Vous êtes sur la page 1sur 5

Power Transmission Line Parameter Estimation and Optimal Meter Placement

Yuan Liao, Senior Member, IEEE Department of Electrical and Computer Engineering University of Kentucky, Lexington, KY, USA yliao@engr.uky.edu

Abstract This paper puts forth a framework for estimating the transmission line parameters (including series resistance and reactance and shunt susceptance), line temperature and sag employing synchronized phasors obtained by Phasor Measurement Units (PMU) in real time. The estimated temperature and sag can be utilized for dynamic thermal rating for increased power transfer. The proposed algorithms harness the non-linear optimal estimation theory, and are capable of detecting and identifying bad measurement data, minimizing impacts of measurement errors and thus significantly improving the estimation accuracy. In addition, this paper proposes an optimal scheme for placing PMUs in the system such that a minimum number of PMUs need to be installed in order to determine the parameters, temperature and sag of all the concerned transmission lines in a power network.

I. INTRODUCTION Transmission line parameters including series resistance, series reactance and shunt susceptance are critical inputs to diverse power system analysis programs. Precision of line parameters is thus essential in ensuring the accuracy of obtained analysis results. Especially in areas of power system protection, many distance relaying algorithms entail line parameters for determining appropriate relay settings, calculating fault distance, and making a sound tripping decision [1-2]. Various algorithms for computing transmission line parameters have been presented in the past literature. An online algorithm for estimating the positive sequence line parameters able to make the most of all the measurements available and minimize impacts of measurement and synchronization errors is presented in [1]. The proposed solution is based on voltage and current phasors obtained by PMUs. The distributed parameter line model is harnessed to automatically and fully consider shunt capacitances and distributed parameter effects of long lines. Algorithms for series-compensated lines and parallel lines are described in [2]. Existing line parameter estimation algorithms require measurements from both ends of the line. This paper proposes new possible methods that do not require measurements from every bus. A novel optimal PMU placement scheme will also

be developed to minimize the number of required PMUs or monitoring costs. The other aspect of this paper is related to dynamic line thermal rating. Online estimation of line temperature and sag could enable us to dynamically change the thermal rating for increased power transfer. There is already published literature on the relationship between line temperature, sag and line resistance [3-5]. The average line temperature can be estimated based on the average line resistance, from which the conductor sag can be derived. This type of method will only require appropriate instrumentation to estimate the line resistance, and eliminate the need of extra tension and weather sensors. One distinctive characteristic of the proposed method in this paper is that it does not require PMU on every bus, which further reduces installation and maintenance costs and communication burdens. In the rest of the paper, Section II will present the overall framework of the proposed monitoring scheme. Section III will describe the algorithm for estimating line parameters. Determination of conductor temperature and sag is described in Section IV. The optimal PMU placement scheme is given in Section V, followed by the conclusion. II. OVERALL FRAMEWORK The overall picture of the proposed framework is described in Figure 1. Synchronized voltage and current phasors from certain locations in the system are transferred to a central computer or data concentrator. Accurate transmission line parameters including the resistance, reactance and susceptance are estimated based on synchronized phasors. The temperature of the line conductor is then derived based on the estimated line resistance. Next, the line sag of selected spans can be estimated. The obtained line temperature and line sag can be used as inputs for dynamic line thermal rating applications. Note that the method does not require additional sensors like tension and weather sensors and thus greatly simplify implementation and reduce maintenance costs. The estimated line parameters can also be exploited for relaying applications, and other power system analysis programs such as power flow and angle stability analysis for enhanced performance. As delineated in reference [1], it is shown that placing PMUs at two ends of a transmission line is adequate to accurately estimate line parameters. Moreover, it may not be necessary to place PMU on each bus in order to estimate

978-1-4244-5853-0/10/$26.00 2010 IEEE

250

parameters of interested lines. To facilitate development of different types of algorithms, and development of an optimal PMU placement scheme, as discussed later, we have identified six typical sub-networks that can constitute a large power network, as shown in Figure 2. The oval labeled sub system represents one part of the remaining network. The subnetwork type 4 contains a series compensation device that can be a simple capacitor bank, or more complicated thyristorcontrolled power flow controller. The sub-network type 5 is a double-circuit line that shares a common bus at one terminal. The sub-network type 6 is a double-circuit line that shares a common bus at both terminals. In the figure, the indicated PMUs are sufficient to estimate line parameters for each sub-network. For example, in subnetwork type 3, two PMUs placed at bus P1 and P2 provide enough measurements to estimate line parameters of line P1P3 and P2-P3. The line parameter estimation algorithm for each of the sub-networks will be proposed in the following sections.

sub system P1

P4

sub system P3

P2 sub system

(b) sub-network type 2

sub system P1

P3 sub system

sub system P2

(c) sub-network type 3

sub system P1 P2

sub system

(d) sub-network type 4


P1 CKT 1 P3 CKT 2 P2

sub system

sub system

sub system

Figure 1. Depiction of the overall framework

(e) sub-network type 5


CKT 1

P1 sub system

P2 sub system sub system P1 P2 sub system CKT 2

PMU

(a) Sub-network type 1

(f) sub-network type 6 Figure 2. Common types of sub-networks

251

III. LINE PARAMETER ESTIMATION ALGORITHMS Line parameter estimation algorithm for sub-network type 1 is referred to [1]. A. Algorithm for sub-network type 3 Referring to Figure 2 (c), based on synchronized measurements we can obtain
(i ) (i ) (i ) Vp 1 cosh( 1l1 ) I p1 Z c1 sinh( 1l1 ) V p 2 cosh( 2 l 2 )

The equation contains only the measurements at bus P1 and P2, and can be separated into two real equations. There are three unknown line parameters. Thus, two or more sets of measurements will enable solution of the unknown parameters. In comparison to the algorithm presented in [2], the voltage cross the series compensation device has been eliminated in (2), resulting in a more robust and efficient algorithm. C. Algorithm for sub-network type 5 and 6 For double-circuit line structures, the afore-mentioned algorithms are applicable since generally there is no or negligible coupling between the two circuits in the positivesequence network. On the other hand, special algorithms may be investigated that require fewer meters [2]. Refer to sub-network type 5 shown in Figure 2 (e), where the two circuits share a common remote bus, only two PMUs are needed to be installed at buses P1 and P2 to estimate the parameters of the two circuits of the parallel line, following equation (1). The rationale is similar to that for sub-network type 3. If a PMU can measure the voltages and currents of both circuits, then one PMU installed at either P1 or P2 will be adequate. Equation (1) is a much simpler formulation than the method introduced in [2], since intermediate variables including voltages and currents at remote terminal P3 are eliminated. Hence, the new formulation is a simpler and more efficient method. Refer to sub-network type 6 shown in Figure 2 (f), where the two circuits share both a local and a remote bus. In this configuration, if the two circuits have identical parameters, then we need to have PMUs installed at both ends of the line. If the two circuits have different parameters, then it is possible to derive line parameters using only one PMU, as discussed below. Based on the voltage and current measurements at terminal P1, we can obtain the voltage at the remote bus, utilizing parameters of either CKT1 or CKT2. As a result, we acquire
(i ) (i ) (i ) Vp 1 cosh( 1l ) I1 p1 Z c1 sinh( 1l ) V p1 cosh( 2 l ) (i ) + I2 p1 Z c 2 sinh( 2 l ) = 0

i) I( p2 Z c2

sinh( 2l 2 ) = 0

(1)

i) (i ) V p(1 , I p1 (i ) V p( i2) , I p 2

i th phasor measurement of positive sequence i th phasor measurement of positive sequence

voltage and current at P1; voltage and current at P2;


(i ) (i ) (i ) Ip 1 V p 2 , and I p 2 ;

i = 1,2,..., N , N

being the total


i) V p(1 ,

number of measurement sets, with each set consisting of

Z c , , l characteristic impedance, propagation constant, and


length of the line, respectively.

Z c and are functions of

unknown line parameters: series resistance, series reactance and shunt susceptance per unit length. Note equation (1) contains only the measurements at bus P1 and P2. The subscript 1 and 2 signifies parameters of line P1-P3 and P2-P3, respectively. For each set of measurements, the above equation can be separated into two real equations. There are six unknown line parameters. Thus, three or more sets of measurements will enable solution of the unknown parameters. Determination of the best number of sets of measurements, and the best interval between selected measurements will be investigated in the future. Similarly, the method as introduced in [1-2] can be utilized to consider possible synchronization and measurement errors. Sub-network type 2 can be handled similarly. B. Algorithm for sub-network type 4 This sub-network is a transmission line containing a series compensation device. Referring to Figure 2 (d), based on synchronized measurements we can get
i) (i ) (i ) I( p1 cosh(l1 ) V p1 sinh(l1 ) / Z c I p 2 cosh(l 2 ) (i ) Vp 2 sinh(l 2 ) / Z c = 0

(3)

Where,
i) V p(1 i th phasor measurement of positive sequence voltage at

P1;

I1(ip)1 i th phasor measurement of positive sequence current at


P1 for circuit 1;
(i ) th I2 p1 i phasor measurement of positive sequence current at

(2)

P1 for circuit 2; Z c1 , Z c 2 characteristic impedance of circuit 1 and 2;

Where, l1 and l2 are lengths of the line segment to the left and to the right of the series compensation device, respectively.

1, 2
length.

propagation constant of circuit 1 and 2, and l for line

252

The above equation can be decomposed into two real equations, and we have six line parameters to solve for. Consequently, three or more sets of measurements will bring about the solution. In comparison with the method presented in [2], equation (3) does not contain any intermediate variables such as current and voltage phasors at terminal P2. As a result, the new formulation is more efficient and robust.

D. Algorithm for transmission lines spanning multiple climate zones The algorithm described in previous sections will yield an average temperature of the line, which may be adequate for many applications. To further improve the estimation accuracy, a transmission line may be partitioned into several sections according to the general climate zone information such as moderate climate zone, Appalachian mountain climate zone, etc. Each section is assumed to have its own average temperature. Figure 3 shows a long transmission line that crosses three climate zones. Estimation of parameters of each section is presented as follows.

For each measurement set, we can obtain two complex equations, which can be separated into four real equations. There are three unknown line parameters for each climate zone, and thus there are a total number of nine unknown line parameters. Therefore, three or more sets of measurements will enable calculation of the line parameters. Once resistance for each climate zone is obtained, the according average temperature can be estimated as shown in the next section. In our future research, we will investigate how the climate zones affect line parameters and how many climate zones the method can handle. One possible way to simplify the solution is to assume that the line reactance and susceptance for each section are the same and once estimated can be treated as constants since temperature has little effect on these parameters; then the only unknown parameters will be the resistance of each section. We will scrutinize how well this hypothesis holds. We will examine this simplified method for all subnetwork types in the future. IV. DETERMINATION OF CONDUCTOR TEMPERATURE AND SAG It is established that the line resistance variation with its temperature is practically linear. Based on the estimated resistance, the temperature of the line can be determined as [5]

Figure 3. Transmission line spanning multiple climate zones Let the temperature for each line section be

T = ( R / Rref 1) / + Tref

(6)

T1 , T2 , and T3 ,

respectively. Let the characteristic impedance of each line section be Z c1 , Z c 2 , and Z c 3 , respectively. Let the propagation constant of each line section be respectively. Based on Figure 3, we can derive the following equations [1-2]:
(i ) (i ) (i ) Vp 1 = {[V p 2 cosh( 3l3 ) + I p 2 Z c3 sinh( 3l3 )] cosh( 2 l 2 ) i) (i ) + [I ( p 2 cosh( 3l3 ) + V p 2 sinh( 3l3 ) / Z c 3 ]Z c 2 sinh( 2 l 2 )}cosh( 1l1 ) (i ) i) + {[I ( p 2 cosh( 3l3 ) + V p 2 sinh( 3l3 ) / Z c3 ] cosh( 2l2 ) (i ) (i ) + [V p 2 cosh( 3l3 ) + I p 2 Z c3 sinh( 3l3 )]sinh( 2l2 ) / Z c 2 }Z c1 sinh( 1l1 )

Where T is estimated line temperature and R is estimated resistance; Rref is line resistance at reference temperature Tref ;

is thermo-resistivity coefficient of the

1 , 2 , and 3 ,
l1 , l2 , and l3 ,

respectively. Let the length of each section be

line conductor. Again, the multiple climate zone algorithms that yield the resistance for each zone will produce average conductor temperature for each zone. In our future research, we will examine feasibility of estimating line temperature based on the estimated resistance, and find out meter accuracy requirements for desired temperature estimation. When performing validation studies, various factors such as skin effects that will affect line resistance will be considered. Based on derived line temperature, the sag of a selected span can be estimated as [6]

D = 3S p {Lref [1 + (T Tref )] S p } / 8 (7)


Where,

(4)
i) I( p1 i) (i ) = {[ I ( p 2 cosh( 3l3 ) + V p 2 sinh( 3l3 ) / Z c 3 ] cosh( 2 l 2 ) (i ) (i ) + [V p 2 cosh( 3l3 ) + I p 2 Z c3 sinh( 3l3 )]sinh( 2l 2 ) / Z c 2 }cosh( 1l1 )
(i ) (i ) + {[V p 2 cosh( 3l3 ) + I p 2 Z c 3 sinh( 3l3 )] cosh( 2l2 ) i) (i ) + [I ( p 2 cosh( 3l3 ) + V p 2 sinh( 3l3 ) / Z c3 ]Z c 2 sinh( 2l2 )}sinh( 1l1 ) / Z c1

D is estimated sag, S p is span length, Lref is line

length at reference temperature, and is thermal elongation coefficient of the line conductor. Since the calculation of sag is based on average line temperature. We will investigate how accurate the method is and feasibility for potential applications, by comparing the results with those obtained by finite analysis, and tension based method.

(5)

253

V. DEVELOPMENT OF AN OPTIMAL PMU PLACEMENT SCHEME This task decides optimal locations to install PMUs, such that the number of required PMUs, or the total monitoring costs are minimized, while obtained measurements are sufficient for estimating parameters of all lines. Let us study a network with n buses. Define Wi as a binary variable whose value equals one if a PMU is placed at bus i and equals zero if there is no PMU at bus i , and i = 1,..., n . The optimal PMU placement problem is formulated as an integer linear programming problem with objective function defined as Minimize

triangular part of the Ybus, and a non-zero element will indicate existence of connection between the corresponding buses. Certainly, we need to judge whether such a connection is really a line or a transformer; if it is a line, then we will add a constraint similar to (10). If there are only one single-circuit line and at least one transformer connected to a bus, then a PMU needs to be placed on that bus. If there are only one double-circuit line and at least one transformer connected to a bus, similar analysis as that for sub-network type 5 and 6 needs to be performed to determine meter placement. VI. CONCLUSION This paper proposes a possible framework for monitoring and estimating transmission line parameters including series resistance, series reactance and shunt susceptance by employing synchronized phasors in real time. All the common types of sub-networks are considered. A possible method considering a line spanning multiple climate zones is also designed. Based on estimated line parameters, the average conductor temperature and sag of selected spans can be calculated and utilized for dynamic thermal rating for increased power transfer. The proposed algorithm harnesses the non-linear optimal estimation theory, and is capable of detecting and identifying bad measurement data, minimizing impacts of measurement errors and thus significantly improving the estimation accuracy. The solution is based on the distributed parameter line model and thus fully considers effects of shunt capacitances of the line. In addition, an optimal approach for placing PMUs in the system such that a minimum number of PMUs need to be installed in order to determine the parameters, temperature and sag of all the concerned transmission lines is proposed. One distinctive feature of the proposed methods is that only PMUs are required and no additional sensors such as tension and weather sensors are needed. This will greatly reduce costs and complexity for practical applications. Implementation and verification of the proposed methods are in progress and research results will be reported in the future. VII. REFERENCES
[1] Y. Liao and M. Kezunovic, Online optimal transmission line parameter estimation for relaying applications, IEEE Trans. Power Delivery, vol. 24, no. 1, January 2009, pp. 96-102. Y. Liao, "Some algorithms for transmission line parameter estimation," in Proc. 2009 41st Southeastern Symposium on System Theory, pp. 127320. M. Weibel, W. Sattinger, P. Rothermann, U. Steinegger, M. Zima, G. Biedenbach, "Overhead line temperature monitoring pilot project", CIGRE 2006, Paris, August 27 September 1, 2006. V. T. Morgan, Effects of alternating and direct current, power frequency, temperature, and tension on the electrical parameters of ACSR conductors, Power Delivery, IEEE Transactions on, Vol. 18, No. 3, July 2003, pp. 859 - 866. Marija Bockarjova, Goran Andersson, "Transmission line conductor temperature impact on state estimation accuracy", IEEE Power Tech, Lausanne, Switzerland, July 1-5, 2007. EPRI, Increased Power Flow Guidebook: Increasing Power Flow in Transmission and Substation Circuits, November 2005.

c W
i i =1

(8)

subject to a set of linear constraints, where ci symbolizes the monitoring cost incurred by having a PMU installed at bus When each

i.

ci is equal, or is simply set to one, the problem is

to minimize the total number of PMUs required. Based on the typical sub-networks delineated in Figure 2, the constraints are generated as follows. First, if a bus is connected to the remaining network through a single line, put a PMU at that bus. Second, if a line is a series compensated line, then place PMUs at both terminals of the line. Next, for other lines, only one terminal of the line needs to have a PMU. For example, for the sample system shown in Figure 1, the constraints will be:

W1 + W2 1 W1 + W3 1 W1 + W4 1 W2 + W4 1 W3 + W4 1
(9)

W3 + W6 1 W4 + W5 1 W5 + W6 1
Assuming equal monitoring cost at each location, for the sample system shown in Figure 1, the optimal scheme is determined as {P1, P4, P6}. That is, PMUs need to be placed at bus P1, P4 and P6 in order to monitor parameters of all the eight lines. An automated approach to generate the linear constraints can be developed using the following idea. We will build a topology processor to identify the connectivity between buses. Any two buses, say bus i1 and i2 , that are connected by a line will generate a constraint
Wi1 + Wi2 1

[2] [3] [4]

[5] [6]

(10)

Another possible way is based on the bus admittance matrix (Ybus) of the network. We can scan through the upper

254

Vous aimerez peut-être aussi