IET Generation, Transmission & Distribution

PMU Performance in the Estimation of Dynamic Line Rating

- A Case Study

Journal: IET Generation, Transmission & Distribution

Manuscript ID GTD-2017-0661

Manuscript Type: Case Study

Date Submitted by the Author: 27-Apr-2017

Complete List of Authors: Alvarez, David L.; Universidad Nacional de Colombia, Facultad de
Ingeniería - Departamento de Ingeniería Eléctrica y Electrónica
Faria da Silva, Filipe
Bak, Claus
Mombello, Enrique; CONICET San Juan, Instituto de Energía Eléctrica
ROSERO , JAVIER ; Universidad Nacional de Colombia
Ólason, Daníel

PHASOR MEASUREMENT UNITS (PMU), POWER OVERHEAD LINES,

Keyword:
ELECTRIC RESISTANCE MEASUREMENT, HEAT TRANSFER, ERRORS

IET Review Copy Only

Page 1 of 9 IET Generation, Transmission & Distribution

IET Generation, Transmission & Distribution

Case Studies

PMU Performance in the Estimation of ISSN 1751-8687

doi: 0000000000
www.ietdl.org
Dynamic Line Rating - A Case Study
David L. Alvarez1 , F. Faria da Silva2 , Claus Leth Bak2, Enrique E. Mombello3 , Javier A. Rosero1 , Daníel
Leó Ólason4
1
Electrical Machines & Drives Group, EM&D, Universidad Nacional de Colombia, Bogotá, Colombia
2
Department of Energy Technology, Aalborg University, Aalborg, Denmark
3
Instituto de Energía Eléctrica, CONICET, Universidad Nacional de San Juan, San Juan, Argentina
4
LANDSNET, Iceland
E-mail: dlalvareza@unal.edu.co

Abstract: This paper analyzes the influence of weather variations in time and space along overhead lines and the error in syn-
chrophasor measurements for estimating conductor temperature in real line. The analysis begins by computing the thermal and
mechanical parameters which change according to the load and atmospheric conditions. Having computed the thermal and me-
chanical parameters, the values of resistance, inductance and capacitance of an overhead line modeled by means of a π equivalent
circuit are estimated, with the purpose of quantifying the sensibility of the electrical parameters to changes in the temperature of
the conductor. Subsequently, the average conductor temperature is estimated from synchrophasor through the relationship be-
tween resistivity and temperature; this temperature is compared with temperature computed from atmospheric conditions in each
ruling that, are interpolated from nearby weather stations. Additionally, the propagation of the error is calculated and compared with
acceptable error margins, because the estimated temperature from synchrophasor is computed from indirect measures. Finally,
the influence of the load variations in the accuracy of the temperature estimation with synchrophasor is analyzed.

1 Introduction
Power systems are facing new challenges in operation, control and
planning. To better face these challenges, it is necessary to optimize
the asset capacity, which have reached their limits as a consequence
of new loads and sources [1]. These new loads and sources increase
the congestion and risk, especially in overhead lines (OHL) [2]. To
push limits in OHL, new technologies and methods have been devel-
oped with the aim of improving their capacity, reliability, safety and
economic operation [3].
Among the technologies for pushing OHL limits is Dynamic Line
Rating (DLR). This technology has the ability to compute the ther-
mal ampacity of the conductors in real time, based on the current
weather [4]. Traditionally, the line ratings are fixed according to ex-
treme climate conditions that rarely happen; however, thanks to the
development of information technologies, it is possible to compute
the real rating online, via measurements of atmospheric conditions
and the current intensity. Two types of measurements for DLR
have been defined: direct and indirect [5]. The indirect method uses
weather stations near the OHL whereas direct direct methods uses
sensors of mechanical tension, temperature, sag or results derived
from these three variables. The devices used for direct methods are
located directly in the OHL, making it difficult to put them into op-
eration and requiring maintenance. Despite this, DLR has low costs
and it is fast to implement, if compared with others methods used
to increase OHL rating [6]. Additionally, DLR is useful when it is
necessary to increase the capacity between 10% and 30%, particu-
larly for wind power integration [7], given the relationship between
wind speed, power generation and cooling. In brief, DLR increases
the capacity of OHL most of the time, achieving asset optimization.
The use of PMUs allows the real-time estimation of the thermal
capacity in conductors. This method is considered as DLR tech-
nology, with the advantage that it uses an existing infrastructure
capable of guaranteeing the functioning and reliability of the DLR
system [8]. With PMU, the conductors’ ratings are estimated from
the impedance of OHL equivalent circuit [9, 10]. The impedance
changes according to the conductors’ temperatures, a parameter that
impacts power state estimation [11] and load flows [12], thereby af-
fecting the losses, power flows, bus voltages, protections schemes
[13] and the line capacity, among others.
The use of PMU for DLR is based on the average conductor
temperature along the OHL. However, this temperature varies along
the line, as a consequence of atmospheric variations in the differ-
ent spans. Reference [14] presents a methodology for incorporating
temperature variations along the line, which consists in the division
of the line in segments based on the temperature gradients obtained
from measurements along the conductor. In the same way, in [15]
critical spans for monitoring OHL are estimated by means of weather
forecasting models, taking into account the climate variations in time
and space. In summary, the variation of the temperature along the
OHL is not negligible when using DLR.
This paper analyzes the influence of the atmospheric variations
in time and space, when PMU is used for a real OHL, showing that
for typical weather conditions, the error in the estimated tempera-
ture using PMU exceeds the acceptable error margins. The paper is
organized as follows: Section 2 discusses multiphysics behavior of
OHLs when changes in weather or load occur. Section 3 describes
the OHL under study and the interpolation method for knowing the
weather parameters along the line. In section 4, the impact of the
weather over conductor temperature is computed in each ruling span
and compared with the temperature calculated from PMU measure-
ments. Finally, Section 5 analyzes the error in the estimation of the
OHL resistance and temperature when PMU measurements are used,
taking into account the accuracy and load variations.
2 Multiphysics Phenomena
During operation, an OHL is under the influence of thermal, mechan-
ical and electrical phenomena [16]. Figure 1 shows the relationship
between these physical phenomena. At first, a heat transfer (Q) is
presented as a product of a heat gain (mainly by Joule effect (P )
and solar radiation (S)) and a heat loss (radiation and convection).
That heat transfer is determined by the current intensity ikm , the

IET Gener. Transm. Distrib., pp. 1–9

c The Institution of Engineering and Technology 2017 IET Review Copy Only 1
 IET Generation, Transmission & Distribution Page 2 of 9

R(TS , ℓ) L(ℓ)
ik im
+ +
Electro-Magnetic
i, v
vk 1 1 vm
C(D, ℓ) C(D, ℓ)
2 2

Ts , P, σ ~
E − −

R, L, C

Thermal Fig. 2: Mechanical and thermal variables that influenced RLC

Mechanical
D, ℓ parameters of an OHL modeled by π equivalent circuit
~a
Q, Ta , S, ϑ H

length, rs is the equivalent ruling span length, and εt is the coeffi-

cient of thermal expansion. As the value of HT ref varies over time,
a continue estimation of these values is necessary.
Fig. 1: Multiphysical phenomena in OHLs as consequence of heat
transfer
EA (rs mc g)2
=
24
properties of the conductor and by atmospheric conditions (ambi-
EA (rs mc g)2
" #
ent temperature (Ta ), solar radiation, and wind speed and direction 2

 
~ a . The heat transfer affects the conductor temperature (TS ), H H − HTref + + EAεt TS − Tref (2)
ϑ 24HTref 2
leading to a variation in the horizontal component of mechanical
tension (H) in the conductor, as a result of changes in the conduc- The conductor length per phase (ℓ) can be computed from the
tor length (ℓ) and in the sag of the catenary (D). Additionally,  the
 OHL geometry and the value of the tension (H) by means of (3),
changes of TS , D, ℓ impact the distribution of the electric field E~ where h is the vertical distance between support elevation points
and the electrical conductivity of the conductor (σ). These varia- (inclined spans) and s is the span length.
tions reflect in the values of voltage (v) and current intensity along s
2
the line. Finally, these three physical phenomena affect the RLC 2H mc gs
 
parameters of the OHL, given that these parameter depends on the ℓ= h2 + sinh (3)
mc g 2H
geometry of the line and the properties of the conductor.
Finally, the OHL sag (D) is computed through (4).
2.1 Thermal phenomena H
 
mc gs
 
D= cosh −1 (4)
CIGRE [17] and IEEE [18] standards are commonly used for com- mc g 2H
puting the temperature in the conductors of OHL. These standards
are both based on the heat balance equation. For thermal steady state, 2.3 Electrical phenomena
(1) is used, where QJ is the heat gain from the Joule effect, QS is
the gain from solar radiation, QC is the loss for convective cooling The electrical parameters of the π equivalent circuit (Figure 2) used
and QR is the loss for radiative cooling. The gain from magnetic for modeling OHLs with medium length are influenced by variations
heating and corona heating, as well as the losses due to evaporative in the load and atmospheric conditions. Rating in medium and short
cooling, are commonly ignored. OHLs is commonly determined by the sag of the catenary [20], a
limit given by a maximum temperature in the conductor. Thus, DRL
is typically used for this kind of OHL. The equivalent resistance of
QJ + QS = QC + QR (1) the conductor (R) varies according to the temperature (TS ) and the
From equation (1), the temperature in the conductor TS and the conductor length (ℓ). These variations are described by (5), where
maximum current intensity can be computed, if the atmospheric con- α is the resistance temperature coefficient. This equation is valid as
ditions, current intensity and the conductor properties (resistivity, long as the conductivity of the material is in the linear zone regarding
temperature coefficient of resistance, solar absorptivity of surface, temperature dependence, which occurs in the normal operation of
solar emissivity of surface, diameter, among others) are known. OHL.

ℓTS
2.2 Mechanical phenomena RTS = RTref 1 + α TS − Tref · (5)
ℓTref
Temperature variations in the conductor result in changes in its The equivalent inductance (L) depends on the arrangement of
length and variations on the forces that act on the catenary. To model the conductors, distances among them, and length of phase conduc-
this behavior, numerical or analytic formulations are used. Numer- tor. This parameter can be computed from (6), where GM D is the
ical methods such as Finite Elements are not commonly used for geometric mean distance and GM R is the geometric mean radius.
DLR, because they require specialized software and large computa-
tional resources when compared with analitycal approximations. As 
GM D

an analytical method, the state change equation (2) is normally used L = 2 · 10−4 ln ·ℓ (6)
GM R
for modeling the tension in a line section (ruling span) [19]. This
equation related the tension HS at a temperature TS by means of a The length of phase conductor and the average distance (havg )
known HT ref at a known temperature Tref , where E is the mod- between conductor and ground influences the value of equivalent ca-
ulus of elasticity of the conductor, A is the conductor cross section, pacitance (C). To calculate C from geometry, (7) can be used [20],
g is the gravitational constant, mc is the conductor mass per unit where k1 depends on havg . Reference [21] uses (8) for computing

IET Gener. Transm. Distrib., pp. 1–9

2 IET Review Copy Only c The Institution of Engineering and Technology 2017

Page 3 of 9 IET Generation, Transmission & Distribution

havg , where Davg is the average sag and hM is the conductor height Table 1 BR1 OHL - Ruling spans characteristics
at the tower. This expression takes into account the sag variation,
Ruling Conductor Capacity Spans Length [m]
which is function of temperature. Span Type [MVA]

0.05556 · 10−6 1 470-AL3 304 289 387 440

C= ·ℓ (7) 2 470-AL3 304 229 394 301 312 392 414 324 340 357
GM D

ln k1 3 470-AL3 304 433 396 453 342 277 180 429 266 190 330
GM Rc 4 470-AL3 304 422 339 398 408 300 394 309 417 376 433
438 424 413 209 393
p 5 470-AL3 304 313 441 385 405 386 444 408 404 453 399
(2hM − Davg ) Davg 420 425 407 422 386 439 414 419 396 359
havg = p ! (8) 357 340 351 376
hM + (2hM − Davg ) Davg 6 470-AL3 304 378 445 316 296 435 313 436 435 414 307
log
hM − Davg 7 470-AL3 304 379 405 442 422 429 295 400 441 376 442
216
8 470-AL3 304 296 231 241 453 273 398 406 364 396 361
263
9 470-AL3 304 410
3 Case Study 10 470-AL3 304 196 215 199
11 470-AL3 304 208 139 137 183 160 137 141 145 130
12 470-AL3 304 407
To analyze the impact of weather along the line in the RLC pa- 13 470-AL3 304 388
rameters, it is studied the OHL identifies as BR1 that belongs to the 14 470-AL3 304 481
Icelandic transmission system operated by Landsnet. This line con- 15 470-AL3 304 261 297 199
16 6469- 304 201 890 167
nects geothermal plants and the substation Brennimelur, and it is AL3134ST4A
considered as the most critical connection in the country [22]. As 17 470-AL3 352 322 280 366 335
shown in Figure 3, BR1 connection crosses mountains, valleys and 18 470-AL3 304 309 313 237 213
the sea, and it was built with three different types of conductors; 19 470-AL3 304 249 377 342 373 243
20 470-AL3 304 415 176
therefore, temperature variations along the conductor occur. Given 21 470-AL3 304 268 281
these characteristics, DLR is an option to increase the reliability and 22 470-AL3 304 296
capacity of the BR1 line. 23 470-AL3 304 379 292 360 385 353 376 300 282 346
24 470-AL3 304 373 329 365 347
25 774-AL3 304 468 329 289 580
3.1 Test Line 26 2X774-AL3 415 222 349 337 387
27 2X774-AL3 830 441 249 288 349
BR1 has a rate voltage of 220 [kV], a length of 59.4 [km] and it 28 2X774-AL3 830 193 398 307 238 351 316
29 2X774-AL3 830 173 260 276 213 297
is suspended at 172 towers divided in 30 tension sections shown in 30 2X774-AL3 830 368 384 398 385 337 340
Table 1. In this work each tension section is approached to a ruling
span [19]. Different types of conductors are used on the OHL; their
properties are shown in Table 2. Weather conditions for static rating
are: ambient temperature Ta = 10 [◦ C], wind speed and angle of
~ a = 0.6∠90◦ [m/s] and solar radiation S = 0 [W/m2 ] for
attack ϑ Table 2 BR1 OHL Conductors
a conductor temperature of TS = 40 [◦ C].
470-AL3 6469-AL3 774-AL3 unit
/134ST4A
3.2 Weather Nowcasting
Type AAAC AACSR AAAC
There are sixteen weather stations close to the OHL; their names and A 469.6 × 10−6 469 × 10−6 774.2 × 10−6 m2
mc 1.294 2.4217 2.140 kg/m
locations are shown in Table 3. The measure records from these sta- E 57000 × 106 67100 × 106 55000 × 106 N/m2
tions are available online at the Islandic Met Office webpage. For εt 23 × 10−6 19.3 × 10−6 23 × 10−6 1/K
DLR it is recommended taking samples every 10 or 15 [min] [23]. R′T 0.07415 × 10−3 0.0768 × 10−3 0.0389 × 10−3 Ω/m
ref
However, as the aim of this paper is to evaluate the performance α 0.0036 0.0038 0.0036 1/K
to use PMU for DLR, the atmospheric conditions between 2016- Tref 25 20 20 ◦
C
04-18 00:00 and 2016-04-18 21:00, with samples taked every three αs 0.5 0.5 0.5 1
ε 0.5 0.5 0.5 1
hours, was considered. The atmospheric conditions were interpo-
d 28.14 × 10−3 32.28 × 10−3 36.18 × 10−3 m
lated through biharmonic splines, evaluating the points located in
the middle of each ruling span using the records and location of the
weather station and the function griddata of Matlab R
. This work
assumes that the atmospheric conditions does not change along each
ruling span, and the conductor temperature (TS ) is computed us- Table 3 Weather station close to the BR1 influence area
ing these measurements. An accurate model of downscale climate
is beyond the scope of this paper, as this work only seeks to ana- Weather station WMO number Latitude [◦ ] Longitude [◦ ]
lyze the influence of weather variations on the conductor capacity.
Rvk 04030 64.1275 -21.9028
Temperature and wind interpolations for the date 2016-04-18 21:00 Holms 04920 64.1085 -21.6864
are shown in Figure 4. Given the climate characteristics of Iceland, Korpa 04132 64.15049 -21.75109
the solar radiation is neglected [22] and normally, the Icelandic Met Geldn 04880 64.1678 -21.8038
Office does not report this parameter. Kjaln 04848 64.2106 -21.7667
Skrau 04818 64.2318 -21.8046
Blikd 04912 64.2664 -21.8329
Sfell 04136 64.2405 -21.4633
4 Impact of atmospheric variations Moshe 04918 64.214 -21.3448
Tingv 04142 64.2807 -21.0875
Akrfj 04926 64.3105 -21.966
To analyze the impact of atmospheric variations on the capacity of Tyril 04806 64.3877 -21.4169
the OHL, thermal, mechanical and electrical variables are calculated Botns 04814 64.4529 -21.4034
for each ruling span based on weather interpolation and the OHL Skahe 04904 64.4902 -21.7621
geometry. Figure 5 shows the flowchart for computing the values Hamel 04128 64.4647 -21.9628
Hveyr 04134 64.567 -21.767
of these variables (TS , ℓ, S, H, R, L, C). Afterwards, these results
are compared with values computed from phasor measurements in

IET Gener. Transm. Distrib., pp. 1–9

c The Institution of Engineering and Technology 2017 IET Review Copy Only 3
 IET Generation, Transmission & Distribution Page 4 of 9

1
64.5° N

-1

-2

-3

-4

-5

-6

21.5° W
Fig. 3: Geographic location of BR1-OHL ruling spans (blue
squares) and nearby weather stations (red diamonds) a

7
order to evaluate the PMU performance. From the PMU measures
of vk , ik , vm , im and with (9) and (10), the value of average tem-
6.5
perature TS is computed through the estimation of the parameters
for circuit of Figure 2. In this work, the PMU values were ob-
tained through SIMULINK R
simulations as follows: a power flow 64.5° N
6

for the circuit of Figure 2 is run initially assuming the design val-
ues of resistance, inductance and capacitance under rate conditions, 5.5

vk = 220 [kV], S = 304 [MVA] and P F = 0.9. Afterwards, an it-

erative script was implemented changing the RLC values of the π 5

model according to (5), (6) and (7), with the aim of updating the
electrical parameters considering the changes in the temperature TS . 4.5

This script runs until the current intensity computed through the load
flow is equal to the current ikm used for calculating the resistance 4

from (5). Figure 6a shows the values of current intensity ikm and
voltage vm for each sample. 3.5

vk 2 − vm 2 3
Z= (9)
vm ik − vk im
2.5

ik + im
 
Y = Im 2 · (10) 2
vk + vm
Figure 6 shows variations of the ℓ, XL , YC , R, D and TS pa- 21.5° W

rameters for each weather sample. The maximum variation of the b

entire phase conductor length is less than 0.02%, which corresponds
to 9 [m]; this means that the inductance of the OHL is not affected Fig. 4: Weather nowcasting for the area of influence of BR1 OHL,
due to normal atmospheric variations and that the phase conductor at 2016-04-18 21:00
length can be assumed constant, as shown in Figure 6b. The vari- a Temperature
ation between the maximum and minimum value of the equivalent b Wind speed
capacitance is less than 0.2%, making negligible the influence of
the sag (D), as shown in Figure 6c, where the sag of the ruling
span number 8 is plotted. This span was chosen because it has the
highest variation within samples, approx. 1.3 [m]. On other hand, ~ S, ikm
Ta , ϑ, Thermal phenomenon TS
the value of resistance changes up to 3.5%, as shown
in Figure 6d.
The average temperature of the conductor TSavg along the OHL
is computed using these resistance values, with the maximum and
Mechanical phenomenon ℓ, D, H
minimum values being 6.7 [◦ C] and 2.6 [◦ C], respectively. In all
samples, the differences between TSavg and TSmax exceed the ac-
ceptable error margin for critical spans of 4 [K] (10% of 40 [◦ C]) Electro-magnetic
R, L, C
proposed in [5]. The maximum temperature (TSmax ) and minimum phenomenon
temperature (TSmin ) were obtained by computing the temperature
from weather in all ruling spans and taking the highest and lowest of Fig. 5: Flowchart for computing each OHL parameter that changes
these values. with the weather and current intensity

IET Gener. Transm. Distrib., pp. 1–9

4 IET Review Copy Only c The Institution of Engineering and Technology 2017

Page 5 of 9 IET Generation, Transmission & Distribution

203.8 717.05 59.795 25.239

717 59.794 25.2385

203.78
716.95
59.793 25.238
203.76 716.9
59.792 25.2375
716.85
203.74 59.791 25.237
716.8
203.72 59.79 25.2365
716.75
59.789 25.236
203.7 716.7
59.788 25.2355
716.65
203.68
716.6 59.787 25.235

203.66 716.55 59.786 25.2345

18 Apr 00:00 18 Apr 06:00 18 Apr 12:00 18 Apr 18:00 18 Apr 00:00 18 Apr 06:00 18 Apr 12:00 18 Apr 18:00
Date Date

a b

9.4 164.99 25 3.74

TS
164.98 avg 3.72
9.2 20 TS
max
164.97
TS 3.7
9 164.96 15 min
OHL R 3.68
164.95
8.8 10 3.66
164.94
8.6 5 3.64
164.93
3.62
8.4 164.92 0
3.6
164.91
8.2 -5
164.9 3.58

8 164.89 -10 3.56

18 Apr 00:00 18 Apr 06:00 18 Apr 12:00 18 Apr 18:00 18 Apr 00:00 18 Apr 06:00 18 Apr 12:00 18 Apr 18:00
Date Date

c d

Fig. 6: Variation of the parameters of the BR1-OHL for each weather sample
a Current intensity (ikm ) flowing through the OHL and voltage at end m (vm )
b Inductance (reactance) and phase conductor length
c Capacitance (admittance) and sag of ruling span number 8
d Resistance, average temperature and, maximum and minimum temperature along the OHL

The value of TS in each ruling span of the OHL under study Start
varies, even if the weather does not change along it; this is a
State estimation
consequence of using different conductors. In this work, this is con- vk , ik , vm , im
Requiv (vk , ik , vm , im )
sidered
by using the following procedure: an equivalent temperature


TSavg is computed with (11) from the resistance Requiv cal- TSavg (11)

culated with (5). Using temperature and supposing initial values

of average ambient temperature (Tak=0 ) and solar radiation,
 the T a0 , S

cooling heat is computed. An equivalent wind speed ϑ ~ equiv is

~ equiv T ak , TS , S, i, Requiv [24]


ϑ
calculated [24] along the entire OHL based on the cooling heat pre- avg

vious computed. With these weather parameters, the temperature T ak+1 =

 
~ equiv [17]
TSi T ak , S, ϑ
in each ruling span (TSi ) is calculated and the resistivity of each T ak − ∆TS

conductor is considered. However, as Ta is originally guessed, it is N

1P
necessary to adjust this value via iterations until the

difference be- ∆TS = TSavg − T · ℓi
ℓ i=1 Si
tween the weighted average of (TSi ) and TSavg is less than an
error (ε). Figure 7 shows this procedure.
|∆TS | < ε
No

N
P
PN
Yes
Requiv − Ri Tref + Ri Tref · αi · Tref Finish
i=1 i=1
TSavg =
N
P
Fig. 7: Flow chart for computing TS in each ruling span from PMU
Ri Tref · αi
i=1
measurements
(11)
Figure 8 shows both the temperature and the error calculated with
the atmospheric conditions in each ruling span and the temperature 8a shows a dynamic behavior in time and space (along the OHL) of
computed from PMU by simulations of the OHL under study. Figure the temperature in the conductor. Moreover, the critical ruling span

IET Gener. Transm. Distrib., pp. 1–9

c The Institution of Engineering and Technology 2017 IET Review Copy Only 5
 IET Generation, Transmission & Distribution Page 6 of 9

25
PMU18.04 00:00 Weather Interpolation18.04 00:00
PMU18.04 03:00 Weather Interpolation18.04 03:00
PMU18.04 06:00 Weather Interpolation18.04 06:00
20 PMU18.04 09:00 Weather Interpolation18.04 09:00
PMU18.04 12:00 Weather Interpolation18.04 12:00
PMU18.04 15:00 Weather Interpolation18.04 15:00
PMU18.04 18:00 Weather Interpolation18.04 18:00
15 PMU18.04 21:00 Weather Interpolation18.04 21:00

10

-5

-10
5 10 15 20 25 30
R
 
a

12
18.04 00:00
18.04 03:00
18.04 06:00
18.04 09:00
10 18.04 12:00
18.04 15:00
18.04 18:00
18.04 21:00

0
5 10 15 20 25 30
Ruling Span
b

Fig. 8: Comparison between temperatures computed from weather interpolation and from PMU estimation in each ruling span for different
times
a Temperature of the conductor - TS
b Error between TS computed from weather interpolation and PMU estimation

changes for each weather sample, and the acceptable error margin in Figure 8b. The critical span was assumed as the span with the
(4[K]) is exceeded between the different critical spans, as shown highest temperature.

IET Gener. Transm. Distrib., pp. 1–9

6 IET Review Copy Only c The Institution of Engineering and Technology 2017

Page 7 of 9 IET Generation, Transmission & Distribution

5 PMU Measurement Error Impact

4.4

5.1 Impact on the accuracy

4.2

This section estimates temperature in conductors using PMU, con-

4
sidering errors in measurements. A normal distribution of error with
mean zero is assumed and the standard deviation is approximated to
3.8
1/3 of meter accuracy, with a typical accuracy of 0.3% assumed for
current and voltage measurements [25]. The angle between phasors 3.6
is taken without error. Additionally, the estimation algorithm pro-
posed in [9] is implemented to reduce the error in the computing of 3.4
OHL resistance and average temperature.
For each weather sample, 1, 000 simulations to estimate the OHL 3.2
parameters were run by adding normal random errors to PMU
measurements. Figure 9a shows the impact of the error on the 3

measurements in the resistance estimation and an uncertainty of ap-

proximately 16 % was obtained. The uncertainty in this work is taken 2.8

as three times the standard deviation (σ). This uncertainty is equiva-

18.04 00:00 18.04 03:00 18.04 06:00 18.04 09:00 18.04 12:00 18.04 15:00 18.04 18:00 18.04 21:00
lent to an error within ±0.6 [Ω], assuming a normal distribution with Date
a mean between 3.6 [Ω] and 3.7 [Ω]. Thus, the error in the estima-
′
a
tion of the equivalent resistance per unit length Reqv is within
±10 × 10−3 [Ω/km]. This error propagates to the computing of
the temperature, reaching errors in the temperature estimation within
±38 [K], as shown in Figure 9b. The sensibility on the temperature 60

estimation due to the resistance is because of the influence of the co-

efficient α in (12). More specifically, the error propagates according
40
to (13). For the 470-AL3, 6469-AL3 /134ST4A and 774-AL3 con-
ductors, which are used in BR1-OHL, the error is within ±37 [K],
±34 [K] and ±36 [K], respectively. These values were calculated by 20
(13) and they are close to the values shown in Figure 9b; the dif-
′
ferences are due to the use of Requiv for computing the standard
deviation σr . 0

′
!
1 RT S -20
TS = ′ − 1 + Tref (12)
α RT ref
-40

dTS 1
σTS = ′ = αR′ σR (13)
dRT S Tref -60

18.04 00:00 18.04 03:00 18.04 06:00 18.04 09:00 18.04 12:00 18.04 15:00 18.04 18:00 18.04 21:00
Date

5.2 Impact of the load on the estimation of the temperature b

of the conductor
Fig. 9: Box plots with the value of both OHL resistance and TSavg
As the values of voltage and current magnitudes depend on the load estimated from PMU for each time sample, assuming an accuracy of
and the impedance of the OHL, the latter influence the estimation of 0.3% in voltage and current measurements
the resistance, and therefore, the computing of the temperature of the a OHL resistance
conductor. A simulation like the one of the previous section is car- b Average temperature of the conductor -TSavg
ried out for the weather sample 2016-04-18 21:00, changing the load
between 0.1 and 1 [pu] and the power factor (PF) between 0.1 and
0.95 . Figure 10 shows the simulations results, with the standard de- the computing of the resistance in a real OHL when PMU measure-
viation σ calculated from 500 runs for each set of
loads and PFs. In ments are used; therefore, it can be deduced that those results are a
the estimation of the equivalent resistance Requiv and the comput- consequence of error propagation in the computing of the resistance.
ing of the temperature in the conductor TS , the minimum standard
deviation was 0.027 [Ω] and 1.9 [K], respectively, for a power factor 
vk − vm

of 0.1 and a load of 1 [pu]. The maximum standard deviation was R ≈ Re (14)
2.41 [Ω] and 172 [K] for a power factor of 0.95 and a load of 0.1 ikm
[pu].
The impact of the load in the estimation of the temperature is s
explained by means of an error propagation on the measurements. 2 2 2
∂R ∂R ∂R
  
To carry out this analysis, it is assumed that the influence of the σR = σv + σv + σi
OHL capacitance is negligible (14) and the uncertainty is thus prop- ∂vk ∂vm ∂ikm km
agated according to (15). Given that σR ∝ 1/ikm , where ikm is ∂R cos (∠vk − ∠ikm )
the current intensity, the uncertainty in the computing of temper- =
∂vk |ikm |
ature is increased at low power flows. Additionally, if the power
factor (PF) is approximated to cos ∠ikm (using ∠vk = 0 as refer- ∂R cos (∠vm − ∠ikm )
ence, ∠vm close to ∠vk and ∠ikm measured with respect to ∠vk ) =−
∂vm |ikm |
the uncertainty increases as PF is close to 1. On the other hand, the
typical ratio between magnitudes of voltage (kV) and current (A) ∂R |vk | cos (∠vk − ∠ikm ) − |vm | cos (∠vm − ∠ikm )
=
in power transmission systems impacts the measurement error in the ∂ikm |ikm |2
resistance computing. For instance, [26] reports negative values in (15)

IET Gener. Transm. Distrib., pp. 1–9

c The Institution of Engineering and Technology 2017 IET Review Copy Only 7
 IET Generation, Transmission & Distribution
2.5
2 tionally, the error in the computing of conductor temperature is lower
1.5
1
0.5
0 Authors thank LANDSNET-Iceland, for providing the test cases in
0 0
0.5 0.5
1 1
a
200
150
100
50
0 sion Distribution, IET, 2013, 7, (2), pp. 130–137.
0 0
0.5 0.5
1 1
b Existing Overhead Transmission Lines’. (CIGRE, 2008).
Fig. 10: Influence of the load on the estimation of both: resistance
and temperature of the conductor using PMU measurements
a OHL resistance
b Average temperature of the conductor -TSavg
6 Discussion
The use of PMU measurements for DLR faces challenges when at-
mospheric conditions and properties of the conductor change along
overhead line, together with inaccuracy due to the error propagation
in the computing of the resistance. Thus, the average value of tem-
perature computed from PMU measurements cannot depict the real
capacity of the conductor, as shown in this paper. However, error
minimization techniques that account weather models, PMU and the
monitoring of critical spans could potentially improve the estimation
of conductor rating.
7 Conclusions
The changes in the load and the atmospheric conditions along an
OHL result in alteration of thermal and mechanical variables, which
affect the electrical RLC parameters. This influence is negligible for
inductance and capacitance under typical atmospheric and load con-
ditions, as a consequence of the small variation of the line length and
low impact of the sag on the capacitance. On the contrary, the value
of the resistance changes in a non-neglected way.
The computing of conductor average temperature solely from the
resistivity using PMUs jeopardizes the OHL when a DLR system
is implemented, as the conductor’s temperature in critical spans can
8 IET Review Copy Only
Page 8 of 9
exceed the accepted security margins. Since the estimation of OHL
resistance and the conductor’s temperature is performed through in-
direct measurements of voltage and current from PMUs, an error
propagation occurs, giving results outside the acceptable margins, a
consequence of the sensibility of temperature with resistance. Addi-
for low power factors and higher when less power is transmitted in a
line, even if state estimation algorithms are used.
Acknowledgment
section 3. This research was supported by the Colombian Depart-
ment of Science, Technology and Innovation (Colciencias) under the
project 617 - National Doctorates.
8 References
1 Winter, W., Elkington, K., Bareux, G., Kostevc, J.:
‘Pushing the Limits: Europe’s New Grid: Innovative
Tools to Combat Transmission Bottlenecks and Re-
duced Inertia’, Power and Energy Magazine, IEEE,
2015, 13, (1), pp. 60–74. Available from: http://
ieeexplore.ieee.org/stamp/stamp.jsp?
tp=&arnumber=6998972&isnumber=6998959
2 Sun, W.Q., Zhang, Y., Wang, C.M., Song, P.: ‘Flex-
ible load shedding strategy considering real-time dy-
namic thermal line rating’, Generation, Transmis-
Available from: http://ieeexplore.ieee.org/
document/6519364/
3 International des grands reseaux electriques. Comite
d’etudes B2, C.: ‘Guidelines for Increased Utilization of
Available from: http://www.e-cigre.org/
4 Douglass, D., Chisholm, W., Davidson, G., Grant, I.,
Lindsey, K., Lancaster, M., et al.: ‘Real-Time Over-
head Transmission-Line Monitoring for Dynamic
Rating’, IEEE Transactions on Power Delivery, 2016,
31, (3), pp. 921–927. Available from: http://
ieeexplore.ieee.org/lpdocs/epic03/
wrapper.htm?arnumber=6991585
5 Rob, S., Jean.Louis, L., Tap, S., Dale, D., Mark, L.,
Gerhard, B., et al.: ‘Guide for Application of Direct
Real-time Monitoring Systems’. (Paris: CIGRE, 2012).
Available from: http://www.e-cigre.org/
6 International des grands réseaux électriques. Joint work-
ing group B2-C1 (19), C.: ‘Increasing Capacity of
Overhead Transmission Lines: Needs and Solutions’.
(CIGRE, 2010). Available from: http://www.
e-cigre.org/
7 Fernandez, E., Albizu, I., Bedialauneta, M.T., Mazon,
A.J., Leite, P.T.: ‘Review of dynamic line rating sys-
tems for wind power integration’, Renewable and Sus-
tainable Energy Reviews, 2016, 53, pp. 80–92. Avail-
able from: http://www.sciencedirect.com/
science/article/pii/S1364032115007960
8 Alvarez, D.L., Rosero, J.A., Silva, F.F., Bak, C.L.,
Mombello, E.E., da Silva, F.M.F., et al. ‘Dynamic
Line Rating - Technologies and Challenges of PMU on
Overhead Lines: A Survey’. In: Proceedings of the
IET Gener. Transm. Distrib., pp. 1–9
c The Institution of Engineering and Technology 2017
Page 9 of 9 IET Generation, Transmission & Distribution
2016 51st International Universities’Power Engineering
Conference (UPEC). (Coimbra: IEEE Press, 2016.
9 Du, Y., Liao, Y.: ‘On-line estimation of transmission
line parameters, temperature and sag using PMU
measurements’, Electric Power Systems Research,
2012, 93, pp. 39–45. Available from: http://www.
sciencedirect.com/science/article/
pii/S0378779612002076
10 Rehtanz, C.: ‘Synchrophasor based thermal overhead
line monitoring considering line spans and thermal
transients’, IET Generation, Transmission & Distri-
bution, 2016, 10, (5), pp. 1232–1239(7). Available
from: http://digital-library.theiet.
org/content/journals/10.1049/iet-gtd.
2015.0852
11 Bockarjova, M., Andersson, G. ‘Transmission Line
Conductor Temperature Impact on State Estimation
Accuracy’. In: 2007 IEEE Lausanne Power Tech.
(IEEE, 2007. pp. 701–706. Available from: http://
ieeexplore.ieee.org/lpdocs/epic03/
wrapper.htm?arnumber=4538401
12 Frank, S., Sexauer, J., Mohagheghi, S.: ‘Temperature-
Dependent Power Flow’, IEEE Transactions on Power
Systems, 2013, 28, (4), pp. 4007–4018. Available from:
http://ieeexplore.ieee.org/lpdocs/
epic03/wrapper.htm?arnumber=6555947
13 Sivanagaraju, G., Chakrabarti, S., Srivastava, S.C.: ‘Un-
certainty in Transmission Line Parameters: Estimation
and Impact on Line Current Differential Protection’,
Instrumentation and Measurement, IEEE Transactions
on, 2013, PP, (99), pp. 1. Available from: http://
ieeexplore.ieee.org/document/6680707/
14 Cecchi, V., Leger, A.S., Miu, K., Nwankpa, C.O.: ‘Incor-
porating Temperature Variations Into Transmission-Line
Models’, IEEE Transactions on Power Delivery, 2011,
26, (4), pp. 2189–2196. Available from: http://
ieeexplore.ieee.org/lpdocs/epic03/
wrapper.htm?arnumber=5960820
15 Matus, M., Saez, D., Favley, M., Suazo.Martinez, C.,
Moya, J., Jimenez.Estevez, G., et al.: ‘Identification of
Critical Spans for Monitoring Systems in Dynamic Ther-
mal Rating’, Power Delivery, IEEE Transactions on,
2012, 27, (2), pp. 1002–1009. Available from: http://
ieeexplore.ieee.org/document/6163401/
16 Mai, R., Fu, L. ‘Dynamic Line Rating estima-
tor with synchronized phasor measurement’. In:
2011 International Conference on Advanced Power
System Automation and Protection. vol. 2. (IEEE,
2011. pp. 940–945. Available from: http://
ieeexplore.ieee.org/lpdocs/epic03/
wrapper.htm?arnumber=6180545
17 Stephen, R., Douglas, D., Mirosevic, G., Argasinska, H.,
Bakic, K., Hoffman, S., et al.: ‘Thermal behaviour of
overhead conductors’. (CIGRE, 2002). Available from:
http://www.e-cigre.org/
18 IEEE Std 738-2006: ‘IEEE Standard for Calculating
the Current-Temperature of Bare Overhead Con-
ductors’, IEEE Std 738-2006 (Revision of IEEE Std
738-1993), 2007, pp. c1–59. Available from: http://
ieeexplore.ieee.org/stamp/stamp.jsp?
IET Gener. Transm. Distrib., pp. 1–9
c The Institution of Engineering and Technology 2017 IET Review Copy Only
tp=&arnumber=6692858&isnumber=6692857
19 CIGRE: ‘Sag-Tension calculation methods for overhead
lines - BROCHURE 324’, , 2007, p. 91. Available from:
http://www.e-cigre.org/
20 Papailiou, K.O.: ‘Overhead Lines: A Cigre Green Book’.
(Cigre, 2014)
21 Tleis, N.: ‘Power Systems Modelling and Fault Analysis:
Theory and Practice’. Newnes Power Engineering Series.
(Elsevier Science, 2007)
22 Jiliusson, S.R. ‘Using PMU Measurements to Assess
Dynamic Line Rating of Transmission Lines’ [M.Sc.
Thesis in Electrical Power Systems and High Voltage
Engineering]. Aalborg University, 2013. Available from:
http://projekter.aau.dk/projekter/
files/77194901/Dynamic_Line_Rating.
pdf
23 Hosek, J., Musilek, P., Lozowski, E., Pytlak, P.: ‘Effect
of time resolution of meteorological inputs on dynamic
thermal rating calculations’, Generation, Transmission
Distribution, IET, 2011, 5, (9), pp. 941–947. Avail-
able from: http://ieeexplore.ieee.org/
stamp/stamp.jsp?tp=&arnumber=6017192&
isnumber=6017187
24 Albizu, I., Fernandez, E., Eguia, P., Torres, E., Mazon,
A.J.: ‘Tension and Ampacity Monitoring System for
Overhead Lines’, Power Delivery, IEEE Transactions
on, 2013, 28, (1), pp. 3–10. Available from: http://
ieeexplore.ieee.org/stamp/stamp.jsp?
tp=&arnumber=6313952&isnumber=6389787
25 IEEE Std C57. 13-2016: ‘IEEE Standard Requirements
for Instrument Transformers’, IEEE Std C5713-2016
(Revision of IEEE Std C5713-2008), 2016, pp. 1–96
26 Carlini, E.M., Pisani, C., Vaccaro, A., Villacci, D.
‘Dynamic Line Rating monitoring in WAMS: Chal-
lenges and practical solutions’. In: 2015 IEEE 1st
International Forum on Research and Technologies
for Society and Industry, RTSI 2015 - Proceedings.
(IEEE, 2015. pp. 359–364. Available from: http://
ieeexplore.ieee.org/lpdocs/epic03/
wrapper.htm?arnumber=7325124
9