Sunteți pe pagina 1din 6

RADIAL TEMPERATURE DISTRIBUTION OF AAAC OVERHEAD LINE

IN STATIONARY AND TRANSIENT CONDITIONS


M. Kang
*
, M. Strobach and C. M. Franck
Power Systems and High Voltage Laboratories, ETH Zurich,
Physikstrasse 3, 8092 Zurich, Switzerland
*Email: kang@eeh.ee.ethz.ch

Abstract: Commonly, dynamic line rating (DLR) calculates conductor temperature with a
basic energy balance equation, which assumes homogeneous conductor temperature.
With this 0D temperature assumption, however, there is a risk that inner layers of the
conductor might exceed the limit temperature, if the rating current approaches its
ampacity. Since conductor overheating should be avoided for operational safety, it is
valuable to investigate radial temperature distributions of conductors. Throughout this
study, therefore, homogeneous 1D conductor model with a radial temperature distribution
is simulated in steady and transient states. For steady state simulations, radial
temperature distributions under various ambient conditions are studied. Higher wind
velocity, lower solar radiation, and lower ambient temperature make the temperature
gradients more pronounced. Also, transient temperature distribution is modeled with a
various stepwise current changes. When the core temperature reaches 80C,
momentous radial temperature distributions are analyzed. Compared to the steady state,
transient phases show smaller temperature differences between core and surface. Both
lower initial and higher final currents result in a flatter radial temperature distribution,
whereas higher final currents cause faster transient heating. Compared to 0D calculations,
the 1D model with a radial temperature distribution reduces the ampacity and prevents
conductor core overheating.


1 INTRODUCTION
Aluminum is the most frequently used material [1,
pp. 200] for overhead lines (OHLs) because of its
advantages. By mixing magnesium and silicon into
aluminum, all aluminum alloy conductor (AAAC)
achieves more than 1.5 times higher strength,
while sacrificing only 5% of ampacity (current
carrying capacity, I
ccc
) [1, pp. 206].
Despite its advantages, the usability of AAAC is
limited by its low operation temperature. AAACs
start recrystallization when they are heated up to
over 100C, and lose the mechanical strength [1,
pp. 202]. Also, its high thermal expansion
coefficient and low specific weight make AAACs
more vulnerable to sagging and vibration [1, pp.
206]. For these reasons, the maximum allowable
temperature T
lim
of AAAC is limited to 80C [3]. T
lim

determines the ampacity under the given ambient
conditions [4], since higher temperatures result in
larger sags and worse mechanical integrity of
conductors.
Therefore, Cigre [5], IEC [6], and IEEE [4]
developed standard OHL models analyzing the
conductor temperature. To simplify the calculation,
the models commonly disregard the radial
temperature distribution, T(r) of the conductor.
These non-dimensional (0D) models are beneficial
for the static line rating (SLR). In SLR, the rating
currents are kept low for security reasons [10] and
enough margins exist between allowable and
actual line temperatures. Therefore, the core
temperatures remain lower than the allowable
temperature in general.
The 0D models might not be suitable for dynamic
line rating (DLR). DLR offers an improved
temperature management of OHL conductors. It is
enabled by accessing and real-time processing of
on-site conductor temperature and weather
information. DLR then not only optimizes current
loading of OHLs but also, minimizes operational
risk by improving security and safety. Therefore,
the conductors can be operated near T
lim
. For this
case, the 0D models may result in conductor core
overheating, because they ignore the T(r)
distribution.
To analyze the limits of 0D models on DLR, a
homogeneous cylindrical conductor is modeled
and its radial temperature distribution is studied.
The radial temperature distribution was researched
extensively by Morgan [7], and its significance on
DLR was already suggested by Douglass [8].
Nonetheless, parametric researches on stationary
and transient behavior of AAACs are still required.
Throughout this paper, therefore, the conductor
temperature T(r) is simulated with a 1D model for
various stationary and transient states. Several
groups of relevant weather conditions are varied
for steady state simulation, and several current
steps are applied to transient phase simulation.
PF-03
1970
2 1D CONDUCTOR MODELING
2.1 System and environment
Conductor: For the modeling, the conductor is
simplified as a homogeneous cylinder with infinite
length. To study T(r) only, a uniform surface
temperature T
s
without angular and axial
temperature distribution is applied. An AAAC with a
conductor cross section A=550mm
2
is used as a
reference material. It consists of 61 strands with
d=3.4mm, and has total conductor diameter
D=30.6mm. As the conductor geometry is
comparable to 61/3.5mm AAC in [7], the same
effective radial thermal conductivity k
r
= 1.23 W/mK
is assumed. The other typical material properties of
AAAC are adopted from [5], and presented in
Table 1.
Table 1: Reference AAAC properties at 20C
Properties Value
Diameter, D 30.51 [mm]
DC Resistivity, 32.7 |nOm]
Density, 2703 |kg/m
3
|
Heat capacity, c 909 |J/kgK|
Absorptivity, o 0.5
Emissivity, c 0.5
Effective radial thermal conductivity, kr 1.23 [W/mK]

Environment: Ambient temperature T
a
,
perpendicular wind velocity V
w
and solar radiation
S are applied around the whole conductor
circumference. Utilities in Switzerland assume
conservative weather condition for the line rating
[9]. For the operational safety, V
w
and S are fixed
to a worst-case value while T
a
varies to apply
seasonal weather differences [10]. The conditions
are shown in Table 2.
Table 2: Seasonal weather conditions
Parameters
Ta [C] Vw [m/s] S [W/m
2
]
Winter 10
0.5 900 Intermediate 20
Summer 40

2.2 Heat equations
The conductor system and environment satisfy the
thermal equilibrium at the steady state [5]. The
power gain of the system is the sum of Joule,
ferromagnetic, solar, and corona heating per unit
time (P
J
, P
M
, P
S
, and P
i
, respectively). It balances
the power loss, which is the sum of convective,
radiative and evaporative cooling per unit time (P
c
,
P
r
, and P
w
, respectively). According to [7], the
reduced equation for AAAC conductors is:


The T(r) distribution is described by the heat
equation in the form of [7]:


where r is the conductor radius (0<r<D/2) and P
J
is
Joule heating power per unit length. The boundary
conditions for (2) are:
T has a finite value at r=0,

, and


For AAACs, [5] assumed P
J
as


where k
j
=1.023 is a factor for accounting skin
effect, I (A) is an effective current, R
dc
(O) is the DC
resistance at 20C ,

(1/K) is the thermal resistive


coefficient, and T
av
is an average line temperature
which is given by:


This is used for conductor temperature in the Cigre
0D model. In addition, the temperature difference
between T
c
and T
s
of AAAC is [5]:


and it is used for analyzing the steady state
temperature gradient.
For the transient phase, any difference between
heat gain and loss will change the conductor
temperature [5]. This can be expressed as:

} {

}
where,

(kg/m).
2.3 Simulation methods
The heat transfer module of COMSOL Multiphysics
is used for the 1D model simulations. For the
steady state and transient phase evaluation,
stationary and time-dependent studies of the
program are used, respectively.
Steady state: The steady state T(r) distribution for
T
lim
(=80C) is simulated with different weather
conditions. Each weather parameter is varied to
analyze the influence on conductor temperature. In
addition to the weather conditions in Table 2,
PF-03
1971
different values of V
w
(0.5, 2.5, and 4.5m/s) and S
(500, 900, and 1200W/m
2
) are used in the
simulation. For each condition, I
ccc
is calculated.
These values are then used as references for
transient phase simulation.
Assuming similar weather conditions, steady state
ampacity from 1D and 0D models are calculated
and compared. All of them use the same maximum
allowable temperature,

1D conductor
model has the maximum temperature at the
conductor core, and the ampacity is the current at

.
Both IEEE and IEC models ignore the T(r)
distribution and the ampacity is the current at

. Among the 0D models, only Cigre model


uses simplified assumptions for the radial
temperature distribution, and its ampacity is
defined as the current at

Although Cigre
model denotes temperature differences between


and

[5], the radial temperature distribution is not


considered for its ampacity analysis.
Transient phase: Transient simulations are done
with the winter weather condition of Table 2.
Therefore, the steady state ampacity of this
condition is used as a reference. For the transient
phase, the allowable time interval t
lim
is defined as
the elapsed time to raise the conductor
temperature to T
lim
. By raising the current stepwise,
transient T(r) distributions at t = t
lim
are
investigated.
The stepwise change is done from initial current I
i

(30% and 60% I
ccc
) to final current I
f
(120%, 150%,
and 200% I
ccc
). These situations should be
investigated extreme situations for a line operating
first in (N)- and then in (N-1)-condition. The plans
for the transient current change are presented in
Table 3.
Table 3: Transient current change plans

(% Iccc)
120% 150% 200%



(
%

I
c
c
c
)

30%
Plan A
(30120)
Plan B
(30150)
Plan C
(30200)
60%
Plan D
(60120)
Plan E
(60150)
Plan F
(60200)

3 RESULTS AND DISCUSSION
3.1 Influences of weather parameters on
steady state temperature distribution
By limiting T
c
=T
lim
=80C, the steady state T(r)
distributions for different weather conditions are
obtained (See Figure 1). Since each weather
parameter influences the elements of heat balance
equation (1), T(r) and ampacity are varied with
parameter change. In addition to seasonal weather
conditions in Table 2, each weather parameter is
manipulated to investigate steady state T(r)
distribution.
Figure 1: T(r), comparison with different T
a
(left),
V
w
(center) and S (right), when

.
Among the weather parameters, V
w
has the largest
influence on T(r), and S has the smallest. The
equilibrium ampacity increases significantly with V
w
,
as the heat loss P
c
increases with the wind velocity.
Analysis for each parameter is done below.
Ambient temperature: To control the influence of
other parameters, fixed value of V
w
=0.5m/s and
S=900W/m
2
are used for T
a
simulation. Lower T
a

results in larger T(r) gradients. Both P
c
and P
r
of (1)
increase with decreasing T
a
[5]. Therefore, higher
currents are needed to increase P
J
for equilibrium.
Since T
a
is the only variable of the static weather
conditions, this part of simulation result can be
used for SLR as well. To reach a core temperature
of 80C under the three different values of T
a
=10,
20, and 40C, current values of I=962.3A, 886.0A,
and 703.8A are needed, respectively.
Wind velocity: To study the influence of V
w
only,
fixed values of T
a
=10C and S=900W/m
2
are used
for this set of simulation. High wind velocities
decrease T
s
significantly and cause larger T(r)
gradients at the ampacity limit. P
c
strongly depends
on V
w
[5]. For the equilibrium of (1), ampacity
increase with V
w
. To reach a core temperature of
80C under the three different values of V
w
=0.5,
2.5 and 4.5 m/s, the current values of I=962.3A,
1395.7A, and 1670.8A are needed, respectively.
As mentioned above, the wind velocity is the most
influential parameter.
Solar radiation: Similarly, to cancel out the
influence of the other parameters, T
a
=10C and
V
w
=0.5m/s are used. Decreasing the solar
radiation causes a reduction of the total heat gain
P
S
. When the other conditions are kept fixed, P
J

has to be increased to satisfy (1) with a higher
current [5]. To reach a core temperature of 80C
under the three different values of S = 500, 900,
and 1200 W/m
2
, current values of I=990.3A,
962.3A, and 940.7A are needed, respectively.
Compared to the other parameters, solar radiation
has relatively small influence on T(r) distribution.
PF-03
1972
3.2 Steady state ampacity comparison
The ampacity values for the standard weather
conditions are presented in Table 4 for the 0D
models according to IEEE and Cigre, as well as for
the 1D model. Respecting the T
lim
=80C value for
the core temperature results in a reduction of line
ampacity as well.
Table 4: Seasonal ampacity comparison of
different models
0D Iccc (A) 1D Iccc (A)

IEEE
(Ts=80C)
Cigre
(Tav=80C)
COMSOL
(Tc=80C)
Winter 1069.9 1024.4 962.3
Intermediate 996.9 953.0 886.0
Summer 825.3 786.5 703.8

As shown in Figure 2, if the ampacity value
calculated from the IEEE (or IEC) 0D model would
be used in the 1D model, the whole conductor
temperature (except its surface) exceeds T
lim
and
reaches a core temperature of 86.7C. When the
Cigre 0D model is used, half of the conductor area
exceeds T
lim
, and T
c
=83.2C Therefore, the T(r)
distribution should be properly considered to
respect the line security.
Figure 2: T(r) comparison of COMSOL 1D (solid,
I
ccc
=1069.9A), Cigre 0D (dashed, I
ccc
=1024.4A),
and IEEE 0D (dot, I
ccc
=962.3A) models at Winter
ambient condition.
The core temperature increase becomes worse
when the weather conditions allow a higher
ampacity, since then P
J
and AT increase in
equations (4) and (6).
To analyze the influence of overrated ampacity, T(r)
distributions with IEEE 0D model ampacity are
presented (see Figure 3). Among the simulation
results, the maximum T
c
=105C is obtained under
the weather condition of T
a
=10C, V
w
=4.5m/s, and
S=900W/m
2
. [9] shows that this weather condition
occurs frequently in Switzerland, and it may even
remain for a long time. According to [1, pp. 203],
the mechanical strength of aluminum decreases
after long-term exposures to high temperature.
After 100 hours, the residual strength is around 80%
of the initial strength, and after 1000 hours, it
decreases to 60%.
Figure 3: T(r) comparison with different T
a
(left),
V
w
(center) and S (right), when

.
3.3 Transient phase analysis
The influence of initial and final current is analyzed.
The ampacity of the 1D model at Winter condition
(I
ccc
=962.3A, see Table 4) is used as a reference
ampacity of the transient phase simulations. From
the initial currents 30% and 60% I
ccc
, Initial steady
state T
c
=23.4C and 38.4C are obtained,
respectively (see Figure 4). Currents for the initial
and final states are presented in Table 5.
Table 5: Initial and final states for transient phase
Initial currents Final currents
% Iccc 30% 60% 120% 150% 200%
I (A) 289 577 1155 1443 1925

The temporal evolutions of

under the 6 different


current steps are depicted in Figure 4. Qualitatively,
higher initial current decreases the heating
duration, and results in smaller

Also, the
temperature rises faster when final current
becomes higher.
Tlim=80C
PF-03
1973

Figure 4: t vsT
c
with various current plans
Influence of I
i
on T(r): With a lower initial current,
it is obvious that

increases. Also, the


temperature distribution becomes slightly flatter
because the initial steady state T(r) distribution is
flatter with a smaller initial current. However, the
influence of I
i
has only minor effect on conductor
temperature distribution (cf. Fig. 5). This is
because the Joule heating power caused by the
overcurrent is not large enough compared to the
radial heat flux to justify complete adiabatic heating.

Figure 5: T(r) comparison of transient case C and
F
Influence of I
f
on T(r): As higher I
f
increases the
temperature rise rate,

decreases. In other
words, the time to lose heat to environment is
reduced and the process becomes more adiabatic.
Therefore, the momentous temperature distribution
is not changed much from the shape of its initial
steady state T(r) distribution, if the applied final
temperature is higher. However, I
f
of 120% to 200%
ampacity is not enough to observe a dramatic
difference in T(r) distribution. For those final
current values, the momentous T(r) distribution at
Tc=80C is not very different form the steady state
T(r) curve (see Figure 6). With higher I
f
, the
surface temperature increases, and

difference
between each transient case is around 0.1C.


Figure 6: T(r) comparison of transient case D, E,
and F

3.4 Time variables of transient phases

Table 6: allowable time interval for different
models
Plans
(change in % Iccc)
0D

(s)
,Ts= Tlim
1D

(s)
,Tc= Tlim

(s)
A (30120) 1779 1479 300
B (30150) 784 684 100
C (30200) 366 328 38
D (60120) 1606 1263 343
E (60150) 647 548 99
F (60200) 288 248 40

with and without T(r) distribution is compared


and their differences are presented in Table 6. t
lim

decreases by increasing I
f
. For all plans, t
lim
is
smaller for the 1D model. If a 0D model would be
used for transient phase analysis, there is a risk of
overheating the wire for t
lim
, as T
c
exceeds T
lim
for
the duration.
4 CONCLUSION
The possible benefits of considering the radial
temperature distribution for DLR are elucidated by
a simple 1D conductor model. If 0D models are
used for steady state ampacity calculation, T
c

might exceed T
lim
by up to 25C even in
reasonable weather conditions. Wind speed is the
most influential parameter for the radial
temperature gradient, as it has the largest effect for
Tlim=80C
PF-03
1974
increasing the ampacity. On the other hand,
changes in solar radiation do not affect the T(r)
distribution significantly.
0D models neglect the T(r) distribution and assume
T
s
or T
c
as conductor temperature. This could be
dangerous, since the long-term exposure to high
temperatures permanently deteriorates the
mechanical integrity of the conductor. This can be
avoided by considering a radial temperature
distribution. Even with the simple 1D model, the
overheating of conductor core can be prevented.
By applying the model to calculation, ampacity at
different weather conditions is reduced by 10 to
18%. This modified steady state ampacity with
each weather condition might offer accurate
information for the overhead line operators.
Also, with a stepwise current change, transient
phases are simulated. Intensity of initial current
and final current is the governing factors for the
transient characteristics. With a lower initial current,
the transient temperature profile is flatter. Higher
final current results in faster temperature increase.
Also, the T(r) becomes flatter with higher I
f
, by
keeping the conductor away from the heat transfer.

can be adjusted with the 1D model. By


regulating current with T
c
, the 1D model is able to
control the overhead line temperature. Compared
to 0D models, it prohibits the extra time for heating
the inner layers of a conductor. This is important
for real time controlling of current with DLR.
5 ACKNOWLEDGMENTS
This study is financially supported by Swiss electric
research, Pfisterer, and Swissgrid.











6 REFERENCES
[1] F. Kiessling et al., Overhead Power Lines,
Springer, 2003.
[2] M. Farzaneh et al., Electrical Design of
Overhead Power Transmission Lines,
McGraw-Hill, 2013.
[3] EN 50331-3-4: Overhead electrical lines
exceeding AC 45 kV. Part 3-4: National
Normative Aspects for Germany. Brussels,
CENELEC, 2001.
[4] Standard for Calculating the Current-
Temperature of Bare Overhead Conductors,
IEEE Std 738, 2006.
[5] Thermal Behavior of Overhead Conductors,
Cigre Working Group 22.12, August 2002.
[6] Overhead electrical conductors-Calculation
methods for stranded bare conductors, IEC TR
61597, 1995.
[7] V. T. Morgan, The radial Temperature
Distribution and Effective Radial Thermal
Conductivity in Bare Solid and Stranded
Conductors, IEEE Transactions on Power
Delivery, pp. 1443-1452, 1990.
[8] D. A. Douglass, Radial and Axial Temperature
Gradients in Bare Stranded Conductors, IEEE
Transactions on Power Delivery, Vol. PWRD-1,
No. 2, April 1986.
[9] Leitungsverordnung LeV, SR 734.31, 30.
March 1994.
[10] Guide for Selection of Weather Parameters for
Bare Overhdad Conductor Ratings, Cigre
Working Group B2.12, 2006.



PF-03
1975

S-ar putea să vă placă și