Long Transmission Lines Fault Location Based on Parameter ...

3 downloads 0 Views 300KB Size Report
Abstract—In this paper, an accurate fault location algorithm based on parameter identification using single-ended data is presented. It is inferred from the ...
Long Transmission Lines Fault Location Based on Parameter Identification Using One-Terminal Data Bin Wang, Jiale Suonan, Heqing Liu, Guobing Song Department of Electrical Engineering Xi'an Jiaotong University Xi'an, Shaanxi, 710049, P.R.China [email protected] Abstract—In this paper, an accurate fault location algorithm based on parameter identification using single-ended data is presented. It is inferred from the differential equations of fault state network and fault component network based on the R-L transmission lines model. And it takes the resistance and inductance of the system behind the fault point, the fault distance and fault resistance as unknown variables. These four unknown variables can be expressed by the coefficients of fault location equations and, therefore, can be obtained by solving those coefficients. When it comes to the distributed parameter transmission lines, an improved algorithm, combining filtering and compensating, is carried out to reduce the model error. Theoretical analysis and simulation results prove that this algorithm is of high accuracy and low time-consumption. Besides, it doesn’t need iteration and the solution is unique. Keywords- long lines; fault location; parameter identification; distributed parameter; compensation algorithm

I.

INTRODUCTION

With the increase of transmission lines voltage level and transmission capacity, modern power systems become increasingly large and complex. Thus the fault location of high voltage transmission lines is becoming more and more important. Finding the fault distance quickly not only contributes to timely lines repairing and stable power supply, but also plays an important role in the power system security, stability and economic operation [1]. The principle of fault location using single-ended data is easy to implement so that it received widespread attentions. Li and Chen [2] proposed a new method for the impedance fault location algorithm using one terminal data. However, this method fails in interphase short circuit fault. Wang and Dong [3] showed a new method for distance measurement while it is lacking of instantaneity. Quan and Li et al. [4] reported a algorithm based on differential equation. But their assumption, that the current flowing through the fault resistance and the zero-sequence current measured by the protection have the same phase, would cause model errors. Another method for accurate transmission lines fault location algorithm using R-L model was investigated in [5]. This method requires the initial value and iteration to solve the nonlinear equations, which calls for a nonignorable time-consumption. Kang and Suonan et al. [6] explored a principle of parameter identification algorithm in the frequency domain. This method also needs to solve the nonlinear equations. Chen and Lv et al. [7] achieved

a method based on parameter identification, but they ignored the influence of long lines’ distributed capacitance. This paper put forward an accurate single-ended fault location algorithm from R-L line model and then utilized in distributed line model. EMTP simulation results prove that this method has high accuracy and low time-consumption. Variations of source impedance, fault types and fluctuations in power frequency have been proved to have little effect on it. II.

THE DERIVATION OF FAULT LOCATION ALGORITHM

First, a transmission lines fault location algorithm of single-phase system using R-L model is presented. Then, algorithm for three-phase system will be given. Finally, algorithm dealing with distributed parameter lines model will be acquired. A. Single-phase system For the sake of simplicity, R-L model transmission lines will be used at first. According to Superposition Theorem, fault network is made of fault state network and fault component network, shown in Fig. 1. Terminal-m is the point where the relay equipment is installed. According to Fig. 1(a), based on the Kirchhoff’s current law at the fault point one obtains:

(a) Fault state network

(b) Fault component network Fig.1 Equivalent circuit of fault condition in system

978-1-4577-0547-2/12/$31.00 ©2012 IEEE

iF = im + in = ( imn + Δim ) + ( −imn + Δin ) = Δim + Δin iF = uF / RF

(1)

(2)

equation and four parameters expressed by coefficients are listed in (9) and (10).

⎡ d ( ima + K L ⋅ 3i0 ) ⎤ uma = x1 ⎢( ima + K R ⋅ 3i0 ) r1 + L1 ⎥ + dt ⎣ ⎦

So the terminal-n’s fault current component is: u Δin = F − Δim RF

⎡ d ( i + K R ⋅ 3i0 ) d 2 ( ima + K L ⋅ 3i0 ) ⎤ x 2 ⎢ ma r1 + L1 ⎥ + dt dt 2 ⎢⎣ ⎥⎦ di du x3 ⋅ im0 + x 4 ⋅ m 0 + x5 ⋅ ma dt dt

(3)

According to the fault component network in Fig. 1(b), based on the Kirchhoff’s voltage law, one obtains: d Δin Ln′ Δin Rn′ + dt d Δim −Δim ( Rm + rp ) − ( Lm + Lp ) = 0 dt

⎧ p = x1 ⎪ ⎪ R ′ = x3 ⋅ ( Lm 0 + L0 p ) − x 4 ⋅ ( Rm0 + r0 p ) n0 x 4 + x3 ⋅ x5 ⎪⎪ ⎨ L ′ = − x5 ⋅ R ′ n0 ⎪ n0 ⎪ x3 ⋅ Rn′ 0 x 4 ⋅ Rn′ 0 + ⎪ RF = 6 ( Rm 0 + r0 p + Rn′ 0 ) 6 ( Lm 0 + L0 p + Ln′ 0 ) ⎪⎩

(4)

Substitute (3) in (4), one obtains: u F Rn′ + +

du F Ln′ = Δim ( Rm + rp + Rn′ ) RF dt

d Δim ( Lm + Lp + Ln′ ) RF dt

(5)

dim Lp dt

(6)

Substitute (6) in (5) the equation can be written as: ⎛ di ⎞ di d 2i ⎛ ⎞ um = x1 ⋅ ⎜ im r + m L ⎟ + x 2 ⋅ ⎜ m r + 2m L ⎟ + ⎜ dt ⎟ dt ⎠ dt ⎝ ⎝ ⎠ d Δim du x3 ⋅ Δim + x 4 ⋅ + x5 m dt dt

(10)

Two-phase short circuit faults represent interphase fault and two-phase-earth-fault. Since they share common fault loops, they can be treated in same ways. So take interphase short circuit faults into consideration. As to two-phase short circuit faults, take mode-1 network as fault component network. The fault location equation is:

In Fig. 1(a), voltage of fault point is: u F = um − im rp −

(9)

(7)

Where x1 − x5 are coefficients of linear equation, the relationship between them and p (fault distance), RF (fault resistance), Rn′ and Ln′ (parameters of remote terminal behind fault) can be written as:

⎡ di di d 2imbc ⎤ ⎡ ⎤ umbc = x1 ⎢imbc r1 + mbc L1 ⎥ + x 2 ⎢ mbc r1 + L1 ⎥ dt dt 2 ⎣ ⎦ ⎥⎦ ⎣⎢ dt d Δimbc dumbc + x3 ⋅ Δimbc + x 4 ⋅ + x5 ⋅ dt dt

(11)

RF should be divided by 2 due to the slight change of fault loop in comparison to the single-phase system. Three-phase short circuit faults can be regarded as twophase short circuit fault, and formula (11) still makes sense.

(8)

C. Distributed parameter transmission lines Still using the algorithm based on R-L model, experiment is conduct to examine the impact of distributed parameter line model: When phase-A to earth fault occurs at 0.205s, the fault point is near the end of the lines (p=260km) and the fault resistance is 200Ω. The calculation results are shown in Fig. 2. It can be seen that this method has a serious error which is caused by distributed capacitance.

According to (7) and (8), solving the five coefficients in (7) can identify the four parameters including fault distance p.

Prefilter is employed to eliminate high-order harmonics. In addition, in order to reduce the model error, long lines compensation is exploited to compensate line on the left side of the fault point. For this purpose, it is needed to acquire an approximate location of the fault from any other less effective algorithm. Detail compensation process is described in [8].

⎧ p = x1 ⎪ ⎪ R ′ = x3 ⋅ ( Lm + Lp ) − x 4 ⋅ ( Rm + rp ) n x 4 + x3 ⋅ x5 ⎪⎪ ⎨ L ′ = − x5 ⋅ R ′ n ⎪ n ⎪ x3 ⋅ Rn′ x 4 ⋅ Rn′ + ⎪ RF = 2 ( Rm + rp + Rn′ ) 2 ( Lm + Lp + Ln′ ) ⎪⎩

B. Three-phase system Single-phase-earth-fault will be discussed first. Suppose phase A is the fault phase, using Clarke matrix to finish the phase-mode transformation and taking mode-0 network as fault component network. Derivation process is similar to that of single-phase system shown in (1)-(8). The fault location

In summary, the algorithm for the distributed parameter is: • • •

Filter the voltage and current data. Calculate the voltage and current at setting point pset . Utilize the algorithm based on R-L lines model to get the relative fault distance p′ .

The fault distance can be calculated by: p = pset + p′ . • This algorithm by its nature is an approximate algorithm. It takes the distributed parameter lines, on the right side of the fault point, as R-L model. And it runs based on an initial fault distance provided by a certain other algorithm.

experiments in cases of different source angles, different fault times and different values of fault resistance have been conduct and indicates that all these changes have no impact on fault location. Tab.1 Simulation results of the parameter identification method-I Fault distance/km

20

100

200

Fig.2 Calculated value of fault distance in distributed parameter lines model based on R-L model’s algorithm

III.

SIMULATION

280

Fault resistance /Ω

Computed location /km A_G

B_C

BC_G

ABC

10

19.990

19.919

20.036

19.918

100

19.995

20.036

20.036

19.981

200

19.996

19.913

20.036

20.073

300

19.997

18.691

20.036

19.921

10

99.983

99.967

99.975

99.967

100

99.989

99.977

99.977

99.995

200

99.984

99.995

99.976

99.977

300

99.966

100.000

99.976

99.982

10

199.976

199.968

199.973

199.969

100

199.981

199.974

199.973

199.982

200

199.988

199.974

199.974

199.978

300

199.990

199.986

199.974

199.729

10

279.971

279.967

279.971

279.967

100

279.976

279.971

279.971

279.977

200

279.977

279.978

279.971

279.958

300

279.975

279.978

279.972

279.786

EMTP and MATLAB are employed to simulate and process data, respectively. Sampled data within 10ms are used and a sliding window analysis is realized. A. Parameter identification algorithm based on R-L model— —method-I

Fig.3 Calculated value of fault distance in single-phase-to-earth fault

It is shown in figure 3 that there is a phase-A earth fault of R-L model lines occurs at 0.2s, while the fault distance is 280km and the fault resistance is 300Ω. It is proved that this algorithm can locate the fault distance rapidly and accurately. The accurate result lasts from 0.2s to about 0.35s, which can be represented by the flat part of the curve. When the lines are regarded as R-L model, the computed fault distance using method-I is shown in Tab.1, while other

Fig.4 The effect of the fault distance

Fig. 4 shows the measured fault distance when singlephase-earth-fault occurs at 20km, 100km, 200km and 280km, respectively, with the fault resistance of 200Ω. It can be concluded from the figure that durations of the accurate fault location result vary: 0.05s, 0.075s, 0.1s to 0.15s.

From the results above, they present a law: The duration of the accurate result increases with fault distance. This law is determined by the time constant of the fault loop response.

• •

Another law can also be found similarly that the duration of the accurate result decreases with the increase of the fault resistance.

• •

B. Parameter identification algorithm based on distributed parameter long line model——the parameter identification method-II Tab.2 Simulation results of method-II Fault distance

Fault resistance

/km

/Ω

60 150 240

A_G

B_C

BC_G

ABC

10

59.747

55.206

55.206

57.461

200

45.793

53.797

55.206

53.495

10

143.802

151.846

148.000

153.811

200

149.332

150.830

153.184

151.427

10

239.957

240.472

240.472

240.441

200

234.474

238.212

240.472

237.778

Tab. 2 shows the result of fault location using the parameter identification method-II. It can be seen that methodII maintains a high accuracy. Its maximum error reaches less than 1% for faults near the terminal end of lines. Tab.3 Fault distance’s impact on method-II Fault distance /km

Fault resistance is10Ω/km

Fault resistance is 200Ω/km

60

61.087

90

93.291

70.739

120

124.322

102.795

210

210.981

220.345

300

301.982

308.196

55.893

Tab. 3 reflects the impact of the fault distance’s change on fault location in case of a phase-A to earth fault. It also indicates that method-II is accurate if the fault occurs at the terminal end of lines. IV.

References [1]

Computed location//km

CONCLUSION

In this paper, a fault location algorithm applicable to long distance transmission lines is proposed. This method has salient features as follows:

It is a time-domain algorithm. This algorithm is of the high accuracy and low timeconsumption. Simulation results prove that this method has a high accuracy and isn’t affected by the variation of system operation modes. This algorithm needs an initial fault distance provided by a certain less effective algorithm.

Ge YZ. New types of protective relaying and fault location: their theory and technique [M]. Xi’an: Xi’an Jiaotong University Press, pp. 217-264, 1996. [2] Li ZM, Chen XY. A novel algorithm for power transmission line fault location using the one terminal current data [J]. Proceedings of CSEE, vol. 17, NO. 6, pp. 416-419, 1997. [3] Wang Bin, Dong Xinzhou, Bo Zhiqian, et al. An impedance fault location algorithm for UHV long transmission lines with single-line-toground faults [J]. Automation of Electric Power Systems, vol. 32, NO. 14, pp. 25-29, 2008. [4] Quan YS, Li P, Zhang Y, Qiu QC, et al. A new one-terminal fault location algorithm based on differential equation [J]. Power System Technology, vol. 28, NO. 21, pp. 47-50, 2004. [5] Suonan JL, Qi J, Chen FF, et al . “An accurate fault location algorithm f or transmission lines based on RL model parameter identification” [J]. Proceedings of the CSE E, vol. 24, NO. 12, pp. 119-125, 2004. [6] Kang Xiaoning, Suonan Jiale. Frequency domain method of fault location based on parameter identification using one terminal data [J]. Proceedings of the CSEE, vol. 25, NO. 2, pp. 22-27, 2005. [7] Chen Kai, Lv Yanping, Luo Zhijuan. Time domain algorithm of fault location based on parameter identification using one terminal data [J]. Electrotechnical Application, vol.28, NO. 23, pp. 40-44, 2009. [8] Song Guobing, Liu Linlin, Suonan Jiale, et al. Long transmission line distance protection based on parameter identification in time domain [J]. vol. 33, NO. 18, pp. 67-70, 2009. [9] Suonan JL, Kang Xiaoning, Song Guobing, Jiao Zaibin. Survey on relay protection using parameter identification [J]. Proceedings of the CSU – EPSA, vol. 19, NO. 1, pp. 14-20, 2007. [10] Xiaoning Kang, Suonan Jiale, Zhiqian Bo, et al. Parameter identification algorithm for fault location based on distributed transmission line model [J]. IEEE Trans. Power Delivery, vol. 15, NO. 4, pp. 1-7, 2008. [11] Song Guobing, Liu Linlin, Suonan Jiale. Study on the frequency band for interpolation method used in distributed parameter line model [J]. Proceedings of the CSEE, vol. 30, NO. 10, pp. 72-76, 2010.