Received 25 August 2022, accepted 3 September 2022, date of publication 8 September 2022, date of current version 19 September 2022.

Digital Object Identifier 10.1109/ACCESS.2022.3205020

Economical Setting-Free Double-Ended Fault

Locator for Transmission Lines: Experiences From
Recent Pilot Installations
O. D. NAIDU 1 , (Senior Member, IEEE), SINISA ZUBIC2 , A. V. S. S. R. SAI1 ,
A. N. PRAVEEN 1 , PATRICK COST2 , AND HÅKAN ERIKSSON3
1 Hitachi Energy, Grid Automation, Bengaluru, Karnataka 560048, India
2 Hitachi Energy, Grid Automation, 72159 Västerås, Sweden
3 Svenska Kraftnät (SVK), 172 24 Sundbyberg, Sweden
Corresponding author: O. D. Naidu ([email protected])

1 ABSTRACT Locating fault precisely on power transmission line is highly beneficial to utilities and accurate
2 fault location can expedite the repair of the faulted components, speed-up restoration, and reduce outage time.
3 In this paper, a setting-free double ended fault location method for power transmission line is proposed using
4 both end voltage and current measurements. This algorithm estimates the required setting parameters using
5 pre-fault data and fault location is calculated using estimated parameters and during fault data. The proposed
6 method is verified using the EMTDC simulations for lines connected with conventional and inverter-based
7 renewable resources. The method is not affected by the error in line parameters and renewable integration.
8 The method is implemented in Intelligent Electronic Device (IED) using 1kHz sampling rate and successfully
9 installed on a 400 kV, 233.07 km line from Power Grid Corporation of India (POWERGRID), India and on a
10 220 kV, 265.6 km line from Svenska Kraftnat (SVK), Sweden. This paper presents laboratory test results as
11 well as field experimental results in which line patrolling crews found the actual location. Pilot installation
12 results match the performance demonstrated with laboratory experimental results. The method gives average
13 fault location error of ∼ 0.1 %. The method can locate the fault within two-tower span distance (∼300m)
14 which is comparable with traveling wave-based method that requires 1000 times higher sampling rate, high-
15 cost hardware, more complex commissioning, and settings. These geographically diverse pilot installation
16 results stand as testimony to the fault location accuracy of the proposed setting-free fault location solution.

17 INDEX TERMS Impedance-based fault locator, inverter-based renewable resources, line fault, parameter
18 estimation, renewables, setting free fault location, transmission lines.

19 I. INTRODUCTION turns out to be tedious and time consuming for long transmis- 28

20 Power transmission lines are exposed to faults due to bad sion lines spread across rugged terrains such as hilly areas, 29

21 weather (hurricanes, lightnings), insulation breakdown, short mountains, and deserts etc. [1], [2]. Visual inspection tech- 30

22 circuits by birds, contacting tree branches with a transmission niques are advanced from road patrols to air patrols and, more 31

23 line and other objects. Temporary faults are cleared by auto- recently, to trials with drones and unmanned aerial vehicles 32

24 reclosing. For permanent faults, the power supply is restored etc. and these methods may not be cost effective for long 33

25 only after the maintenance crew detects and replaces the transmission lines. 34

26 failed component. For this purpose, precise fault location Identification of fault location with high precision on 35

27 should be known, else, the fault location identification job power transmission line is of great value to transmission 36

asset owners and maintenance crew as it helps in expediting 37

The associate editor coordinating the review of this manuscript and maintenance work and achieving quick restoration of the line 38
approving it for publication was Ali Raza . [2]. Quick restoration of service improves the reliability of 39

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
VOLUME 10, 2022 96805
 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

40 power supply and reduces the financial loss for the utilities sequence current at the terminal [20]. These assumptions may 96

41 and end users [3]. not hold for lines connected with inverter-based renewables 97

42 Traveling wave (TW)-based fault locators are used to resources as the phase angle is modulated by the inverter con- 98

43 locate the fault within 1-2 towers span distance [4], [5], [6], trols and grid codes [23], [24], [25]. The single-ended fault 99

44 [7], [8], [9], [10], [11], [12], [13], [14], [15]. The perfor- location principles are greatly affected with the integration of 100

45 mance of TW based fault location (FL) methods depends renewables [25]. During fault, the controlled phase angle of 101

46 on accurate detection of wavefront arrival times, sampling inverter-based resources (IBRs) current introduces the high 102

47 rate, and data synchronization [4]. Based on the availabil- non-homogeneity in impedance calculation [25]. This in turn 103

48 ity of input to fault locator, traveling wave-based methods causes large error in the apparent impedance calculation, 104

49 can be classified into two types viz. 1) single-ended [5], especially for resistive faults. This results in the calculated 105

50 [6], [7], [8], and 2) double-ended [4], [9], [10], [11], [12], fault location to be lesser or more than the actual fault location 106

51 [13], [14], [15]. The single-ended TW-based methods use the value. Therefore, single-ended FL methods will not provide 107

52 incident and reflected wave arrival times to locate the fault. the desirable FL accuracy for lines connected with renewable 108

53 The single-ended TW-based techniques pose challenges in power plants [25]. 109

54 identifying the waves reflected from the location of the fault A single-ended-fault location method for grid-connected 110

55 and the remote substation terminals, as well as the waves converter system based on control and protection coordina- 111

56 reflected from the buses of adjacent networks connected to tion is presented in [26]. A single-terminal fault location 112

57 the protected line [5], [8]. Therefore, the practical implemen- method for transmission lines integrated by the inverter-type 113

58 tation of the single-ended TW-methods is challenging and source is proposed in [27]. These methods use the converter 114

59 limited. control information in fault location formulation and solves 115

60 Double-ended TW-based methods [9], [10], [11], [12], the fault location iteratively. These methods may not be easy 116

61 [13], [14], [15] are more accurate and these methods are in to implement in IEDs as they require control information of 117

62 practical usage [4], [10], [11], [12]. However, accuracy of the inverters. 118

63 TW-based FL algorithms depends on traveling wave detec- Present communication technology allows for use of data 119

64 tion, data synchronization, IED hardware, processing of mul- from both ends of the line to calculate the location of the fault 120

65 tiple reflections, filtering of noise, correction of substation [2]. The accuracy of double-ended methods [30], [31] is not 121

66 secondary cable delays [13], and wave speed [14], etc. These affected by fault resistance, loading, fault loop information, 122

67 data synchronization and wave speed errors can be corrected and system non-homogeneity. These methods are economical 123

68 by conducting experiments [15] such as creating a fault at alternative to traveling wave-based methods that requires 124

69 known distance and calculating the secondary cable delays high sampling rates. Rearrangement method of reducing fault 125

70 and wave speed. This process is difficult, requires more engi- location error in tied uncompleted parallel lines is presented 126

71 neering effort and not cost-effective. Besides, they are not in [32]. A robust fault location method using sparse set of 127

72 economical solutions for transmission asset owners due to the digital fault recorders for radial or meshed power networks 128

73 high cost involved in dedicated hardware and solution tuning is presented in [33]. This method requires network level data 129

74 efforts. which is not available in IEDs. The negative sequence-based 130

75 Impedance-based FL methods are most employed by trans- methods are more advantageous as they are not affected by 131

76 mission system operators (TSOs) because of ease of use, low load current and zero sequence impedance [34], [35]. How- 132

77 sampling rate data requirement, and low hardware cost [2], ever, these methods may not be suitable for lines connected 133

78 [3]. These methods are classified into two types viz. 1) Single- with converter interfaced renewable power plants (CIRPPs) 134

79 ended [16], [17], [18], [19], [26], [27], and 2) Double-ended as most of these plants will not supply negative sequence 135

80 [28], [29], [30], [31], [32], [33], [34], [35]. The single-ended current [23], [24], [25]. The positive-sequence impedance- 136

81 methods are commonly available as an inbuilt function in based algorithms are used to locate the faults for lines con- 137

82 the protection IEDs and not required any communication nected with CIRPPs since positive sequence quantities are 138

83 and GPS [2], [19]. However, the precision of single-ended available for lines connected with conventional sources and 139

84 FL algorithms [20], [21], [22] is influenced by the type of CIRPPs. There are several fault location methods for trans- 140

85 fault, arc resistance, impedance angles of line and source, mission lines reported in literature [1], [28], [29], [30], [31], 141

86 mutual compensation, uncertainty of the line parameters, and [32], [33], [34], [35] for utilizing synchronized measurements 142

87 system non-homogeneity. As a result, the accuracy of the from both ends. The accuracy of the two-ended impedance- 143

88 commercially available single-ended impedance-based fault based fault location algorithms depend on the accuracy of line 144

89 locators are in the order of 1.5 to 2.5 percent. For a 300- parameters such as inductance, capacitance, and resistance. 145

90 kilometer transmission line, a ±1.5 percent error still requires The electrical parameters of a transmission line are not known 146

91 9-kilometer length to be inspected (about 30 towers) which is with great precision. It is noted in case studies that there 147

92 not cost-effective. is up to 25-30% of error in actual and the stored values of 148

93 And also, the main assumption of single-ended fault loca- the line parameters [36]. Line parameter with little accuracy 149

94 tor techniques is that the fault current at the FL is in phase with is sufficient for protection applications but for accurate FL, 150

95 fault current at the terminals [16] and in phase with negative precise line parameters are required. These line parameters 151

96806 VOLUME 10, 2022

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

152 are not constant and varies with practical conditions such as
153 weather, age of the conductor etc.
154 The line parameters estimation may help to improve the
155 accuracy of the double-ended fault location. Several methods
156 have been proposed for line parameter estimation for trans-
157 mission lines [37], [38], [39], [40], [41], [42], [43], [44],
158 [45]. The online tracking of line parameter using SCADA FIGURE 1. Two terminal transmission line single line diagram.
159 data is proposed in [37] and this technique requires multiple
160 measurement sets. In [38], multiple measurement scans are (i) parameter estimation using single set of pre-fault 208

161 used with Lagrangian multipliers for improving the accuracy data, (ii) fault location calculation using during fault data, 209

162 of the line parameter estimation. A more accurate calcula- (iii) implementation in IED platform, (iv) experimental val- 210

163 tion of the transmission line parameters can be performed idation using two pilot installations in Indian and Swedish 211

164 by using synchronized current and voltage phasor measure- grid. 212

165 ments provided by the PMUs [39], [40] and these meth- The proposed algorithm is verified through the EMTDC 213

166 ods require multiple data sets which may not be available simulations covering various fault scenarios for 220-kV, 214

167 all the time and it is difficult implement in IED platforms. 240 km line connected with conventional and inverter-based 215

168 To mitigate requirement of multiple data set, a method to renewables resources. The method is implemented in IED and 216

169 estimate the parameters estimation for three-terminal mixed successfully commissioned on a 400-kV 233.07 km line from 217

170 line using pre and during fault phasors is presented in [41]. Power Grid Corporation of India (POWERGRID), India and 218

171 This method requires the negative sequence currents, and it on a 220 kV 256.6 km line from Svenska Kraftnat (SVK), 219

172 will not work for lines connected with CIRPPs. Impedance Sweden. The parameter estimation and fault location algo- 220

173 estimation method for transmission line by assuming linearly rithm, hardware, software requirements, experimental setup, 221

174 changing parameters for short periods is presented in [42]. and results are provided in this paper. The simulation and 222

175 A technique for estimation of electrical parameter of trans- experimental results are consistent, and the proposed algo- 223

176 mission line using synchronized sampled data proposed in rithm can locate the fault within two-tower span distance 224

177 [43]. This method can obtain the steady state values of voltage using 1kHz sampling data which is comparable with traveling 225

178 and current of the transmission line as a function of time and wave-based methods that requires thousand times (∼1MHz) 226

179 line length. In [44], Kalman filter (KF) and extended Kalman higher sampling rates. The proposed method does not require 227

180 filter (EKF) have been applied for state and parameter esti- information about line parameters or settings, saves on engi- 228

181 mation of non-uniform transmission line. A stochastic based neering cost and does not require any additional hardware 229

182 state equations are derived using Telegrapher’s equation for beyond what is required for a modern line differential pro- 230

183 transmission line. Nonlinearities in line voltage and current tection solution [47], i.e., a 2 Mbps communication link. 231

184 have been taken into account by implementing perturbation

185 theory. A data driven based method for detection and identifi- II. PROPOSED METHOD 232

186 cation of line parameters of a transmission line is presented in The paper is organized as follows: (i) parameter estimation, 233

187 [45] using PMU and unsynchronized SCADA measurements. (ii) fault location formulation, (iii) implementation in IED 234

188 Data driven methods may not be suitable for implementation platform, (iv) simulation result studies for lines connected 235

189 in IED platforms. From the literature review, there is a scope with conventional generation and renewable power plants, 236

190 for improvement in double-ended fault locator accuracy by (v) pilot installation in Indian and Swedish grid, and (viii) 237

191 eliminating the dependency on parameters and settings. conclusions. The details of the modules are provided below. 238

192 In this paper, a setting-free fault location method for trans-

193 mission lines using double-ended data is proposed. This A. PROPOSED PARAMETER ESTIMATION METHOD 239

194 algorithm estimates the required setting parameters using This section describes the details of the line parameter esti- 240

195 pre-fault data and fault location is calculated using estimated mation algorithm using a two-step approach. In step one, line 241

196 parameters, and during fault phasors. The proposed algo- parameters are calculated using a lumped model of the line. 242

197 rithm aims to estimate the line parameters using a two-step In step two, the line parameters obtained from step one is 243

198 approach using one set of pre-fault data. Line parameters are given as an initial guess to the least-squares method and the 244

199 calculated using lumped line model in step one. In second accurate line parameters are estimated using complete model 245

200 step, the parameters obtained from step one is given as ini- (distributed parameter model). The entire approach requires 246

201 tial guess and the nonlinear equations formulated using dis- one set of measurement data (pre-fault) only. The details are 247

202 tributed parameter line model are solved. The FL is calculated provided as below. 248

203 using the estimated parameters and during fault phasors.

204 The proposed method does not require information about 1) INITIAL GUESS CALCULATION USING LUMPED MODEL OF 249

205 line parameters or settings and saves the engineering cost THE LINE 250

206 of the solution. The main contributions of this work are as Consider a transmission line connected between buses M and 251

207 follows: N as shown in Fig.1 and two-port lumped π model as shown 252

VOLUME 10, 2022 96807

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

FIGURE 3. Two terminal transmission line distributed model.

FIGURE 2. Two terminal transmission line lumped model.

2) LINE PARAMETER ESTIMATION USING DISTRIBUTED 284

253 Fig.2. From Fig.2, the fault current flowing in the series
1pre MODEL OF THE LINE 285
254 impedance (IML ) is calculated from bus M and N as in (1)
In this section, the line parameters are estimated using the 286
255 and (2) respectively.
distributed model of the line as shown in Fig.3. The objective 287
1pre 1pre lMN  1   1pre  functions are formulated using the basic two-port network
256 IML = IM − YL VM (1) 288
2 relationships. 289
1pre 1pre lMN   
1pre

257 IML = −I N + YL1 VN (2) From the two-port network shown in Fig.3, we have, 290
2
1pre 1pre 1pre 1
258 where, VM = VN cosh(γ lMN ) − IN Zch sinh(γ lMN ) (7) 291
1pre 1pre 1pre
259 VM and VN are pre-fault positive sequence voltage 1pre VN 1pre
1pre 1pre IM = sinh(γ lMN ) − IN cosh(γ lMN ) (8) 292
260 measured at bus M and N respectively; IM and IN are 1
Zch
261 pre-fault positive sequence current measured at bus M and
1pre where, 293
262 N respectively; IML is pre-fault ccurrent flowing through
series impedance; ZL1 and YL1 are positive sequence series
s
263
ZL1
q
264 impedance and shunt admittance per meter of the line respec- 1
Zch = γ = ZL1 × YL1 ; 294

265 tively and lMN is the length of the line in meters. YL1
1
266 Line admittance is obtained as in (3) using (1) and (2), Zch = Characteristic impedance of the line 295

1pre 1pre γ = propagation constant of the line

!
296
2 I + I
267 YL1 = M N
(3)
lMN V 1pre + V 1pre The non-linear equations (7) and (8) are solved by using 297
M N
268 From Fig.2, we have, least square estimation technique to obtain the actual line 298

parameters of the line. The initial guess required for this 299
1pre 1pre 1pre
269 VM − VN = IML lMN ZL1 (4) solution is obtained from the previous step. The detailed steps 300

1pre for solving these equations are mentioned in the previously 301
270 Substitute IML and YL1 in (4), we get (5)
published paper [46]. The line parameters are obtained by 302
1pre 1pre
! !
1pre 1pre 1pre IM + IN 1pre solving the objective functions (7) and (8). This is under- 303
271 VM − VN = IM − 1pre 1pre
VM lMN ZL1 determined problem with set of two equations to solve for 304
VM + VN
four unknowns and that cannot be solved directly. Now let us 305
272 (5)
separate objective function in to real and imaginary parts as 306

273 Series impedance can be obtained as in (6) after simplifica- in (9) and (10), 307

274 tion of (5) 1pre 1pre


1pre 1pre

VMr + jVMi = VNr + jVNi cosh (a + jb) 308
1pre 2 1pre 2
     
1  VM − V N − (c + jd) × sinh (a + jb) 309
ZL1 = (6)

275
1pre 1pre
lMN V 1pre I 1pre − V 1pre I 1pre
 
N M M N
× (INr + jINi ) (9) 310

1pre 1pre 1pre 1pre sinh (a + jb)

276 Line parameters are obtained using the following relations: IMr + jIMi = (V Nr + jVNi ) 311
(c + jd)
imag(ZL1 ) imag(YL1 )
 
1pre 1pre
277 R1 = real(ZL1 ); L1 = ; C1 = − INr + jINi cosh (a + jb) (10) 312
2πf 2πf
278 where, where, 313
1 = c + jd; γ l 1pre 1pre
279 R1 , L 1 , C 1 are positive sequence resistance, inductance, Zch MN = a + jb; VM = VMr + 314
1pre 1pre 1pre 1pre 1pre 1pre 1pre 1pre
280 and capacitance per meter of the transmission line respec- jVMi ; IM = IMr + jIMi ; VN = VNr + jVNi ; IN = 315
1pre 1pre 1pre 1pre 1pre 1pre 1pre 1pre
281 tively; f is system frequency in Hz. These parameters INr + jINi ; VMc = VMcr + jVMci ; IMc = IMcr + jIMci 316

282 obtained using the simplified lumped model of the line are and suffix ‘‘r’’ and ‘‘i’’ represents the real and imaginary 317

283 used as initial guess in the next step. quantities. 318

96808 VOLUME 10, 2022

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

319 Elaborating the hyperbolic functions, we get (11) and (12) The updated value of variable matrix, X in next iteration is- 352

320 cosh (a + jb) = cosh (a) × cos (b) + j sinh (a) × sin (b) X = X = (0) + s (21) 353

321 = +jq (11) Initial guess X = (0) is obtained from step one. 354

322 sinh (a + jb) = sinh (a) × cos (b) + j cosh (a) × sin (b)   
a a0
 
1a

323 = r + js (12)  b   b0   1b 
 =
 c   c0  +  1c 
   (22) 355

324 Substituting (11) and (12) in (9) and (10), we get d d0 1d

 
1pre 1pre 1pre 1pre
325 VMr + jVMi = VNr + jVNi (p + jq) The final parameters are determined using the following rela- 356

1pre 1pre tionships. 357

326 − (c + jd) (r + js) × (INr + jINi )
1
Zch = c + jd; γ = (a + jb)/lMN ; 358
327 (13)
1pre 1pre 1pre 1pre (r + js) ω = 2πf ; ZL1 = Zch
1
× γ = R1 + jX 1 ; 359
328 IMr + jIMi = (V Nr + jVNi )
(c + jd) 1
Y = γ /Zch
1
= G1 + jB ; 1 1
R = Re(ZL1 ); 360

1
329
1pre
− (INr
1pre
+ jINi ) (p + jq) (14) L = Im(ZL1 )/ω; 1
C = Im(YL1 )/ω. 361

330 Separating real and imaginary parts, we get These estimated parameters are used as an input for fault 362

  locator method which is described in the next section. 363

1pre 1pre 1pre 1pre 1pre
331 VMr = pVNr − qVNi cINr − dINi r
  B. DOUBLE ENDED FAULT LOCATION METHOD 364
1pre 1pre
332 + cINi + dINr s (15) This section presents a fault location method for two-terminal 365

lines using both end measurements. The idea is derived from

 
1pre 1pre 1pre 1pre 1pre 366
333 VMi = pV Ni − qV Nr cINr − dINi s
[46] and is developed into a setting free fault location formu- 367
1pre 1pre lation. The main idea is that the fault location formulation is
334 − (cINi + dINr )r (16) 368

1pre cr + ds 1pre 1pre

derived by equating the fault point voltage calculated from 369
335 IMr = 2 V ((cs − dr)/(c2 + d 2 ))VNi both ends of the line. Consider a two-terminal transmission
c + d 2 Nr 370

1pre 1pre line with two IEDs, and each one installed at the terminals 371
336 − pINr + qINi (17) of the transmission line. Let us consider a-g fault on the 372
 
1pre cr + ds 1pre cs − dr 1pre transmission line at a distance ‘d’ from the terminal M. 373
337 IMi = 2 V + 2 VNr
c + d 2 Ni c + d2 Calculate the fault point voltage and current using ABCD 374
1pre 1pre parameters from terminal M
338 − pINi − qINr (18) 375
 1f   1
−B1d
  1f 
VFM Dd VM
339 Four non-linear equations (15)-(18) of the from F(X ) = 1f = (23) 376
IFM −C 1d A1d 1f
IM
340 0, where F = (f 1, f 2, f 3, f 4)T and X = (a, b, c, d)T , are
341 solved for the four unknowns using the least square estima- From (23), calculate the fault point voltage from terminal M 377

342 tion (LSE) method. as in (24) 378

343 For the non-linear set of equations F (X ) = 0, 1f

VFM = D1d VM1f − B1d IM
1f
(24) 379
 −1
344 s = − HT H HT F (19) where, positive sequence A1d , B1d , Cd1 ,
and parameters are D1d 380

defined up to fault point ‘d’ and ABCD parameters are cal- 381
345 where, H is the Jacobian matrix. culated as in (25). 382
346 Therefore equation (19) can be written as-
A1d = D1d = cosh (γ dlMN ) ; B1d = Zch
1
sinh (γ dlMN ) ; 383

1a
  T  −1
H11 · · · H14 H11 · · · H14 1
 1b   . . Cd1 = 1 sinh (γ dlMN ) (25)
. . ...   . . 
384

347
 1c  = −  ..
     .. . . . ..  
Zch

1d H41 · · · H 44 H41 · · · H44 Similarly, calculate the fault point voltage and current from 385

 T  f1  terminal N as in (26) 386

H11 · · · H14
V1f D11−d V1f − B11−d I1f
×  ... . . . ...  
  f2  FN = N N (26) 387

 f3  (20)

348

H41 · · · H44 where ABCD parameters are defined as in (27) 388
f4
A11−d = D11−d = cosh (γ lMN (1 − d)) ; 389
349 where,
B11−d = 1
Zch sinh (γ lMN (1 − d)) ; 390
1pre 1pre 1pre 1pre
350 f 1 = VMr VMcr ; f 2 = VMi VMci ; 1
1pre 1pre 1pre 1pre Cd1 = 1
sinh (γ lMN (1 − d)) (27) 391
351 f 3 = IMr IMcr ; f4 = IMi IMci Zch

VOLUME 10, 2022 96809

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

FIGURE 4. Positive sequence network during fault.

392 Substitute (25) in (24) and (27) in (26), we get (28)

393
1f
VFM = VM1f cosh (γ dlMN ) − IM Zch sinh (γ dlMN ) ;
1f 1

394
1
VFN = VN1f cosh (γ lMN (1 − d))−IN1f Zch
1
sinh (γ lMN (1 − d))
395 (28)
396 Expand the cosh and sinh terms and equate the fault point
397 voltages calculated from both terminals as in (29)
FIGURE 5. Flow chart of the proposed setting-free fault location.
398 VM1f cosh (γ dlMN ) − IM Zch sinh(γ dlMN )
1f 1

399 = VN1f (cosh (γ lMN ) cosh (γ lMN d)

400 − sinh (γ l) sinh (γ lMN d)) − (IN1f Zch
1
)(sinh (γ lMN )
401 × cosh (γ lMN d) − cosh (γ lMN ) sinh (γ lMN d)) (29)
402 Fault location is obtained as in (30) from (29)
 
1 −1 K1
403 d= tanh (30)
γ lMN K2
404 where, K1 = VN1f cosh (γ lMN ) + (I 1f
N Zch )(sinh (γ lMN ) − VM ;
1 1f
1f
405 K2 = VN1f sinh (γ lMN ) − (I 1f 1
N Zch )(cosh
r (γ lMN ) − Zch1I ;
M
1 + jωL 1 


1 R
Zch = jωC 1 ;
406


q
γ = R1 + jωL 1 jωC 1 R1 , L 1 and C 1 are resistance,



407

408 inductance and capacitance per unit length of the line which
409 are estimated using the pre-fault voltage and current mea- FIGURE 6. Pre and during fault voltage and current signals measured at
410 sured at both ends of the line. The proposed method does not (a) Bus M and (b) Bus N for a-g fault at 60km from M.

411 require any settings to locate the fault. The complete algo-
412 rithm steps are depicted in the flow diagram shown in Fig.5.
413 The proposed method is implemented in IED [47] platform faults on a 220-kV, 50Hz, 240km two-terminal transmission 427

414 using 1kHz sampling rate. The practical implementation, line. The simulations are done in the EMTDC software with 428

415 communication network and data synchronization details are frequency dependent transmission line models. The fault data 429

416 provided below. is recorded as COMTRADE99 format and Discrete Fourier 430

Transform (DFT) is applied to obtain the current and voltage 431

417 III. SIMULATION RESULTS AND ANALYSIS phasors. The main inputs to the algorithm are synchronized 432

418 This section provides the illustrative examples, detailed study pre and during fault voltage and current phasors and line 433

419 on effect of variation in power system parameters on FL length. 434

420 accuracy, comparative assessment with existing practically

421 available methods, and results for lines connected with 1) ILLUSTRATIVE CASE 435

422 inverter-based renewables resources. A-g fault case: Consider a-g fault at 60km from bus M on the 436

transmission line with fault resistance of 25, fault inception 437

423 A. SIMULATION STUDY FOR LINES CONNECTED WITH angle is 0 degree and source to line impedance ratio (SIR) is 438

424 CONVENTIONAL SOURCES 1.0. The voltage and current signals are as shown in the Fig.6. 439

425 To test and illustrate the validation of the method in the previ- The calculated (lumped model) and estimated (dis- 440

426 ous section, EMTDC/PSCAD has been used to simulate the tributed line model) line parameters and corresponding errors 441

96810 VOLUME 10, 2022

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

TABLE 3. Study of effect of fault resistance variation on proposed

method.

FIGURE 7. Fault location errors for different fault locations for a-g and
bc-g faults.

TABLE 1. Calculated line parameters using lumped model.

50, 100, 150, 200 and 215 km) are considered to test the 466

effect of fault distances on the proposed solution. Phase a- 467

to-ground (a-g) and double phase to ground (bc-g) faults are 468

considered. In the simulation, the values of the fault inception 469

and fault resistance are set to 600 and 10  respectively. The 470

same simulation system described in the Section A is used 471

TABLE 2. Estimated line parameters using distributed line model.
for simulations. Fig.7 shows the performance of the proposed 472

solution with different fault distances for two types of the 473

faults. The FL errors are less than 0.3% and observed slightly 474

high fault location error for faults closer to fault locator due to 475

high DC component in the voltage and current signals. High 476

DC components in the fault signals affects the estimation of 477

the voltage and current phasors. 478

442 compared to actual line parameters for this case are provided
443 in the Table 1 and 2 respectively. From Table 1 and 2, 2) STUDY OF THE EFFECT OF FAULT RESISTANCE ON THE 479

444 the calculated and estimated parameters are close to actual PROPOSED METHOD 480

445 parameters. The estimated parameters are more accurate than Fault resistance (RF ) is one of the most negative influences on 481

446 calculated and these parameters are used for fault location the impedance-based fault location methods [21]. To evaluate 482

447 calculation. The calculated fault location using the estimated the performance of the proposed method with various fault 483

448 line parameters and post fault phasors is 59.85 km. The abso- resistances, various values of fault resistances 0.1 , 5 , 484

449 lute percentage of fault location error calculated using (31) is 10 , 50 , and 100 , have been considered. The source 485

450 0.06% for the 240 km transmission line for this case. impedance parameters are kept constant, and the fault loca- 486

tions are varied from 10-240 km with an interval of 10km 487

451 %Error
Actual Fault Location − Calculated Fault Location for a-g and bc-g faults. A total of 240 cases are simulated. 488

452 = Table 3 shows the maximum and average fault location errors 489
The total length of the line
for various fault resistances. The maximum FL error is 0.37% 490
× 100 (31)
when fault resistance is 100  for a-g fault. The average
453
491

error is 0.1% which is acceptable. The results show that the 492

454 B. STUDY ON EFFECTIVENESS OF THE PROPOSED accuracy of the proposed method does not change notably 493

455 METHOD FOR VARIOUS FAULT CONDITIONS with change in fault resistance. 494

456 This section presents the effectiveness of the proposed

457 method on variation of different fault conditions (i) influence 3) STUDY OF THE EFFECT OF SOURCE TO LINE IMPEDANCE 495

458 caused by fault location, (ii) effect of fault resistance, and (iii) RATIO ON THE PROPOSED METHOD 496

459 effect of source to line impedance ratios (SIRs). The results The source strength is adversely impacting the impedance- 497

460 are provided below. based FL methods [21]. In this section, two source to line 498

impedance ratios (SIRs) are considered to verify the perfor- 499

461 1) STUDY OF THE INFLUENCE CAUSED BY FAULT LOCATION mance of the proposed method. Various FLs (5, 20, 50, 100, 500

462 This section provides the results for effect of variation fault 150, 200 and 235 km) and two fault types (a-g and bc) are 501

463 location on the proposed method. Location of the fault is considered for EMTDC simulations. In the simulation, the 502

464 one of the negative effects on impedance-based fault location values of the fault inception and fault resistance are set to 503

465 methods [21]. In this section, various fault locations (5, 20, 600 and 20  respectively. Fig.8 shows the performance of 504

VOLUME 10, 2022 96811

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

FIGURE 8. Fault location errors for different source to line impedance

ratio for a a-g fault.
FIGURE 10. Comparative assessment of the proposed method vs practical
methods for variation in the line parameters (a) a-g fault at 60km and
(b) bc-g fault at 180km from Bus M.

resistance, fault type, line to source impedance ratio (SIRs), 526

and loading of the line. 527

D. COMPARATIVE ASSESSMENT WITH PRACTICAL 528

METHODS FOR VARIATION IN LINE PARAMETERS 529

FIGURE 9. Performance of the proposed fault location method for various The practically available single-ended [19] and double-ended 530

fault situations. [30], [31] methods need line parameters as an input. Any 531

error in these parameters will affect the accuracy of the FL 532

TABLE 4. Fault situations considered for simulation study. methods. Such inaccuracies may occur due to aging, tem- 533

perature, sag of the conductor, etc., of the line. This section 534

compares the proposed method with commercially estab- 535

lished single-ended (SEFL) [19] and double-ended positive 536

sequence (DEFL) [30] and double-ended negative sequence 537

(NSFL) [31] methods. For this purpose, we have consid- 538

ered phase-to-ground (a-g) and phase-to-phase-ground (bc-g) 539

faults at 60 and 180km from bus M respectively. The fault 540

resistances considered for a-g and bc-g fault are 20 and 10 541

respectively. 542

Accuracy of the proposed method (PM) is assessed for 543

505 the proposed solution with different source to line impedance different errors in the line parameters (resistance, inductance, 544

506 ratios (1 and 3) for phase a to ground faults. The FL error capacitance) ranging from 0% to 50%. The evaluation results 545

507 notably increases with increase in SIR for faults closer to for a-g and bc-g fault cases are shown in Fig.10 (a) and (b) 546

508 terminals of the line due to high transients in fault signals. respectively. 547

509 High transients will affect the phasor estimation for faults From the figure, the fault location errors of both single- 548

510 closer to line terminals with high source to line impedance ended [19] and double- ended [30], [31] increases with 549

511 ratios. The maximum FL error observed is 0.37% for high increase in the error in line parameters whereas proposed 550

512 SIR case which is acceptable accuracy method accuracy is independent on the parameter variation 551

for both cases. For existing methods, the FL error is greater 552

513 C. DETAILED STUDY USING VARIOUS TEST CONDITIONS than 2% for variation of 10% in line parameters. This requires 553

514 The proposed method is tested with different power system inspection of more towers and is time consuming for the exist- 554

515 fault situations. The fault situations are tabulated in Table 4. ing methods. The FL accuracy claimed in existing methods is 555

516 Total 2304 fault cases are tested, and fault location error is highly dependent on the accuracy of the line parameters. 556

517 presented in Fig.9.

518 From figure 9, the average and maximum fault location E. COMPARATIVE ASSESSMENT WITH PRACTICAL 557

519 errors are 0.1% and 0.6% respectively. The fault location METHODS FOR VARIATION IN SOURCE IMPEDANCES 558

520 error is 0.1% (∼240m) for 1756 (76.21%) cases which is less Practically available single-ended [19] and double ended 559

521 than one tower span (300m) distance for a 240km line. The negative sequence-based methods [31] requires the two-port 560

522 accuracy is comparable to traveling wave-based fault locators equivalent (source) impedance as a setting. These source 561

523 [9], [10], [11], [12], [13], [14] which require ∼1000 times impedances vary with structural and operational changes in 562

524 higher sampling rates and high communication bandwidth. power system. This section provides the performance of the 563

525 The fault location accuracy is not influenced by the fault available and proposed fault location methods for variation 564

96812 VOLUME 10, 2022

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

FIGURE 11. Comparative assessment of the proposed method vs practical

methods for variation in the source impedance for (a) a-g fault at 50km
and (b) bc-g fault at 150km from bus M.

FIGURE 13. Pre and during fault voltage and current signals measured at
(a) Bus M and (b) Bus N for a-g fault at 48km from M.

TABLE 5. Calculated line parameters using 5 model.

FIGURE 12. Two terminal line connected with CIRPP at bus M.

565 in source impedances. For this purpose, we have consid-

566 ered phase-to-ground (a-g) and phase-to-phase-ground (bc-g) TABLE 6. Estimated line parameters using distributed line model.
567 faults at 50 and 150km from bus M respectively. The fault
568 resistance for a-g and bc-g faults are 10 and 5 respectively.
569 Accuracy of the proposed method is verified for different
570 errors in source impedances, ranging from 0% to 50%. The
571 evaluation results for a-g and bc-g fault cases are shown in
572 Fig.11 (a) and (b) respectively. From the figure, the fault loca-
573 tion accuracies of both single-ended [19] and double ended
574 negative sequence [31] are increasing with increasing error in at the today’s situation of renewable power plant integration. 596

575 source impedance parameters whereas double-ended [30] and Let us investigate the performance of proposed and existing 597

576 proposed method accuracies are independent of the source methods for this configuration. Type IV wind turbine gener- 598

577 parameter variation for both cases. The single-ended method ator, which is a full converter model, with 200 units of 2MW 599

578 [19] is more sensitive with the source impedance angle varia- each in compliance with various grid codes [23], [24] and the 600

579 tion and double-ended negative sequence-based method [31] FRT characteristics are modeled for the study. 601

580 sensitive with source impedance magnitude variation. The

581 source strength is significantly varied with integration of the a: PROPOSED METHOD 602

582 renewables into grid and the accuracy of these methods will Consider a-g fault at 48km from bus M on the transmission 603
583 further worsen for future power grids. line with fault resistance of 50, fault inception angle is 00 . 604

The voltage and current wave forms are shown in Fig.13. 605
584 F. STUDY OF THE PROPOSED METHOD FOR LINES From the figure, the current wave form is modulated, and 606
585 CONNECTED WITH RENEWABLE POWER PLANTS voltage wave has more transients at wind farm side. The 607

586 This section provides the analysis of proposed and existing calculated (lumped model) and estimated (distributed line 608

587 methods for lines connected with converter interfaced renew- model) line parameters and corresponding errors compared to 609

588 able power plants (CIRPPs). Two cases are considered such actual line parameters for this case are provided in the Table 5 610

589 as renewable power plant at one side and both sides of the and 6 respectively. From Table 5 and 6, the calculated and 611

590 line. estimated parameters are close to actual parameters as for 612

a line connected with a renewable resource. The calculated 613

591 1) CASE 1: TRANSMISSION LINE CONNECTED WITH fault location using the estimated line parameters and post 614

592 RENEWABLE POWER PLANT AT ONE END OF THE LINE fault phasors is 48.24 km. The absolute percentage of fault 615

593 Consider a two-terminal line with wind type IV renewable location error is calculated using (31) is 0.1% for a 240 km 616

594 power plant connected at Bus M and N connected to power transmission line for this case. The method is not affected by 617

595 grid as shown in Fig.12. This configuration is very common renewable integration. 618

VOLUME 10, 2022 96813

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

FIGURE 15. Line connected with CIRPP at both buses.

FIGURE 14. Shows (a) angle difference between fault current and IED
measured current at bus M and (b) negative sequence current measured
at Bus M for a-g fault at 48km from M.

619 b: SINGLE ENDED METHOD [19]

620 The calculated fault distance using single ended FL method
621 [19] is 40.25 km. The absolute percentage of fault location
622 error is calculated using (31) is 3.23% for a 240 km transmis-
623 sion line. The basic principle of single ended fault location
624 depends on the accuracy of apparent impedance calculation.
625 The apparent impedance for a a-g fault is obtained as in (32)
 
VAM IF
626 Zapp = = ZF + RF (32)
IAM + K0 I0 IAM + K0 I0
627 where VAM , IAM is phase A voltage and current phasors at M,
628 I0 is zero sequence current at M, ZF is the line impedance
629 till fault point, RF is the fault resistance. Fig. 14(a) shows the FIGURE 16. Pre and during fault voltage and current signals measured at
630 angle difference (α) between fault current, IF and IED mea- (a) Bus M and (b) Bus N for a-g fault at 180km from M.
631 sured current, (IAM + K0 I0 ). For comparison purposes, the
632 figures include the respective plots when the CIRPPs at bus
633 M is replaced by a conventional generation of same capacity. systems are shown in Fig 14. (b). From the figure, the negative 660

634 As seen from Fig.13(a) the value of α is higher (−560 ) for sequence current measured for lines connected with renew- 661

635 the CIRPPs-connected system when compared to the con- able resources is almost zero and the method proposed in [31] 662

636 ventional generation-connected system (−90 ). To  analyses  will not work for this case. The most of the renewable power 663
IF plants (wind type IV and Solar PV) will not produce negative
637 the observations made, let us denote the term IAM +K 0 I0
664

638 in (32) as D. When α is very small, as in case of conven- sequence currents [23], [24]. Therefore, the method described 665

639 tional generation-connected lines, D is mostly real. However, in [31] may not work for full converter based [23], [24] (wind 666

640 in CIRPPs-connected lines, since the current is controlled by type IV and Solar PV) renewable resources. 667

641 the inverter control strategies, grid codes and FRT character-
642 istics, α is high and consequently D is not real. This causes 2) CASE 2: TRANSMISSION LINE CONNECTED WITH 668

643 a significant negative reactance shift in apparent impedance RENEWABLE POWER PLANT AT BOTH ENDS OF THE LINE 669

644 estimated, which can potentially cause the IED at bus M Consider a transmission line with wind type IV renewable 670

645 calculates less impedance than actual value. Therefore, the power plant connected at the Bus M and solar PV power plant 671

646 single-ended fault location method [19] is calculates the fault connected at Bus N as shown in Fig.15. This configuration 672

647 location (40.25km) less than the actual value (48km) for this may be not practical today, however, this would be a possible 673

648 case. situation in future. Let us investigate the performance of 674

proposed and available FL methods for this configuration. 675

649 c: DOUBLE ENDED METHOD [30]

650 The calculated fault location using double ended method [30] a: PROPOSED METHOD 676

651 is 48.69 km. The absolute percentage of fault location error is Consider ab fault at 180km from bus M on the transmission 677

652 calculated using (31) and it is 0.3% for this case. The double line with fault resistance of 10, fault inception angle is 600 . 678

653 ended method is not impacted by the renewable integration; The voltage and current signals as shown in the Fig.16 for this 679

654 however, the accuracy of the method depends on the line case. The calculated and estimated parameters are tabulated 680

655 parameter accuracy. in Table 7 and 8 respectively. From the Tables, the calculated 681

and estimated parameters are close to actual parameters as in 682

656 d: DOUBLE ENDED NEGATIVE SEQUENCE CURRENT BASED the case of line connected with a conventional resource. The 683

657 METHOD [31] calculated fault location using the estimated line parameters 684

658 The negative sequence current measured at bus M for lines and post fault phasors is 179.48 km. The absolute percentage 685

659 connected with renewable power plants and conventional of fault location error is 0.2% for this case. The method 686

96814 VOLUME 10, 2022

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

TABLE 7. Calculated line parameters using 5 model.

FIGURE 17. Data message structure.

TABLE 8. Estimated line parameters using distributed line model.

protection by comparing the currents entering and leaving the 726

protected line [47]. Current from local end are connected to 727

the analog input module or a transformer module of an IED. 728

These are then digitized and forwarded as local end current 729

687 provides the desired result. Therefore, the proposed method to a differential protection function and to the remote end 730

688 is suitable for future renewable energy configurations. IED via Line Differential Communication Module (LDCM). 731

Over the same LDCM, digitized current from remote end is 732

689 b: SINGLE ENDED METHOD [19] received and forwarded to the differential. 733

690 The calculated fault distance using single ended method [19] In 64 kb mode, the format for communication is 734

691 is 169.36 km. The absolute percentage of FL error for the case C37.94 and the data are exchanged every 5 ms. The transmit- 735

692 is 4.43%. The single ended fault location is not dependable ted data contains three currents and voltages, sent as sampled 736

693 for lines with CIRPPs connected at both ends. values, clock information, trip, block, and alarm signals. The 737

data message as shown below. 738

694 c: DOUBLE ENDED METHOD [30] In 2 Mb mode, the format for communication is still 739

695 Double ended method proposed in [30] is calculated the fault C37.94 and data sent every 1 ms i.e. telegrams are sent 740

696 location is 179.32 km and absolute for this case is 0.2%. The as soon as data is available. The data consists of nine 741

697 method is not affected by the renewable integration at both analog sampled values and binary signals. The commu- 742

698 sides of the line. nication between the IED, can either be direct of via a 743

multiplexer-telecommunication network interface. In either 744

699 d: DOUBLE ENDED NEGATIVE SEQUENCE CURRENT BASED case, sampling of the analog values shall be synchronized in 745

700 METHOD [31] all IED to ensure proper functioning of the fault locator. 746

701 The negative sequence measurement-based method will not With networks using symmetric or fixed routes, echo- 747

702 work for this situation as there is no significant negative based synchronizing allows the internal clock in each local 748

703 sequence current available like previous case. And also, the IED to act as a master and the internal clock in each remote 749

704 method [31], requires the source impedance as an input. The IED operates as a slave. Time deviations between the internal 750

705 source impedance for lines connected with renewable power clocks are monitored continuously and compensated for with 751

706 plants varies continuously during the fault. The variation echo messages between all ends at 40 ms intervals over the 752

707 depends on the grid codes and inverter control systems [23], telecommunication network. With networks using unspeci- 753

708 [24]. Therefore, this method may not give the accurate result fied route switching, reference for the internal clock comes 754

709 when the lines are connected with CIRPPs. from global time provided by, for example, a built-in GPS 755

receiver. The internal clock in each IED [47] is thus set 756

710 G. SUMMARY OF COMPARATIVE ASSESSMENT according to the GPS systems’ global time. The maximum 757

711 Identifying the best method for fault location also depends time deviation between internal clocks is set using a param- 758

712 on the application scenario. For example, location estimates eter MaxtDiffLevel in IED. If the actual time deviation is 759

713 from two-ended algorithms are not always accurate as FL more than this parameter, the IED will not provide the fault 760

714 accuracy of these methods depends on the line parameters location. There are two main types of telecommunication 761

715 and data synchronization. Therefore, when implementing networks used by electric power utilities: Plesiochronous 762

716 fault location methods, end users should be aware of the Digital Hierarchy (PDH) networks and Synchronous Digital 763

717 application scenario, identify possible error sources, and then Hierarchy (SDH). PDH networks are used with 64 kbps com- 764

718 choose the method that is robust to those error sources. The munication. Proper synchronization of PDH networks must 765

719 proposed method is not influenced by the most source of be available so that they can be used with protection and fault 766

720 errors including lines connected with renewables whereas location applications. SDH networks are used with 2 Mbps 767

721 existing practically available methods influenced by many communication. The communication structure between the 768

722 factors. fault locators (IEDs) as shown in Fig.18. 769

723 H. IMPLEMENTATION IN IED PLATFORM IV. FIELD INSTALLATIONS AND EXPERIENCES 770

724 The line differential protection is based on Kirchoff’s law. The proposed double ended fault location solution is imple- 771

725 A protection IED provides a phase segregated differential mented in IED and tested in laboratory. After success in 772

VOLUME 10, 2022 96815

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

TABLE 9. Summary of factors effecting the fault location accuarcy.

FIGURE 18. Communication structure between fault locator over SDH.

FIGURE 20. Pilot implementation setup at POWERGRID.

FIGURE 19. Pilot installation between station A and station B at power

grid corporation of india limited (POWERGRID).

773 laboratory evaluation, the solution is installed on a 400-kV,

774 233.07 km line in Indian POWERGRID and on a 220-kV,
775 265.6 km line in Swedish Utility. Recently, two fault events
776 were recorded in September 2021 on the transmission line FIGURE 21. Pre and during fault voltage and current signals measured at
(a) Station A and (b) Station B for c-g fault at 33km from station B for
777 where the solution is installed in the Indian system and one event 1.
778 fault event on the line of Swedish installation in July 2021.
779 The detail of the pilot experiences is described in following
780 section. the map and power network as shown Fig.19. The pilot setup 786

including communication, configuration details are shown in 787

781 A. PILOT 1: INDIAN PILOT INSTALLATION EXPERIENCE Fig.20. Two fault events are recorded, and the details are 788

782 A 400-kV, 233.07 km transmission line from Power Grid provided below. 789

783 Corporation of India (POWERGRID) is chosen. The trans- Fault event 1- Phase-c to-ground fault (c-g): The first fault 790

784 mission line is in central part of India, and it connects between occurred on 21st September 2021 and Fig.21 shows voltage 791

785 station A and B. The pilot installation setup is marked on and current signals recorded at station A and B terminals for 792

96816 VOLUME 10, 2022

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

TABLE 10. Estimated line parameters using proposed method. TABLE 12. Estimated line parameters using proposed method.

TABLE 11. Calculated fault location using proposed method. TABLE 13. Calculated fault location using proposed method.

FIGURE 23. Pilot installation line of length 265.6km connected between

station A and B at SVK, Sweden.

TABLE 14. Estimated line parameters using proposed method.

FIGURE 22. Pre and during fault voltage and current signals measured at
(a) Station A and (b) Station B for c-g fault at 33km from station A
Substation for event 2.
or ± 1 tower span from the substation staff claimed value for 815

this case. 816

793 c-phase to ground fault. The line positive sequence parame- Developed fault locator calculated and indicated the fault 817

794 ters (resistance (R), reactance (X ), and susceptance(B)) and location as 32.87 km and 33.23, difference of only 130 and 818

795 fault location calculated by the IED at station A and station B 230-meters differences respectively or ± 1 tower span from 819

796 are provided in Table 10 and 11 respectively. The estimated the claimed value in both cases. The line search required to 820

797 fault distance using the proposed method is 32.87km and locate the fault is only one tower span distance that signif- 821

798 198.20km from station A and station B terminals respectively. icantly saves the line patrolling time and cost. The calcu- 822

799 When the line crew patrolled the line, they found a fault at lated line parameters are consistent for both cases and these 823

800 33km from station A. The difference of actual to estimated parameters can be used to update the protection settings The 824

801 fault location is 130 meters or ± 1 tower span from the end customer (user) acknowledged that printed parameters 825

802 claimed value for this case. The line search required based are very valuable for relay settings and other power system 826

803 on proposed method is two tower span distance for this case. monitoring applications. 827

804 The calculated line parameters can be used to update the

805 protection settings. B. PILOT 2: SWEDISH PILOT INSTATION EXPERIENCE 828

806 Fault event 2- Phase-c to-ground fault: The second fault A 220-kV, 265.6 km transmission line chosen to install pilot 829

807 also occurred at same location as previous event and Fig.22 at Swedish network. The line connects between Substation A 830

808 shows voltage and current signals recorded at station A and B and B as shown in Fig.23. In this pilot installation, the IED is 831

809 terminals for this event. The line parameters and fault location configured with proposed double-ended solution and single- 832

810 calculated by the proposed method at station A and station B ended solution [19]. 833

811 are provided in Table 12 and 13 respectively. The estimated Fault event 1- Phase-a to-phase-b fault (ab): Double phase 834

812 fault distance using the proposed method is 33.23km and fault occurred on 14th July 2021 and Fig.24 shows the voltage 835

813 200.80km from station A and B terminals respectively. The and current recorded at station A and B for this event. The line 836

814 difference of actual to estimated fault location is 230 meters parameters and fault location calculated by the IED (proposed 837

VOLUME 10, 2022 96817

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

TABLE 16. Source impedances.

of the method is compared to practically available methods 867

for impact of error in line and source parameters. For the 868

available methods, the fault location error increases for higher 869

inaccuracies in line parameters whereas, the proposed method 870

is independent of these parameter changes. 871

The proposed method is implemented in existing line pro- 872

tection IED using 1kHz sampling rate. The proposed method 873

FIGURE 24. Pre and during fault voltage and current signals measured at does not require information about line parameters or set- 874
(a) station A and (b) station B for ab fault at 226.6km from substation A.
tings, saves on engineering cost and does not require any 875

additional hardware beyond what is required for a modern 876

TABLE 15. Calculated fault location using proposed method. line differential protection solution, i.e., a 2 Mbps commu- 877

nication link. The method is installed at Indian and Swedish 878

power system, the early field results confirm the validity of 879

the method. The field experimental results are matching with 880

simulation results. It helped in pilot installation to locate the 881

faults within one tower span distance which in-turn aids in 882

quick restoration of the power supply. Quick restoration of 883

power supply reduces customer complaints, outage time, and 884

838 method) at both substations are provided in Table 14 and 15 therefore minimizes revenue loss, outage management costs 885

839 respectively. The estimated fault distances using the proposed for utilities and the end users. These geographically diverse 886

840 method are 226.82km and 38.69km from station A and B pilot installation results stand as testimony to the accuracy of 887

841 terminals respectively. When the line crew patrolled the line, the developed fault location method. 888

842 they found a fault at 226.6 km and 39km from station A and B. In a more general testing, where different fault conditions 889

843 The difference of actual to estimated fault location is 220 and are considered, the fault location error in majority of cases 890

844 310 meters or ± 1 tower span from both the substations for is ∼0.1%. Based on these results and confirmation from the 891

845 this case. Calculated fault location using single-ended fault pilot installations it is expected that the fault location accu- 892

846 locator [19] are 248.6 and 40.37 km from station A and B racy will be within 2 tower spans even in more problematic 893

847 respectively. After analysis and discussion with end customer, situations with higher fault resistances and different fault 894

848 we concluded that the main reason for the higher FL errors types. The innovative and accurate line parameter estimation 895

849 (8.3% and 0.5% from station A and B respectively) of the in the proposed method enables achievement of accuracy 896

850 SEFL [19] for this case is due to incorrect parameter settings. levels comparable to the traveling wave fault locator technol- 897

ogy that uses 1000 times higher sampling frequency, more 898

851 V. CONCLUSION costly hardware and commissioning process. The solution is 899

852 In this work, a double-ended setting-free fault location suitable to locate the fault in different geographical regions 900

853 method for transmission line connected with conventional or (example, snowy areas, coastal, hilly terrains, deserts, etc.) as 901

854 inverter based renewable resources is presented. The voltage it is independent of the transmission line parameters (which 902

855 and current signals from both terminals are the only inputs change with atmospheric conditions). Further, the calculated 903

856 to the fault locator. The parameters of the line, or the source parameters of the line can be used for protection and power 904

857 impedances at terminals are not required to be known which system monitoring application settings. 905

858 avoids the engineering effort and cost. As a future work, the proposed model-free approach can 906

859 The performance of the method is evaluated using the be implemented in SCADA/EMS control centers or cloud- 907

860 EMTDC simulations. The simulation results show that the based platforms. Setting-free fault location for three termi- 908

861 performance of the method is accurate for lines connected nal and non-homogeneous/mixed (combined overhead and 909

862 with conventional and renewable power plants for various underground cable) circuits is difficult as the impedance 910

863 fault scenarios. The results are not affected by the renew- parameters are not same for all sections. This work can be 911

864 able integration whereas negative sequence based double extended for these complex power networks. The issues with 912

865 ended, and single-ended method fail or inaccurate for the data synchronization and measurement errors can be elimi- 913

866 lines connected with renewable power plants. The accuracy nated in future works. 914

96818 VOLUME 10, 2022

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

915 APPENDIX [17] L. Eriksson, M. M. Saha, and G. D. Rockefeller, ‘‘An accurate fault locator 982

916 See Table 16. with compensation for apparent reactance in the fault resistance resulting 983
from remore-end infeed,’’ IEEE Trans. Power App. Syst., vol. PAS-104, 984
no. 2, pp. 423–436, Feb. 1985. 985
917 ACKNOWLEDGMENT [18] L. Ji, X. Tao, Y. Fu, Y. Fu, Y. Mi, and Z. Li, ‘‘A new single ended 986

918 The authors express gratitude to Power Grid Corporation of fault location method for transmission line based on positive sequence 987
superimposed network during auto-reclosing,’’ IEEE Trans. Power Del., 988
919 India Ltd. (POWERGRID), India, for their support and coop- vol. 34, no. 3, pp. 1019–1029, Jun. 2019. 989
920 eration during the installation of IEDs with double–ended [19] Hitachi Energy Line Distance Protection REL670 Manual. [Online]. 990

921 fault locator in their substations. Their valuable and timely Available: https://search.abb.com/library/Download.aspx?DocumentID= 991
1MRK506372-BEN&LanguageCode=en&DocumentPartId=&Action= 992
922 inputs have enabled in validation of the performance of pro- Launch 993
923 posed double ended fault locator solution. And also, they [20] O. E. Schweitzer, ‘‘A review of impedance-based fault locating experi- 994

924 express gratitude to Svenska Kraftnat (SVK) Team for their ence,’’ Proc. 15th Annu. Western Protective Relay Conf., Spokane, WA, 995
USA, Oct. 1988, pp. 24–27. 996
925 continuous support for experimenting the double ended fault [21] S. Das, S. Santoso, A. Gaikwad, and M. Patel, ‘‘Impedance-based fault 997
926 locator solution. location in transmission networks: Theory and application,’’ IEEE Access, 998
vol. 2, pp. 537–557, 2014. 999
[22] A. L. Dalcastagne and S. L. Zimath, ‘‘A study about the sources of error of 1000
927 REFERENCES impedance-based fault location methods,’’ in Proc. IEEE/PES Transmiss. 1001
928 [1] M. M. Saha, J. Izykowski, and E. Rosolowski, Fault Location on Power Distrib. Conf. Expo., Latin Amer., Bogota, Colombia, Aug. 2008, pp. 1–6. 1002
929 Networks (Power Systems). London, U.K.: Springer, 2010. [23] A. Hooshyar, M. A. Azzouz, and E. F. El-Saadany, ‘‘Distance protec- 1003
930 [2] IEEE Guide for Determining Fault Location on AC Transmission and tion of lines emanating from full-scale converter-interfaced renewable 1004
931 Distribution Lines, Standard C37.114–2014 (Revision of IEEE Std energy power plants—Part I: Problem statement,’’ IEEE Trans. Power Del., 1005
932 C37.114–2004), IEEE, Jan. 2015, pp. 1–76. vol. 30, no. 4, pp. 1770–1780, Aug. 2015. 1006
933 [3] H. Panahi, R. Zamani, M. Sanaye-Pasand, and H. Mehrjerdi, ‘‘Advances [24] S. Paladhi and A. K. Pradhan, ‘‘Adaptive distance protection for lines 1007
934 in transmission network fault location in modern power systems: Review, connecting converter-interfaced renewable plants,’’ IEEE J. Emerg. Sel. 1008
935 outlook and future works,’’ IEEE Access, vol. 9, pp. 158599–158615, 2021. Topics Power Electron., vol. 9, no. 6, pp. 7088–7098, Dec. 2021. 1009
936 [4] E. O. Schweitzer, A. Guzman, M. V. Mynam, V. Skendzic, B. Kasztenny, [25] N. George and O. D. Naidu, ‘‘Distance protection issues with renewable 1010
937 and S. Marx, ‘‘Locating faults by the traveling waves they launch,’’ in Proc. power generators and possible solutions,’’ in Proc. 16th Int. Conf. Develop. 1011
938 67th Annu. Conf. Protective Relay Eng., Mar. 2014, pp. 95–110. Power Syst. Protection (DPSP ), Mar. 2022, pp. 1–6. 1012
939 [5] A. Guzman, B. Kasztenny, Y. Tong, and M. V. Mynam, ‘‘Accurate and [26] P. Chang, G. Song, J. Hou, and R. Xu, ‘‘A single-ended fault location 1013
940 economical traveling-wave fault locating without communications,’’ in method for grid-connected converter system based on control and protec- 1014
941 Proc. 71st Annu. Conf. Protective Relay Eng. (CPRE), College Station, tion coordination,’’ IEEE Trans. Power Del., vol. 37, no. 4, pp. 3071–3081, 1015
942 TX, USA, Mar. 2018, pp. 1–18. Aug. 2022. 1016
943 [6] C. Zhang, G. Song, T. Wang, and L. Yang, ‘‘Single-ended traveling wave [27] P. Chang, G. Song, J. Hou, and R. Xu, ‘‘A single-terminal fault location 1017
944 fault location method in DC transmission line based on wave front informa- method for transmission lines integrated by inverter-type source,’’ IEEE 1018
945 tion,’’ IEEE Trans. Power Del., vol. 34, no. 5, pp. 2028–2038, Oct. 2019. Trans. Power Del., vol. 37, no. 3, pp. 1704–1713, Jun. 2022. 1019
946 [7] H. Shu, X. Liu, and X. Tian, ‘‘Single-ended fault location for hybrid feeders [28] A. A. Girgis, D. G. Hart, and W. L. Peterson, ‘‘A new fault location 1020
947 based on characteristic distribution of traveling wave along a line,’’ IEEE technique for two-and three-terminal lines,’’ IEEE Trans. Power Del., 1021
948 Trans. Power Del., vol. 36, no. 1, pp. 339–350, Feb. 2021. vol. 7, no. 1, pp. 98–107, Jan. 1992. 1022
949 [8] O. D. Naidu and A. K. Pradhan, ‘‘Precise traveling wave-based transmis- [29] C.-W. Liu, T.-C. Lin, C.-S. Yu, and J.-Z. Yang, ‘‘A fault location technique 1023
950 sion line fault location method using single-ended data,’’ IEEE Trans. Ind. for two-terminal multisession compound transmission lines using syn- 1024
951 Informat., vol. 17, no. 8, pp. 5197–5207, Aug. 2021. chronized phasor measurements,’’ IEEE Trans. Smart Grid, vol. 3, no. 1, 1025
952 [9] F. V. Lopes, K. M. Silva, F. B. Costa, W. L. A. Neves, and D. Fernandes, pp. 113–121, Mar. 2012. 1026
953 ‘‘Real-time traveling wave-based fault location using two-terminal unsyn- [30] T.-C. Lin, J.-Z. Yang, C.-S. Yu, and C.-W. Liu, ‘‘Development of a 1027
954 chronized data,’’ in IEEE Trans. Power Del., vol. 30, no. 3, pp. 1067–1076, transmission network fault location platform based on cloud computing 1028
955 Jun. 2015. and synchrophasors,’’ IEEE Trans. Power Del., vol. 35, no. 1, pp. 84–94, 1029
956 [10] F. V. Lopes, P. Lima, J. P. G. Ribeiro, T. R. Honorato, K. M. Silva, Feb. 2020. 1030
957 E. J. S. Leite, W. L. A. Neves, and G. Rocha, ‘‘Practical methodology [31] D. A. Tziouvaras, ‘‘New multi-ended fault location design for two-or three- 1031
958 for two-terminal traveling wave-based fault location eliminating the need terminal lines,’’ in Proc. 7th Int. Conf. Develop. Power Syst. Protection 1032
959 for line parameters and time synchronization,’’ IEEE Trans. Power Del., (DPSP), 2001, pp. 395–398. 1033
960 vol. 34, no. 6, pp. 2123–2134, Dec. 2019. [32] H. Torkaman, E. Zeraatkar, N. Deyhimi, H. H. Alhelou, and P. Siano, 1034
961 [11] Qualitrol Traveling Wave Locator and Monitor. Accessed: Sep. 2016. ‘‘Rearrangement method of reducing fault location error in tied uncom- 1035
962 [Online]. Available: https://www.qualitrolcorp.com/wp-content/uploads/ pleted parallel lines,’’ IEEE Access, vol. 10, pp. 51862–51872, 2022. 1036
963 2016/09/IP-F21-02L-01E_TWS_FL8_FL1.pdf [33] C. Galvez and A. Abur, ‘‘Fault location in power networks using a sparse 1037
964 [12] Reason RPV311 Digital Fault Recorder With Fault Location and PMU. set of digital fault recorders,’’ IEEE Trans. Smart Grid, vol. 13, no. 5, 1038
965 Accessed: May 2018. [Online]. Available: https://www.gegridsolutions. pp. 3468–3480, Sep. 2022, doi: 10.1109/TSG.2022.3168904. 1039
966 com/products/brochures/reason%20rpv311-brochure-en-2018-05-33113- [34] B. R. Jeffrey, B. Gabriel, and T. Dmetrios, ‘‘Multi-ended fault location 1040
967 a4.pdf system,’’ U.S. Patent 6 256 592 B1, Jul. 3, 2001. 1041
968 [13] O. Naidu and A. K. Pradhan, ‘‘A traveling wave-based fault location [35] J. P. William, Z. K. Bogdan, and G. A. Mark, ‘‘Multi-ended fault location 1042
969 method using unsynchronized current measurements,’’ IEEE Trans. Power system,’’ U.S. Patent 7 472 026 B2, Dec. 30, 2008. 1043
970 Del., vol. 34, no. 2, pp. 505–513, Oct. 2018. [36] G. L. Kusic and D. L. Garrison, ‘‘Measurement of transmission line 1044
971 [14] O. D. Naidu and A. K. Pradhan, ‘‘Model free traveling wave based fault parameters from SCADA data,’’ in Proc. IEEE PES Power Syst. Conf. 1045
972 location method for series compensated transmission line,’’ IEEE Access, Expo., New York, NY, USA, Oct. 2004, pp. 440–445. 1046
973 vol. 8, pp. 193128–193137, 2020. [37] Y. Wang, W. Xu, and J. Shen, ‘‘Online tracking of transmission-line 1047
974 [15] E. O. Schweitzer III, M. V. Mynam, A. Guzman-Casillas, V. Skendzic, parameters using SCADA data,’’ IEEE Trans. Power Del., vol. 31, no. 2, 1048
975 B. Z. Kasztenny, and D. E. Whitehead, ‘‘Power line parameter adjust- pp. 674–682, Apr. 2016. 1049
976 ment and fault location using traveling waves,’’ U.S. Patent 8 990 036 B1, [38] L. Zhang and A. Abur, ‘‘Identifying parameter errors via multiple mea- 1050
977 Mar. 24, 2015. surement scans,’’ IEEE Trans. Power Syst., vol. 28, no. 4, pp. 3916–3923, 1051
978 [16] T. Takagi, Y. Yamakoshi, M. Yamaura, R. Kondow, and T. Matsushima, Nov. 2013. 1052
979 ‘‘Development of a new type fault locator using the one-terminal voltage [39] M. Asprou and E. Kyriakides, ‘‘Identification and estimation of erroneous 1053
980 and current data,’’ IEEE Power Eng. Rev., vol. PER-2, no. 8, pp. 59–60, transmission line parameters using PMU measurements,’’ IEEE Trans. 1054
981 Aug. 1982. Power Del., vol. 32, no. 6, pp. 2510–2519, Dec. 2017. 1055

VOLUME 10, 2022 96819

 O. D. Naidu et al.: Economical Setting-Free Double-Ended Fault Locator for Transmission Lines

1056 [40] Y. Liao and M. Kezunovic, ‘‘Online optimal transmission line parameter A. V. S. S. R. SAI received the master’s degree 1109
1057 estimation for relaying applications,’’ IEEE Trans. Power Del., vol. 24, in power systems from the National Institute of 1110
1058 no. 1, pp. 96–102, Jan. 2009. Technology (NIT), Tiruchirappalli, India, in 2007. 1111
1059 [41] O. D. Naidu, N. George, and P. Yalla, ‘‘Parameter estimation, selective He is currently working as a Senior Research 1112
1060 auto-reclosing and fault location for three-terminal mixed transmission and Development Engineer with Hitachi Energy, 1113
1061 lines using synchronised data,’’ IET Gener., Transmiss. Distrib., vol. 14, Grid Automation Research and Development. 1114
1062 no. 25, pp. 6049–6060, Dec. 2020. He joined the ABB Research and Development 1115
1063 [42] D. Ritzmann, P. S. Wright, W. Holderbaum, and B. Potter, ‘‘A method for
Center, Bengaluru, in 2007, as a Domain Engi- 1116
1064 accurate transmission line impedance parameter estimation,’’ IEEE Trans.
neer. He is involved in the development of relay 1117
1065 Instrum. Meas., vol. 65, no. 10, pp. 2204–2213, Oct. 2016.
1066 [43] A. Bendjabeur, A. Kouadri, and S. Mekhilef, ‘‘Novel technique for trans- application function library (AFL) related to line 1118

1067 mission line parameters estimation using synchronised sampled data,’’ IET protection and monitoring. His research interests include power system 1119

1068 Gener., Transmiss. Distrib., vol. 14, no. 3, pp. 506–515, Feb. 2020. protection and monitoring applications and specialized in fault location. 1120

1069 [44] A. K. Gautam, S. Majumdar, and H. Parthasarathy, ‘‘State and parameter

1070 estimation of non-uniform transmission line using Kronecker product
1071 based modeling,’’ IEEE Trans. Power Del., early access, Feb. 7, 2022, doi:
1072 10.1109/TPWRD.2022.3149137.
1073 [45] J. Sun, Q. Chen, and M. Xia, ‘‘Data-driven detection and identification of
1074 line parameters with PMU and unsynchronized SCADA measurements in
1075 distribution grids,’’ CSEE J. Power Energy Syst., early access, May 6, 2022,
1076 doi: 10.17775/CSEEJPES.
1077 [46] O. Naidu, P. Yalla, A. Sai, and S. Sawai, ‘‘Model-free fault location for
1078 transmission lines using phasor measurement unit data,’’ CIGRE Study A. N. PRAVEEN received the B.E. degree in elec- 1121
1079 Committee B5 Colloq., vol. 5, pp. 1–12, Jun. 2019. trical and electronics from the National Institute 1122
1080 [47] Hitechi Energy Product Guide, Line Differential Protection RED6702.1. of Engineering, Mysuru, India, in 2009. He is 1123
1081 [Online]. Available: https://search.abb.com/library/Download.aspx? currently pursuing the Ph.D. degree in protection 1124
1082 DocumentID=1MRK505346-BEN&LanguageCode=en&Document applications for smart grid with the Indian Insti- 1125
1083 PartId=&Action=Launch tute of Technology (IIT), Madras, India. He joined 1126
ABB, in 2009, and worked as a Commissioning 1127
Engineer for UHV and EHV substations and then 1128
as a Senior Application Engineer for grid automa- 1129
tion systems. He is currently working as the Tech- 1130
nical Support Line Manager with Hitachi Energy, Grid Automation Products. 1131

1084 O. D. NAIDU (Senior Member, IEEE) received the

1085 Ph.D. degree from the Indian Institute of Technol-
1086 ogy (IIT), Kharagpur, India. From 2009 to 2012,
1087 he was a Senior Power System Application Devel-
1088 opment Engineer at the ABB India Development
PATRICK COST joined ABB, the Netherlands, 1132
1089 Center, Bengaluru, India. From 2012 to 2019,
in 2006, where worked on various projects as a 1133
1090 he was the Principal Scientist at the ABB Cor-
Product and Project Engineer. From 2010 to 2015, 1134
1091 porate Research Center, Bengaluru. He is cur-
he was an Application Specialist and the Regional 1135
1092 rently working as a Senior Principal Engineer with
Technical Marketing Manager for the Central 1136
1093 Hitachi Energy, Grid Automation Research and
Europe and Sub-Sahara Africa, Vasteras, Sweden. 1137
1094 Development, Bengaluru. He is the author of more than 50 scientific papers
In 2015, he joined the Global Product Manage- 1138
1095 and 40 patent applications. He holds 15 granted patents. His research inter-
ment Team for bay level products, with a focus 1139
1096 ests include power system protection, fault location, renewable integration
on protection application development. He is cur- 1140
1097 and monitoring, artificial intelligence applications to power system protec-
rently the Global Product Manager with Hitachi 1141
1098 tion, and monitoring.
Energy, Vasteras. 1142

1099 SINISA ZUBIC received the Ph.D. degree from HÅKAN ERIKSSON received the B.S. degree in 1143
1100 the University of Belgrade, Serbia, in 2013. computer science from the University of Uppsala, 1144
1101 From 2014 to 2018, he joined ABB, where he was Sweden, in 1990. He was worked at ASEA and 1145
1102 a Senior Scientist at the ABB Corporate Research ABB in various positions related to relay protec- 1146
1103 Center, Poland. Before joining ABB, he was a tion systems including test and RTDS simulation, 1147
1104 Teaching and Research Assistant at the Faculty of research and development, technical support, and 1148
1105 Electrical Engineering, Banja Luka, Bosnia and lead engineer for over 32 years (1977–2010). Since 1149
1106 Herzegovina. He is currently the Research and 2010, he has been working with the Swedish TSO 1150
1107 Development Manager in application software at Svenska kraftnät as a Relay Protection Engineer. 1151
1108 Hitachi Energy, Vasteras, Sweden. 1152

96820 VOLUME 10, 2022