUs, compared with the reference case (w/o PMU). Contingency Reference 2 PMUs in buses: level (w/o PMU) 13 + 6 6 + 14 -100% 0.417 0.006 0.920 -90% 0.363 0.008 0.328 -80% 0.313 0.009 0.234 -70% 0.256 0.010 0.171 -60% 0.215 0.010 0.138 -50% 0.176 0.009 0.108 -40% 0.140 0.008 0.080 -30% 0.108 0.007 0.055 -20% 0.078 0.005 0.034 -10% 0.050 0.003 0.015 Table 3.b IEEE 57-bus. Voltage errors [%] in the ISys, computed with eq. (9), for different different location of 2 PMUs, compared with the reference case (w/o PMU). Contingency Reference 2 PMUs in buses: level (w/o PMU) 58 + 8 2 + 12 4 + 10 -100% 1.293 0.918 1.571 5.028 -90% 1.254 0.801 1.291 3.976 -80% 1.185 0.567 0.877 3.344 -70% 1.126 0.333 0.775 2.969 -60% 1.039 0.351 0.737 1.427 -50% 0.973 0.432 0.799 0.614 -40% 0.906 0.603 0.904 0.224 -30% 0.847 0.756 0.934 0.257 -20% 0.794 0.747 0.885 0.439 -10% 0.628 0.621 0.692 0.541

ISSN: 1790-5117

96

ISBN: 978-960-474-130-4

Proceedings of the 9th WSEAS/IASME International Conference on ELECTRIC POWER SYSTEMS, HIGH VOLTAGES, ELECTRIC MACHINES

Table 4 IEEE 57-bus. Voltage errors [%] in the ISys, computed with eq. (9), for different levels of nodal contingency (loss of generation unit in bus #58) and different location of 1 PMU, compared with the reference case (w/o PMU). 1 PMU in bus: Contingency Reference level (w/o PMU) 58 2 12 3 4 6 10 -100% 1.293 0.350 0.564 1.023 0.881 1.068 0.905 0.403 -90% 1.254 0.381 0.529 0.885 0.731 0.875 0.812 0.291 -80% 1.185 0.468 0.552 0.662 0.471 0.538 0.638 0.173 -70% 1.126 0.565 0.635 0.504 0.267 0.287 0.480 0.248 -60% 1.039 0.617 0.742 0.413 0.211 0.210 0.359 0.389 -50% 0.973 0.641 0.767 0.347 0.272 0.277 0.237 0.489 -40% 0.906 0.656 0.776 0.334 0.394 0.411 0.149 0.552 -30% 0.847 0.669 0.782 0.386 0.506 0.526 0.096 0.583 -20% 0.794 0.681 0.782 0.469 0.610 0.623 0.139 0.587 -10% 0.628 0.653 0.695 0.525 0.776 0.770 0.282 0.594

As to the placement of two or more PMUs in the ESys, the same behavior was observed as in the previous case. However, due to larger size of the system (for instance, the ISys of the IEEE 57-bus system has 43 buses and 46 branches, compared with 5 buses and 7 branches in the previous case, the negative effect of the fixed values of the voltages imposed by PMU measurements is lower, as shown in Table 3.b. Based on the case study presented above, it can be concluded that the applications of real-time measurements provided by PMUs installed in the ESys can considerably improve simulation results produced by traditional REI equivalents. However, if measurement data from two or more PMUs installed in the ESys are to be used, then PMUs location must be optimized for better results. As a general rule, the use of a single PMU in the ESys is sufficient and its best location corresponds to the bus the most exposed to outages, with a high value of generation or load.

5 Conclusion
This paper presents a new static network equivalencing technique that take into consideration the influence of changes occurring in the external network based on voltage phasors measurements in the external systems. The method was developed around the REI equivalent, but in principle it can be used with any other equivalencing technique. Numerical simulations have shown that real-time measurements provided by PMUs installed on the buses of the external system can considerably improve simulation results produced by traditional REI equivalents. References: [1] Baldwin T.L., Mili L., Phadke A.G., Dynamic Ward equivalents for transient stability analysis, IEEE Trans. on Power Systems, No. 1, 1994, pp. 59-67. [2] Dimo P., Nodal Analysis of Power Systems, Abacus Press, Kent, England, 1975.

[3] Donolo M.A., Centeno V.A., A Fast Quality Assessment Algorithm for Phasor Measurements, IEEE Transactions on Power Delivery, vol. 20, no. 4, October 2005, pp. 2407-2413. [4] EPG, Phasor Technology and Real-Time Dynamics Monitoring Systems, Consortium for Electric Reliability Technology Solutions, Electric Power Group, 2006. [5] Giri J., Sun D., and Avila-Rosales R., Wanted: A More Intelligent Grid, IEEE Power and Energy Magazine, Vol. 7, No. 2, Mar-Apr 2009, pp. 34-40. [6] Web address, available at: http://www.cnrcme.ro/foren2008/CD_ROM_1/ s3_ro/s3-15-ro.pdf [7] Jordan J., van Wyk A., van Wyk B., Nonparametric Time-Varying Phasor Estimation Using Neural Networks, Lecture Notes in Computer Science, Volume 4985/2008, June 29, 2008, ISSN 0302-9743, pp. 693-702. [8] Oatts M.L., Erwin S.R., Hart J.L., Application of the REI equivalent for operations planning analysis of interchange schedules, IEEE Trans. On Power Systems, Vol. 5, No. 2, 1990, pp. 547-555. [9] Podmore R., Germond A. Development of dynamic equivalents for transient stability studies, EPRI research Projects 763, April 1979. [10] Rau N.S., Radial equivalents to map networks to market formats an approach using quadratic programming, IEEE Trans. On Power Systems, Vol. 16, No. 4, 2001, pp. 856-861. [11] Semitekos D., Avouris N., Power systems contingencies analysis using ANNs, Workshop on Computer Science and Information Technologies, CSIT 2002, Patras, Greece, CD-ROM. [12] Singh H.K., Srivastava S.C., A Sensitivity Based Network Reduction Technique for Power Transfer Assessment in Deregulated Electricity Environment, Transmission and Distribution Conference and Exhibition 2002: Asia Pacific. IEEE/PES, pp. 1976-1981, 2002. [13] Ward J.B., Equivalent circuits for power-flow studies, AIEE Trans., Vol. 68, pp. 373- 384, 1949.

ISSN: 1790-5117

97

ISBN: 978-960-474-130-4