Determining high voltage cable conductor temperatures

DETERMINING HIGH VOLTAGE CABLE
CONDUCTOR TEMPERATURES.
Guy Van der Veken.
Euromold, Belgium.
INTRODUCTION.
INVESTIGATIONS.
Type tests on MV cable accessories are
described in CENELEC HD628 and HD629
documents.
Some of the tests described require elevated
conductor temperatures within strict limits (e.g.
5K to 10K above the maximum permissible
operating temperature of the extruded cable
insulation).
Following points have been evaluated:
1) Validity of the methods as described.
2) Uncertainty of the results obtained.
3) Comparison of the two methods.
To accomplish this, over the allowed range of
ambiant temperatures, the heating current is to
be regulated.
Due to the presence of high test-voltages across
the cable’s insulation, the on-line measurement
of the conductor temperature on the tested cable
is not possible using standard measuring
techniques.
3 methods for determining the cable temperature
are given in the document HD628:
1) Method 1 using the relationship between the
conductor temperature, the heating current and
the ambiant temperature.
2) Method 2 using the relationship between the
conductor temperature, the heating current and
the cable-jacket temperature.
(These two methods require a preceding
calibration of the cable, to establish these
relationships.)
3) Method 3 using a parallel loop of same cable
in the same environment that is heated with the
same current, but is not carrying high voltage.
For this purpose, following variables have been
examined:
a) Thermocouple materials.
b) Thermocouple execution.
c) Thermocouple placement.
d) Effect of conductor cross-section.
e) Number of thermocouples used.
Uncertainty factors evaluated include:
a) Uncertainty of the measuring equipments.
b) Uncertainty of the measurements.
c) Uncertainty of the calculated temperature.
RESULT
Evaluation of the data leads to following
conclusions:
1) The method 2, using jacket temperature
measurement, results in the lowest deviation.
2) Uncertainty of the temperature determined
(± 3K for small crossections to ± 5K for large
crosssections) is found to be high when
compared to the temperature range given (5K).
DÉTERMINATION DE LA TEMPÉRATURE DE
CÂBLES CONDUCTEURS À HAUTE TENSION
Guy Van der Veken.
Euromold, Belgique
INTRODUCTION.
INVESTIGATIONS.
Les tests de type sur les accessoires de câbles
MV sont décrits dans les documents CENELEC
HD628 et HD629.
Certains
tests
décrits
requièrent
des
températures élevées des conducteurs dans des
limites strictes (p.e. 5K ou 10K au-dessus de la
température de service maximum autorisée de
l’isolation extrudée du câble).
Pour réaliser ceci dans les limites de température
ambiante autorisées, le courant de chauffage doit
être réglé.
En raison de la présence de hautes tensions
d’essai sur l’isolation du câble, la mesure en
ligne de la température du conducteur du câble
testé est impossible avec des techniques de
mesure standard.
Le document HD628 indique 3 méthodes pour
déterminer la température du câble :
1) La méthode 1 utilise la relation entre la
température du conducteur, le courant de
chauffage et la température ambiante.
2) La méthode 2 utilise la relation entre la
température du conducteur, le courant de
chauffage et la température du manteau du
câble.
(Ces deux méthodes requièrent un calibrage
préalable du câble pour établir ces relations.)
3) La méthode 3 utilise une boucle parallèle de
câble identique dans le même environnement,
qui est chauffé avec le même courant, mais ne
transporte pas de haute tension.
Les points suivants ont été évalués :
1) La validité des méthodes décrites.
2) L’incertitude des résultats obtenus.
3) La comparaison des deux méthodes.
Pour ce faire, les variables suivantes ont été
examinées :
a) Les matériaux des thermocouples.
b) L’exécution des thermocouples.
c) L’emplacement des thermocouples.
d) L’effet de la section du conducteur.
e) Le nombre de thermocouples utilisé.
Les facteurs d’incertitude évalués comprennent :
a) L’incertitude des appareils de mesure.
b) L’incertitude des mesures.
c) L’incertitude de la température calculée.
RÉSULTAT
L’évaluation des données mène aux conclusions
suivantes :
1) La méthode 2, qui utilise la mesure de la
température du manteau, produit le plus faible
écart.
2) L’incertitude de la température déterminée (±
3K pour de petites sections à ± 5K pour de
grandes sections) s’est révélée élevée par
rapport aux limites de température données (5K).
DETERMINING HIGH VOLTAGE CABLE CONDUCTOR TEMPERATURES.
Guy Van der Veken.
Euromold, Belgium.
PROBLEM DEFINITION
Temperature
Heating
General
The tests for checking high voltage products
(joints, terminations, connectors, bushings) are
described in standard HD628-S1. Some tests require an elevated temperature. For these tests, we
must guarantee that the conductor temperature
remains within strict limits (5K to 10K above the
maximum conductor temperature).
To achieve this in fluctuating ambient conditions,
we must regulate the current. This means that we
must measure the following values simultaneously:
conductor temperature and current. However, this
is not possible because the voltage in the cable is
too high. This makes it impossible to measure the
conductor temperature directly. We must determine the conductor temperature (which must be
kept within strict limits) in another way. The answer
can be found in Appendix A of this standard. We
will carry out a no-voltage pre-test (calibration) i.e.:
st
A 1 method enables us to obtain the relationship
between the conductor temperature, the ambient
temperature and the current.
nd
With a 2 method, we obtain the relationship between the conductor temperature, the cable-jacket
temperature and the current.
Specific test.
The heat-cycle test is one of the tests in which we
use an increased conductor temperature. This test
consists of 128 cycles. Each cycle (see figure 1)
lasts for 8 hours and consists of the following
steps:
* Heat the cable so that the conductor temperature
is within the increased temperature zone for at
least 2 hours.
* Leave the cable to cool down naturally for at least
3 hours, until the temperature difference between
the conductor and the environment is maximum
10K.
Cooling
Increased
temperature
10K
Environment-
Time
2 hrs
3 hrs
8 hrs
Figure 1: Representation of the heat-cycle test.
This test shows that it is also very important to
know the relationship between the conductor temperature and the current. If the current estimate is
wrong, there is a risk that a whole series of cycles
of the heat-cycle test must be repeated until we
have 128 good cycles.
RESEARCH.
In order to make a choice between the aforementioned methods, we investigated the following elements:
1. Checking the existing methods.
2. Which method gives the most accurate conductor temperature.
3. Is the 5K range feasible? In order to check this,
we carried out an uncertainty study according to
standard XP X07-020 of 1996.
The tests.
The test configuration. To carry out the novoltage pre-test (calibration of the cable), the test
configuration in figure 2 was used.
At each measurement point, 1 thermocouple (calibrated) of each thermocouple group (see next
paragraph) was attached. This enables us to establish which thermocouple group gives the most
accurate temperature reading and whether there
are major temperature differences between the
different thermocouple groups.
MP1
current
source
70cm
make the conductor visible) is closed and the
jacket is put back in its original position.
MP2
90cm
MP4
Figure 2: Test configuration.
70cm
Jacket
Insulation
Thermocouple
MP3
Conductor
The thermocouple groups. Thermocouple group
1 (= Cenelec method): J-type, point soldered (figure 3). Thermocouple group 2 (= point method): Jtype, unsoldered (figure 4). Thermocouple group 3:
(window mehtod): J-type, twisted, soldered (figure
5). Thermocouple group 4: strip thermocouple (this
is a thermocouple that is attached onto a copper
layer, making it possible to measure the jacket
temperature with these thermocouples).
junction
Figure 3: Cenelec method.
conductor
Figure 4: Point method.
junction
Figure 7: Presentation of the thermocouple positioning with the window method.
Calibration. During calibration, the following
measurement values are registered every 5 minutes: current, jacket-, environment- and conductortemperature. The thermocouples used are described in paragraph “The thermocouple groups”.
The test configuration used is given in figure 2 and
the positioning of the thermocouples is described
in the previous paragraph. Calibration was carried
out for Al cables with a 50mm², 240mm² and
630mm² section. With this choice, we cover a wide
range of high voltage cables regarding the cable
section. In addition, it gives us a good idea of the
better method for determining the conductor temperature and of the uncertainty about the conductor temperature we can expect when testing with
an increased temperature.
Figure 5: Window method.
The uncertainty.
Positioning of the thermocouples. The thermocouples in groups 1 and 2 are connected to the
conductor through a small hole, drilled in the cable
(see figure 6). This figure also shows how the
thermocouple must be positioned, i.e. where the
thermocouple wire comes out of the cable, it will be
bent. When taping the thermocouple, this bend will
provide a pressure point.
Thermocouple
Jacket
Insulation
Conductor
Figure 6: Presentation of the thermocouple positioning with the drilled method.
The thermocouples in group 3 are in contact with
the conductor by inserting them between the different conductor wires (see figure 7). When the
thermocouple is put in place, the window (rectangular cut-out in the cable-jacket and insulation to
To determine the uncertainty of the conductor
temperature, we used the French standard XP
X07-020 of 1996. This standard is based on establishing the variances on the variables needed for
determining the conductor temperature, i.e. variance on current, ambient temperature and jacket
temperature. A factor we must certainly take into
account is the variance on the model. This is the
difference between the conductor temperature
measured during calibration and the calculated
conductor temperature on the basis of the measured current and the measured jacket temperature
(or ambient temperature according to the method
used). Once these variances are known, we can
establish the variance on the conductor temperature. Then, the uncertainty is indicated by:
1/2
∆ϑconductor = k * (V[ϑconductor]) . (with ∆ϑconductor : uncertainty on the conductor; k: widening factor;
V[ϑconductor]: variance on the conductor temperature.)
RESULTS.
Cenelec ↔ Points ↔ Window.
The results are discussed on the basis of figures.
They are the result of the calculations based on
the values measured during the tests. However, it
is impossible to explain for each figure where all
the values come from.
In this paragraph, we will only consider the results
obtained with method 2 (ϑConductor based on ϑJacket).
The reason for this is given in the previous paragraph.
124°C
122°C
Method 1 ↔ method 2.
ϑCond and ∆ϑCond
120°C
118°C
116°C
When looking at figure 8, we can clearly see that
the variance on the model with method 1 (ϑConductor
based on ϑEnvironment) is always bigger than with
method 2 (ϑConductor based on ϑJacket).
114°C
112°C
110°C
108°C
106°C
104°C
1.6
1.4
Cenelec Points
Window Cenelec Points
Cable 50mm²
Temperature of conductor
Window Cenelec Points
Cable 240mm²
Uncertainty
Window
Cable 630mm²
Variance on model (°C²)
1.2
Figure 10: Representation of the average conductor temperatures and the uncertainties for the different cables and thermocouple positioning.
1.0
0.8
0.6
0.4
0.2
0.0
50²
Method 1
240²
630²
50²
CENELEC
Method 2
240²
630²
50²
POINTS
240²
630²
WINDOW
Figure 8: The variances on the model.
This greater variance on the model has a direct
impact on the total uncertainty ∆ϑConductor (see figure 9). Here too, we see that method 1 always
gives greater values.
6
∆ϑConductor (°C)
5
When looking at this graph, we notice the following:
* The conductor temperature is highest when the
thermocouples are positioned according to the
point method (for all three cables).
* The point and window methods are two equivalent methods: they have nearly the same conductor temperature and uncertainty.
* The conductor temperatures calculated on the
basis of the Cenelec method are 6 to 10°C lower
than the conductor temperature calculated on the
basis of the point and window methods.
* The uncertainty on the conductor temperature is
approximately identical with the 3 methods
(Cenelec, points, window).
So we can conclude that the positioning of the
thermocouples is better with the point or window
method.
4
3
2
Impact of the partial factors
1
0
50²
240²
630²
CENELEC
Method 1
50²
240²
630²
POINTS
50²
240²
630²
WINDOW
Method 2
Figure 9: The uncertainties in graph.
We can conclude from figures 8 and 9 that method
2 (ϑconductor based on ϑJacket) should be preferred for
determining ϑConductor as this method gives us the
smallest variance on the model and the smallest
uncertainty.
From paragraph "Method 1 ↔ method 2" we know
that method 2 is the most appropriate to determine
ϑConductor. Paragraph "Cenelec ↔ points ↔ window" gives us the positioning of the thermocouples
(points or window). When discussing the results,
we will only consider these methods.
With method 2, V[ϑConductor]Jacket (= variance on the
conductor temperature, whereby the conductor
temperature is established on the basis of the
jacket temperature) is determined as follows:
2
 ∂ϑ 
 ∂ϑ 
V [ϑ ] = 
 .V [ϑ ] + 
 .V [I ]
 ∂I 
 ∂ϑ 
 ∂ϑ  . ∂ϑ  .
+ 2. 
 
 V [ϑ ] . V [I ] + V [ function ]
 ∂ϑ   ∂I 
2
Cond
50mm² - Window
Cond
Cond Jacket
Jacket
V[ ϑCond] = 2.71 °C²
Jacket
Cond
∆ϑCond,Jacket = +/- 3.3 °C
Factor 4
9%
Cond
Jacket
Jacket
Whereby:
(
∂ ϑ Cond
∂ ϑ Jacket
)
(
∂ ϑ Cond
∂I
)
(
)(
2.
∂ϑCond
∂ϑJacket
2
2
.V [ϑ Jacket
]
Factor 3
26%
= factor 1
Factor 1
62%
.V [I ] = factor 2
.
∂ϑCond
∂I
)
. V [ϑJacket]. V [I ] =factor 3
Factor 2
3%
Factor 1
V [ function] = factor 4
Factors 1, 2, 3, 4 are represented in fig. 11.a → f.
Factor 2
Factor 3
Factor 4
Figure 11b: Percentage of the partial factors for the
50mm²-Al cable (window method).
240mm² - Points
50mm² - Points
V[ϑCond] = 2.9 °C²
V[ϑCond] = 3.26 °C²
∆ϑCond,Jacket = +/- 3.41 °C
∆ϑCond,Jacket = +/- 3.62 °C
Factor 4
11%
Factor 4
35%
Factor 1
44%
Factor 3
26%
Factor 1
60%
Factor 2
3%
Factor 1
Factor 3
19%
Factor 2
Factor 3
Factor 4
Figure 11a: Percentage of the partial factors for the
50mm²-Al cable (point method).
Factor 1
Factor 2
Factor 2
2%
Factor 3
Factor 4
Figure 11c: Percentage of the partial factors for the
240mm²-Al cable (point method).
630mm² - Window
240mm² - Window
V[ϑCond] = 3.13 °C²
V[ϑCond] = 5.46 °C²
∆ϑCond,Jacket = +/- 3.54 °C
∆ϑCond,Jacket = +/- 4.68 °C
Factor 4
29%
Factor 1
29%
Factor 1
48%
Factor 4
45%
Factor 2
4%
Factor 3
21%
Factor 1
Factor 2
2%
Factor 2 Factor 3
Factor 4
Figure 11d: Percentage of the partial factors for the
240mm²-Al cable (window method).
630mm² - Points
V[ϑCond] = 5.52 °C²
∆ϑCond,Jacket = +/- 4.7 °C
Factor 1
29%
Factor 4
45%
Factor 2
4%
Factor 3
22%
Factor 1
Factor 2
Factor 3
Factor 4
Figure 11e: Percentage of the partial factors for the
630mm²-Al cable (point method).
Factor 1
Factor 2
Factor 3
22%
Factor 3 Factor 4
Figure 11f: Percentage of the partial factors for the
630mm²-Al cable (window method).
We can draw the following conclusions from figure
11:
* Factor 2 is very small for all cables and positioning methods of the thermocouples.
* The factors that contribute most to the uncertainty ∆ϑConductor are:
* For cables with a small section: factors 1 and
3.
* For cables with a medium section: factors 1
and 4.
* For cables with a large section: factors 1 and
4.
This means we can make the uncertainty ∆ϑCond
smaller by reducing factors 1, 3 and 4. We can do
this as follows:
* Use class 1 thermocouples instead of class 2.
(This has an impact on factors 1 and 3.)
* Adapting the current transformer class (this reduces factor 3):
* the transformer ratio ≤ 400 ⇒ class 0,5.
* the transformation ratio > 400 ⇒ class =
200/transformation ratio).
* When calibrating the cable, we will:
* for a cable section < 240mm² => set 6 current
values instead of 5.
* for a cable section ≥ 240mm² => set 7 current
values instead of 5.
(This will reduce V[function].)