Professional Documents
Culture Documents
Abstract. In this paper a new method is proposed that allows There are several important implications of the proposed
estimation of conductor temperature and cable losses as a change. Two of those will be discussed in this paper. One is a
function of conductor current and some parameters prediction of the conductor temperature and losses from the
corresponding to the nominal loading conditions. Such
current measurements alone and the other is the modification
parameters include cable rating and loss factors computed by
cable manufacturer for the laying conditions specified by the of the Neher-McGrath procedure for the calculation of the
user. There are two major applications of the proposed external thermal resistance.
approach. One relates to rating of power cables with current We will start by presenting mathematical apparatus
measurements alone and the other to the computation of the required to make these modifications and conclude by
external thermal resistance with a non unity load factor used in considering several practical examples.
the Neher/McGrath approach. Both applications are examined
in this paper.
Index terms. Power cables, real time rating systems, Neher- II. PREDICTION OF CONDUCTOR TEMPERATURE FROM THE
McGrath method, loss factor. CONDUCTOR CURRENT
Conductor temperature rise is related to the conductor
I. INTRODUCTION losses through the following equation (see Chapter 4 of [1])
The losses in a cable conductor are dependent on the
∆θ = n (WcTt + ∆θ d ) − θ x (1)
operating temperature. Since in field installations this
temperature is unknown, the estimation of the losses is a where
difficult task and, therefore, usually not performed at all or n = number of conductors in the cable.
some approximations are used [2]. The losses are also a ∆θ d = temperature rise caused by dielectric losses, (ºC).
function of daily load fluctuations. In the Neher/McGrath
Tt = total thermal resistance defined in (2), (K·m/W).
method these fluctuations are taken into account in the
calculation of the external thermal resistance by considering a T1
loss-load factor (referred to as a loss factor, for short). In Tt = + (1 + λ1 ) T2 + (1 + λ1 + λ2 )(T3 + T4 ) (2)
n
rating computations, the loss factor is calculated from the load
shape alone without consideration of the conductor A possible drying-out of the surrounding soil is taken into
temperature. In reality, the definition of the loss factor account in (1), following the two-layer-model of Cox and
requires consideration of the conductor resistance that is Coates [3] (see Section 4.2 in [1]), by means of the variables
temperature dependent. ν and θx with
The aim of this paper is to propose an improvement in the
accuracy of the rating calculations by an introduction of the ρ2
θ x = (vx − 1) ⋅ ∆θ cr and vx = (θ x = θ xT ) for ∆θ e > ∆θ cr
temperature dependence of the conductor losses in the rating ρ1 (3)
formulae. Since, as mentioned above, the conductor vx = 1 (θ x = 0) for ∆θ e ≤ ∆θ cr T
temperature is usually unknown, we will need some other
information to account for the absence of the real time where
temperature values. The additional information comes from ∆θ e = temperature rise at the cable surface, (ºC).
standard rating conditions as specified by the cable ∆θ cr = critical temperature rise of the soil, (ºC).
manufacturer and involves cable rating values at nominal θ xT = correction temperature, (ºC).
conditions.
ρ1 , ρ 2 = thermal resistivities of the moist and dry soils,
respectively, (K·m/W).
The thermal resistivity of the dry soil is normally larger
than that of the moist soil by a factor of 2 to 6. An example of
G.J. Anders is with Kinectrics Inc., Toronto, Ontario, Canada (e-mail: the standard values of these quantities is given in Table 1 [5]]:
george.anders@kinectrics.com
H. Brakelmann is with the Department of Electrical Engineering,
Duisburg University, Germany, (email: brakelmann@uni-duisburg.de).
TABLE 1 STANDARD VALUES FOR SOIL PROPERTIES
cm = 1 + α T ⋅ (θ max − 20 ) (11)
variable unit value The values Tt , R , ∆θ d , R , θ x , R are defined above, but here
θ amb °C 20
they are related to the rated load. We will compute now the
∆θ cr K 15 value of Rt 20 from equations (10) and (9) and substitute it in
ρ1 (K·m)/W 1 equation (7).
ρ2 (K·m)/W 2.5
∆θ − ∆θ d + θ x = Tt , x ⋅ n ⋅ Rt 20 ⋅ I 2 ⋅ (α T ⋅ ∆θ + cα )
θx K 22.5
(12)
2
We will rewrite (1) to relate conductor losses to the current vT I
= (∆θ max − ∆θ d , R + θ x , R ) ⋅ ⋅ ⋅ (α T ⋅ ∆θ + cα )
as a straightforward multiplication of the a.c. resistance and cm I R
squared current I 2 R . In order to achieve this, we will first
introduce the following revised designation of the total In equation (12) the variable ν T is given by
thermal resistance of the cable
Tt , x
T1 T ⋅ (1 + λ1 ) νT = for ∆θ e ≤ ∆θ cr , and
Tt , x = + 2 + T3 + vx ⋅ T4 = TI + vx ⋅ T4 (4) Tt , R
n ⋅ (1 + λ1 + λ2 ) 1 + λ1 + λ2
(13)
and the electrical resistance representing all ohmic losses in ν T = 1 for ∆θ e > ∆θ cr
the cable
A caution is required when interpreting equation (13).
Rt 20 = R20 ⋅ (1 + λ1 + λ2 ) (5) Three different situations may arise. If there is already drying
out for current I (we assume that I ≤ I R ) then ν T = 1 for both
R is temperature-dependent. Hence, the conductor
temperature θ changes the total ohmic losses of the cable, currents. In another case, we may have no drying out for
WI , according to current I but the soil dries out for the rated current. In this
case, the nominator in the first equation in (13) corresponds to
WI = Wt − Wd = W20 ⋅ 1 + α T ⋅ (θ − 20 ) the moist soil and the denominator to the dried out soil. The
temperature inequality associated with the first equation holds
(6) only for current I but not for the rated current. Finally, the
= n ⋅ Rt 20 ⋅ I ⋅ 1 + α T ⋅ (θ − 20 )
2 third possibility exists when no drying out occurs for the rated
current. In this case, ν T = 1 . In the reasoning made above we
where W20 are conductor losses at 20ºC, (W/m) and αT is the implicitly assumed that the loss factors that appear in the
temperature coefficient of resistance (1/ºC) of the conductor definition of the total thermal resistances (2) and (4) do not
material. depend on the current values. In reality, this is not true but the
With these notations, equation (1) can be rewritten as error introduced by this assumption is negligible.
Solving now equation (12) for the temperature rise, we
∆θ = Tt , x ⋅ n ⋅ Rt 20 ⋅ (α T ⋅ ∆θ + cα ) ⋅ I 2 + ∆θ d − θ x obtain
(7) 2
I
= Tt , x ⋅ (Wt − Wd ) + ∆θ d − θ x vT ⋅ cα ⋅ (∆θ max − ∆θ d , R + θ x , R ) ⋅ − cm (θ x − ∆θ d )
∆θ = IR (14)
with 2
I
cm − vT ⋅ α T ⋅ (∆θ max − ∆θ d , R + θ x , R ) ⋅
cα = 1 + α T ⋅ (θ amb − 20 ) (8) IR
For the nominal load, characterized by the rated current I R Making substitutions
and the nominal total losses WtR , the cable conductor reaches θ1 = cα ⋅ (θ max − θ amb − ∆θ d , R + θ x , R ) ⋅ vT (15)
its maximum stationary temperature θ max . In this case,
equation (7) takes the form θ 2 = cm ⋅ (∆θ d − θ x ) (16)
∆θ max = Tt , R ⋅ (WtR − WdR ) + ∆θ d , R − θ x , R (9)
(
c1 = 1 + α T ⋅ θ amb − 20 + ∆θ d , R − θ x , R
) (17)
whereas
c1 + c2 ⋅ 1 − vT ⋅ T1
Td = + T2 + T3 (27)
I R 2⋅n
From equations (24) to (27) we can compute the critical
2
I current as
= Wd + (WtR − Wd ) ⋅ ⋅ vθ
IR 1
I cr =
(20) n ⋅ Rt 20
(28)
where the temperature coefficient vθ is defined as ∆θ cr / T4 − Wd
⋅
1 + α T ⋅ [ ∆θ cr + θ amb − 20 + TI ⋅ ∆θ cr / T4 − (TI − Td ) ⋅ Wd ]
c3
νθ = (21)
I
2
T1 T ⋅ (1 + λ1 )
Tt , R = + 2 + T3 + vx ⋅ T4
n ⋅ (1 + λ1 + λ2 ) 1 + λ1 + λ2
Figure 1 Conductor temperature as a function of conductor current for the 3-
conductor armored cable
= 0.269 + 3 ⋅ 0.299 = 1.165 K ⋅ m/W
The temperature coefficient ν T is calculated from equation The cable losses are computed next. First, we obtain the
temperature coefficient from equation (21)
(13) as
c3 0.983
νθ = = = 0.841
Tt , x 0.568 I
2
600
2
On the other hand, for the current I = 600 A , no drying out From equation (20), we now compute the total losses in the
occurs and θ x = 0. cable
From equations (15) to (18), we obtain I
2
600
2
Wt = Wd + (WtR − Wd ) ⋅ ⋅ vθ = 3 ⋅ 0.52 + (41.53 − 0.52) ⋅ ⋅ 0.841 = 41.49 W/m
I
R 966
The cable losses computed from (20) as a function of IV. PRACTICAL APPLICATION OF THE TEMPERATURE
conductor current are shown in Figure 2 (dashed line). The PREDICTION EQUATION
solid line represents the joule losses computed without A profitable use of equations (19), (23) and (28) is possible,
considerations of temperature variation of resistance; i.e., for if the corresponding cable data are known. In this case, for all
νθ = 1 . operating conditions, the stationary conductor temperature as
well as the cable losses can be determined by inserting the
actual current values into these simple analytical formulae.
One possibility is to execute a complete current rating
analysis, thus determining all thermal resistances, loss factors
and other relevant parameters. This is normally done by the
cable manufacturer, since they have to their disposal all
construction data as well as a suitable rating program.
The extraction of the necessary coefficients
θ1 ,θ 2 , c1 and c2 from the cable data without any rating
calculations is possible, if the following nominal values for
full-load operation are given:
• the rated current , I R
Figure 2 Total losses for the 3-conductor armored cable as a function of
conductor current and the ratio of losses. • the dielectric losses (p.u.l.), We ,
• ambient soil temperature, θ amb .
The line at the top of the graph shows the ratio of the losses
computed without and with the temperature correction. The These three parameters are easily obtainable from the cable
maximum error, in this case, exceeds 20% for the currents in specifications. With these values, equation (19) can be used
the usual loading range of this type of the cable; that is, for the prediction of conductor temperature for a given current
between 300 and 600 A. with an assumption that the soil dry out does not occur. Since
Assuming that the critical temperature rise at which drying this is the situation most often encountered in practice, the
out starts for this installation is 35ºC, the critical loading method presented in this paper can be used as a simple
current can be computed from (28). In order to perform this alternative to the full real-time temperature monitoring
calculation, we need to obtain first the internal thermal system.
resistances from (4) and (27), and the conductor resistance at If, on the other hand, there is a need for the calculation of
20ºC. The internal equivalent thermal resistance was cable losses, or if a soil dry out may occur during cable
computed above as TI = 0.269 K ⋅ m/W . operation, we will require two additional quantities; namely,
• the total rated losses (p.u.l.), WtR of the cable
T1 0.754 system, and
Td = + T2 + T3 = + 0.04 + 0.047 = 0.213 K ⋅ m/W • the cable surface temperature, θ e .
2⋅n 6
R ⋅ (1 + λ1 + λ2 ) From these data, the external thermal resistance T4 can be
Rt 20 = computed as
1 + αT ⋅ (θ max − 20°C)
θ e − θ amb
T4 = (29)
0.000033 ⋅ (1 + 0.123 + 0.210) WtR
= = 0.0000345 ohm/m
1 + 0.00393 ⋅ (90 − 20)
The total thermal resistances of the cable can be
approximated by
The critical conductor current is equal to
′ = (θ max − θ e ) / WtR + T4
Tt , x ≈ Td , x ≈ Tapp (30)
1
I cr = The approximation given by equation (30) results in a very
n ⋅ Rt 20 small error for the cables with small dielectric losses.
∆θ cr / T4 − Wd As a substitute for the cable surface temperature and the
⋅ total cable losses, we can assume soil thermal resistivities for
1 + α T ⋅ [ ∆θ cr + θ amb − 20 + TI ⋅ ∆θ cr / T4 − (TI − Td ) ⋅ Wd ]
moist and dry conditions, and proceed with the calculations
accordingly.
1
=
3 ⋅ 0.0000345 V. EVALUATION OF THE LOSS FACTOR CALCULATIONS IN
RATING STANDARDS
35 / 0.299 − 0.52
⋅ = 952A
1 + 0.00393 ⋅ [35 + 15 − 20 + 0.269 ⋅ 35 / 0.299 − (0.269 − 0.213) ⋅ 0.52] A. Modification to the Neher-McGrath formula
The developments presented in the preceding sections will
now be used to discuss an issue of loss factor application in
cable rating.
In North America1, the variations of the daily load cycle are
taken into account by modifying the external thermal
resistance T4 , as follows [4].
ρ s Dx 4L
T4 = ln + µ ln (31)
2π De Dx
where
ρs = soil thermal resistivity, K·m/W,
Dx = diameter at which the effect of daily load
variations starts, mm. Its value is normally taken
to be 211 mm [4].
L = depth of cable center, mm,
De = external diameter of the cable or pipe, mm, and
µ = the loss factor.
The loss factor is defined by
τ
1
µ=
Wmax ⋅τ
⋅ ∫ W (t ) ⋅ dt
t =0
(32)
Figure 3 Two-step load function applied to the 3-conductor armored cable.
This is usually approximated by
The load in Figure 3 is characterized by 8 hours of full load
τ
µI =
1
⋅ ∫I
2
(t ) ⋅ dt (33) ( I = I1 ) , followed by 16 hours of a lower load ( I = I 2 ;
2
I max ⋅τ t =0 possibly an example for an industrial load). It is assumed, that
However, taking into account variations of the resistance the peak value I max of the load cycle corresponds to 60% of
with temperature, equation (33) can be written as the rated current I R of the cable.
τ The load and loss factors calculated from the traditional
1
µθ = ∫I (t ) ⋅ν θ (t ) ⋅ dt
2
⋅ (34) equations (36), (35) and (33), are computed next.
2
I max ⋅τ t =0
τ 24
1 1
In engineering practice, the loss factor is approximated LF =
I max ⋅τ
⋅ ∫ I (t ) ⋅ dt = 1⋅ 24 ∑ I (t ) = 0.73
t =1
from the known or assumed load factor LF as (Anders, 1997) t =0
c1 + c2 ⋅ 1 − ν T
I R
0.983
= = 0.836
1
Most European standards (excluding Germany), which in the majority of 1
2
cases are based on the IEC standards, do not introduce a notion of the loss 0.712 + 0.563 ⋅ 1 − 0.488 ⋅
factor. Instead, calculations involving varying loads are referred to a separate 1/ 0.6
standard using full time dependent analysis [6].
From equation (34), we have At low load conditions, the error is the largest. It can also be
τ
quite large for the usual cable loading conditions in the range
1 1 24 2
µθ =
I 2
⋅τ
⋅ ∫I
2
(t ) ⋅ν θ ⋅ dt = ∑ I (t ) ⋅νθ = 0.469
1 ⋅ 24 t =1
of 40 to 80% of the rating. The effect on ampacity will
depend on the cable construction. In the case presented in this
max t =0
paper, the error was in the range of 4%. Fortunately,
The corresponding values of the external thermal resistance
traditional calculation methods give conservative ratings.
and the ratings are shown in Table 3.
VIII. BIOGRAPHIES
George J. Anders (M’74, SM’84,
F’00) received a M.Sc. degree in EE from
Technical University of Lodz in Poland in
1973, and an M.Sc. degree in Mathematics
and Ph.D. degree in Power System
Reliability from the Univ. of Toronto in
1977 and 1980, respectively. Since 1975,
he has been employed by Ontario Hydro
first as a System Design Engineer in
Transmission System Design Dept. of
System Planning Division and currently as a Principal Engineer in Kinectrics
Inc. (a successor company to Ontario Hydro Technologies). Dr. Anders is the
author of two books “Probability Concepts in Electric Power Systems",
published by John Wiley & Sons, NY, 1990, and “Rating of Electric Power
Figure 4 Ratio of traditional loss factors to the temperature-dependent one. Cables”, published by IEEE Press, New York, 1997 and McGraw-Hill, New
York, 1998. He is a registered Professional Engineer in the Province of
We can observe that substantial errors can occur in the Ontario.
Heinrich Brakelmann (54), VDE, is
evaluation of the loss factor when the temperature dependence head of the Institute for Energy-Transport
is neglected. and -Storage of the faculty for engineering
sciences of the Gerhard-Mercator-University
of Duisburg. His main research fields are the
VI. CONCLUDING REMARKS power cable technologies (ETG-award in
In order to obtain accurate conductor operating temperature 1982). He received a M.Sc. degree in EE
(Dipl.-Ing.) from the Technical University of
and corresponding losses temperature sensing devices are Aachen, where he was promoted to Dr.-Ing.
sometimes installed on a cable or pipe surface. Installation of in the field of circuit-breakers in 1973.
such devices is often very costly, or outright impossible. In Between 1973 and 1977, he was
employed by the cable-company
such cases, an approximate method of real-time rating analysis Felten&Guilleaume, Cologne, as a Principal Engineer for power cable
based on the current measurements alone may prove to be systems. Since 1977 he is a member of the faculty of the Gerhard-Mercator-
very beneficial. Such a method has been presented in this University of Duisburg and since 1992 professor for electrical power
transmission. At time he is a member of CIGRE WG21-05 („transient effects
paper. on long cables“). Prof. Brakelmann is author of more than 100 technical
The calculation of the loss factor as a function of conductor publications and of three book-publications in the field of cable technique as
temperature has also been introduced. The error that is well as high-voltage and high-current problems
obtained in estimating the conductor temperature will depend
on the loading of the cable with reference to the nominal load.