The 20th International Symposium on High Voltage Engineering, Buenos Aires, Argentina, August 27 – September 01, 2017

DIELECTRIC LOSSES OF MEDIUM VOLTAGE CABLE ACCESSORIES

AND THEIR INFLUENCE ON ON-SITE DISSIPATION-FACTOR MEAS-
UREMENTS ON XLPE-INSULATED MEDIUM VOLTAGE CABLE LINES

Georg Frübing1*, Ronald Plath1, Wilfried Kalkner1, Thomas Kumm2

and Vladimir Stamenkovic1
1Technische Universität Berlin, Berlin, Germany
2EWE NETZ GmbH, Oldenburg, Germany

*Email: [email protected]

Abstract: The paper presents a strategy to approach the in general unknown influence of
medium-voltage cable accessories on dissipation-factor measurements at 0.1 Hz on XLPE-
insulated medium-voltage cable lines. The strategy consists of three steps, which are pre-
sented exemplary for a large distribution network in Germany with about 60,000 km of ca-
bles produced since the middle of the 80s according to current quality standards. The first
step is the analysis of the frequency of occurrence of accessory types in the network, from
which the dominant accessory types can be identified. The second step contains the ex-
traction of samples of these accessories from the network and their investigation in the
laboratory to derive the dielectric losses at 0.1 Hz. In the final step it is possible to correct
on-site tanδ-measurements utilizing the dielectric loss data of the accessory samples.

1 INTRODUCTION groups with equal name (selling/product name) and

producer. In this paper producers are represented
XLPE-insulated medium voltage cables manufac- with capital letters and names are represented with
tured in Germany according to current quality stand- small letters.
ards, e.g. IEC 60502-2 or CENELEC HD 620, seem
to show only minor aging even after up to 30 years Figure 1 shows the installed number of accessories
in service. However, for network operators the in the network for all categories. The categories are
knowledge of the cable condition and its possible ranked according to the number along the abscissa.
decrease with time is important for a reliable plan- More than 100 categories of joints and more than
ning of investments in the network infrastructure. 200 categories of terminations/connectors were ex-
Dissipation-factor measurements at 0.1 Hz of cable tracted from the database, however, most of them
lines can help to determine the cable condition with a comparatively small number. Many catego-
[1, 2], provided that clear diagnosis criteria are ries with only few occurrences were due to errone-
known. Thereby, ageing must be clearly distin- ous or uncomplete recording of names. It should be
guished from other effects or influences. mentioned that additionally about 2500 terminations
and connectors as well as about 5000 joints are
On-site measurement results differ in principle to marked as unknown in the database.
those obtained in the laboratory due to the exist-
ence of cable accessories (joints, terminations, con-
nectors). The influence of these cable accessories
is not fully known, especially not in case of XLPE-
insulated cables of new generation. In the following
a three-step procedure is described to improve the
quest for the unknown influence of accessories in
on-site dissipation factor measurements. These
three steps are described in sections 2, 3 and 4, re-
spectively.

2 ANALYSIS OF THE FREQUENCY OF OC-

CURRENCE OF DIFFERENT ACCESSO-
RIES
Figure 1: Number of occurrences for all categories
As first step data extracted from the asset’s data of terminations/connectors and joints installed from
base of the network operator were analysed for the 1981 to 2013
installation years from 1981 to 2013. 126,099 joints
and 127,276 terminations (mainly indoor types) and The most important finding from Figure 1 is that only
separable connectors were registered in the data a few accessories types account for the majority of
base. The accessories were then categorized in all accessories. If we exclude all categories of joints
with less than 6000 occurrences only five categories 3 LABORATORY INVESTIGATIONS OF AC-
remain. The percentage of all joints in each of these CESSORIES
five categories is shown in Figure 2. It can be seen
that almost every second joint in the network is of Since the relevant accessories are known, samples
type “B a”. The joint name “a” occurs two times in of these accessories were extracted from the net-
Figure 2: First with producer “A” and a second time work between 2013 and 2016 and analysed dielec-
with producer “B”. The reason for that can be found trically in the laboratory. Until now this was done for
in the history of the producers, since producer “B” four joints (two of them of the same type but different
bought producer “A” at a specific time. Together the installation year), one connector and one indoor ter-
four groups represent about 85% of all joints and mination. The accessories were extracted with
therefore in the majority of on-site measurements about 1 m of cable (in case of joints on both sides).
corrections of measured data will be necessary for In case of the termination and the connector guards
these types of joints. were installed 5-10 cm next to the end of the termi-
nation/connector to minimize the influence of the ca-
ble insulation in the following measurements. In
case of joints the guards were installed 5-10 cm
next to the ends of the outer shrinking tube of the
joint, which is normally about 80-100 cm long. It
should be mentioned that the dissipations-factor of
the whole specimen will also depend on the dissipa-
tion factor of the cable(s) on which the accessory is
installed. However, if the dissipation factor of the
short specimen is not dominated by the dissipation
factor of the accessory, this accessory will not be
able to significantly influence the dissipation factor
of a whole cable line. Therefore the influence of the
cables will stay unconsidered here.
Figure 2: Number of relevant types of joints in
percentage of the number of all joints

For terminations and connectors all categories with

less than 5000 occurrences were excluded, after-
wards only four categories remain. The percentage
of all terminations/connectors in each of these cate-
gories is shown in Figure 3. These four types in-
clude 70% of all terminations and therefore in the
majority of on-site measurements corrections of
measured data will be necessary for these types of
terminations and connectors.

Figure 4: tanδ of accessory specimen at 2 kV and

20 °C (term. = termination, conn. = connector)

After preparation of the specimen, measurements of

dissipation factor and capacitance where performed
in the laboratory at a comparatively small voltage of
2 kV and in the frequency range from 10 mHz to
100 Hz. The results for the dissipation factor are
Figure 3: Number of relevant types of terminations plotted in Figure 4. The installation year of the in-
and connectors in percentage of the number of all vestigated accessories is given in the legend.
terminations / connectors (“e” and “f” are indoor ter-
minations, “g” and “h” are connectors) The specimen “B a 1998” has the smallest dissipa-
tion factor, which can be found in a range which was
already observed for new cables and cables ex-
tracted from the same network [3], i.e. this joint can-
not be expected to have a significant influence in on-
site measurements. Joint specimen “B a 1998” and
“B a 1987” show a very different behaviour despite
the fact that both are the same joint type (same
name and producer). Aging or a change of material
or field grading by the producer might be a possible
explanation for this difference.

The joint “C b 1990” shows rather flat dielectric

losses in the observed frequency range. The joints
“A c 2010” and “B a 1987” as well as the connector
“B h 1998” and the indoor termination “B f 2003”
show rising losses with decreasing frequency, a be-
haviour often observed in previous investigations
[4–6]. It should be added that for the indoor termi-
nation “B f 2003” no surface-current detection was
applied, hence an influence of the surface current
on the dissipation factor cannot be excluded.

The main scope of this investigation are on-site

Figure 6: Capacitance of accessory specimen at
measurements at 0.1 Hz with voltages up to two
0.1 Hz and 20 °C
times the service voltage (12 kV), i.e. 24 kV. For this
reason, the dissipation factor and the capacitance
4 ESTIMATION OF THE INFLUENCE ON ON-
of all six specimen were measured in the laboratory
SITE MEASUREMENTS
at 0.1Hz at voltages from 1 to 24 kV. The results are
displayed in Figure 5 and Figure 6 respectively. It
Form the on-site measured capacitance 𝑪𝐥𝐢𝐧𝐞 and
can be observed that both quantities do not change
significantly with voltage. This is advantageous the dissipation factor 𝐭𝐚𝐧 𝜹𝐥𝐢𝐧𝐞 of a cable line and the
since the possible increase of the dissipation factor known capacitance 𝑪𝐚𝐜𝐜,𝒏 and dissipation factor
with voltage is an accepted criterion for condition as- 𝐭𝐚𝐧 𝜹𝐚𝐜𝐜,𝒏 of the 𝑵 accessories in this cable line, the
sessment of the cable insulation [2]. global dissipation factor 𝐭𝐚𝐧 𝛿cable of only the cables
in the cable line can be calculated using the follow-
ing formula [7]:

𝑪𝐥𝐢𝐧𝐞 ⋅ 𝐭𝐚𝐧 𝜹𝐥𝐢𝐧𝐞 − ∑𝑵

𝒏=𝟏 𝑪𝐚𝐜𝐜,𝒏 ⋅ 𝐭𝐚𝐧 𝜹𝐚𝐜𝐜,𝒏
𝐭𝐚𝐧 𝜹𝐜𝐚𝐛𝐥𝐞 =
𝑪𝐥𝐢𝐧𝐞 − ∑𝑵
𝒏=𝟏 𝑪𝐚𝐜𝐜,𝒏

In the following this correction will be applied to an

on-site measurement conducted in 2016 on a cable
line with a length of 283 m and a capacitance of
68 nF with one joint of type “B a” installed 1997 and
two connectors of type “B h” installed 1997 and
1998 at both ends of the cable line. Voltages of 4,
12, 18, 24 kV were applied. Utilizing the values of
tanδ and C in Figure 5 and Figure 6 the dissipation
factor was corrected using the above formula for
each voltage. The correction was calculated for
three cases: In the first case only the joint was con-
sidered with the values for “B a 1987”. In the second
case again only the joint was considered but with
Figure 5: tanδ of accessory specimen at 0.1 Hz and the values for “B a 1998”. In the third case only the
20 °C connectors at the two ends of the cable line were
considered with the values of “B h 1998”. The origi-
In Figure 6 the joint specimen have higher capaci- nally measured data and the results of the three cor-
tance values since the capacitance of the two about rections are presented in Figure 7.
0.5 m cable ends at both sides of the joint is added.
It will be described in the next section that the prod- For the correction with the values of the joint speci-
uct of tanδ and C is the relevant value for the calcu- men “B a 1987” and “B a 1998” the change in dissi-
lation of the influence on the measurement. pation factor is negligible. For the latter the dissipa-
tion factor is even increasing a bit through the cor-
rection, since the dissipation factor measured on-
site is higher than that of the specimen in the labor-
atory. However, for the correction with the values of
the connector specimen “B h 1998” a significant re- REFERENCES
duction is observed. The reduction is almost inde-
[1] CIGRE Working Group D1/B1.20, “Non-de-
pendent of voltage since the specimen exhibited an
structive water-tree detection in XLPE cable
almost constant dissipation factor and capacitance
insulation” Technical Brochure No. 493, 2012.
over voltage.
[2] IEEE Guide for Field Testing of Shielded
Power Cable Systems Using Very Low Fre-
quency (VLF) (less than 1 Hz), IEEE Std.
400.2™, 2013.
[3] G. Frübing, R. Plath, W. Kalkner, and T.
Kumm, “The Influence of Different Types of
Voltage Loading on Subsequent Dissipation-
Factor Measurements on XLPE-Insulated Me-
dium-Voltage Cables” in International Sympo-
sium on High Voltage Engineering, Pilsen
2015.
[4] F. Mauseth, K. D. Hammervoll, and S. Hvid-
sten, “Dielectric properties of service aged
medium voltage XLPE cable joints” in 10th
IEEE International Conference on Solid Die-
lectrics, 2010, pp. 1–4.
[5] N. Amyot, E. David, D. Fournier, D. Jean, and
D. Lalancette, “Dielectric response of cable
accessories and its influence on cable diag-
nostics” in Annual Report Conference on
Figure 7: Example for the correction of the on-site Electrical Insulation and Dielectric Phenom-
measured dissipation factor utilizing the laboratory ena, 2002, pp. 434–437.
measurements presented in Figure 5 and Figure 6 [6] A. Avellan, P. Werelius, and R. Eriksson,
“Frequency domain response of medium volt-
5 CONCLUSIONS age XLPE cable terminations and its influence
on cable diagnostics” in Conference Record
From the presented results it can be concluded that of the IEEE International Symposium on Elec-
our strategy works for the investigated network in trical Insulation, 2000, pp. 105–108.
principal. The analysis of the frequency of occur- [7] D. Eisemann, Henningsen C.-G., W. Geyer,
rences results in a small number of accessory types, W. Kalkner, and T. Kumm, “Untersuchungen
which include the majority of all accessories in the zum dielektrischen Verhalten von Mittelspan-
network. Some of these accessories were investi- nungskabelgarnituren” in ETG-Fachtagung,
gated in the laboratory. It was observed that the dis- Grenzflächen in elektrischen Isoliersystemen
sipation factor for most of them is significantly higher 2008.
compared to that of the cables, however, a notable
dependence on voltage was not found. A correction
of an on-site measurement was successfully exam-
ined utilizing the values obtained during the labora-
tory investigations. These calculation shows a neg-
ligible influence of the joints but a substantial contri-
bution of the connectors at the ends of the cable
line.

6 OUTLOOK

It is planned to extend the proposed analysis to all

accessories mentioned in section 2. Further on, the
correction will be applied to more than 130 on-site
dissipation-factor measurements carried out be-
tween 2014 and 2016 and the results will be evalu-
ated statistically.

ACKNOWLEDGMENTS
The authors would like to thank Enrico Klemm, Tino
Rücker and Weng Ying Tam for lab assistance.