Upload
azrul
View
216
Download
4
Embed Size (px)
Citation preview
Development of Treeing Detection Method for Cable
with Three Insulation Regions Using TDR Technique
Tze Mei Kuan* and Azrul Mohd Ariffin Universiti Tenaga Nasional
Jalan IKRAM-UNITEN, 43000 Kajang, Selangor
Abstract- Treeing is a typical factor that contributes to aging of
polymeric insulated cables. The occurrence and proliferation of
these water trees cause electricity disruptions which interfere
with the proper operation of the lines. This has resulted in a
strong need for techniques to identify treeing degradation in
polymeric insulated cables. In previous studies, a method that
uses time domain reflectometry (TDR) technique to detect and
locate water tree within two insulation regions has been shown to
be feasible. Thus, this paper attempts to investigate on the
potential of using the same method on cable with three insulation
regions. Cable is modelled using MATLAB simulation tool to
obtain the cable impedance for PSpice set-up simulation where
the TDR technique is applied. The pulse reflections from the
PSpice simulations are recorded and analysed to investigate
whether this technique is capable of detecting water tree within
three insulation regions.
Keywords- Insulation; reflection; trees
I. INTRODUCTION
Water treeing is one of the main forms of polymeric
insulation degradation in medium voltage cables. Water trees
gradually reduce the electrical breakdown strength of cables
and eventually lead to cable failures that cause electricity
disruptions which interfere with the proper operation of the
lines. This has resulted in a strong need for techniques to
identify treeing degradation in polymeric insulated cables so
that appropriate service time can be scheduled to prioritize
cable replacement. However, such degradation is difficult to
detect in the cables using non-destructive techniques as partial
discharges are not involved in the growth of water tree.
Nevertheless, studies in [1] have shown that insulation
containing water trees has lower insulation resistance and
higher dielectric losses. Thus, the measurement of these
properties can be used to assess the condition of the insulation.
Diagnostic tests have been identified which can detect and
evaluate the degradation phenomena that causes failure in
cables. These tests include destructive and non-destructive
techniques which can be carried out on-site or in the
laboratory. However, most of the techniques could only give
average information on the status of the whole cable and
cannot detect which part along the cable that is degraded the
most. In previous studies [2,3], a method that uses time
domain reflectometry (TDR) technique to detect and locate
water tree within two insulation regions has been shown to be
feasible. Thus, this paper attempts to investigate on the
potential of using the same method to detect and locate water
tree within three cable insulation regions. Cable is modelled
using MATLAB simulation tool to obtain the cable impedance
for PSpice set-up simulation where the TDR technique is
applied. Fig. 1 shows the TDR system circuit model that has
been proposed. The basic principal on how this TDR
technique works has been discussed in [3]. Using the time
delay equation in [3], water tree along the cable can be located
from the PSpice results.
Fig. 1. Cable Model for Un-degraded Condition [4]
II. METHODOLOGY
The cable type and size used in this study is cross linked
polyethylene (XLPE), 9.0mm, 630mm2
(33kV) with a length
of 30m. Before performing simulations using PSpice, the
XLPE cable has to be modelled by the MATLAB simulation
tool to obtain the cable characteristic impedance. In order to
investigate for the potential of the TDR technique to detect
and locate water tree within the cable insulation, the cable has
to be modelled in two conditions where the first is assuming
the cable with a clean or un-degraded insulation and the
second is assuming a portion in the cable insulation has been
degraded due to the growth of water tree.
In [4], the cable model for the un-degraded condition as
illustrated in Fig. 2 has been proposed from the application of
circuit and transmission line theories.
Fig. 2. Cable Model for Un-degraded Condition [4]
978-1-4673-5074-7/13/$31.00 ©2013 IEEE
2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June2013
172
Another cable model is proposed in [2] by assuming part of
the cable insulation has been degraded due to the growth of
water tree in the XLPE cable insulation. Fig. 3(a) shows part
of the XLPE cable insulation has been degraded due to water
tree while Fig. 3(b) shows the model for this cable insulation.
Fig. 3. (a) XLPE Insulation Partially Degraded due to Water Tree
(b) XLPE Insulation Model with Water Tree
Fig. 4 below shows the full cable model incorporating the
water tree from Fig. 3(b) into Fig. 2.
Fig. 4. Cable Model for Degraded Condition
From the MATLAB simulations on these two cable
conditions, the characteristic impedance of the cable can be
obtained using equation (1) in [2] for PSpice simulation. As
mentioned earlier, studies in [2,3] have investigated on the
potential use of TDR technique to detect and locate water tree
within XLPE insulated cable that has two insulation regions as
shown in Fig. 5.
Fig. 5. Cross Section of Partially Degraded XLPE Cable
(2 Insulation Regions)
As this paper attempts to test the capability of this method in
detecting and locating water tree within the cable insulation, a
third region is introduced to the cable in Fig. 5 which is
illustrated in Fig. 6 below.
Fig. 6. Cross Section of Partially Degraded XLPE Cable
(3 Insulation Regions)
Thickness of
XLPE insulation XLPE insulation
still in good
condition
Part of XLPE insulation
that has been infected by
water tree
Length of Ttest
(Region 1)
Length of Ttest2
(Region 2)
Conductor
screen
Thickness of
XLPE insulation XLPE insulation
still in good
condition
Part of XLPE insulation that has
been infected by
water tree
Length of Ttest
(Region 1)
Length of
Ttest3
(Region 3)
Conductor
screen
Length of
Ttest2
(Region 2)
XLPE
insulation still in good
condition
(a)
(b)
2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June2013
173
Studies in [2,3] performed the PSpice simulations using
three different length combinations for Ttest and Ttest2 from
Fig. 5. The cable lengths are 10/20m, 15/15m and 20/10m
where the first figure represents the length for Ttest while the
second length is referring to Ttest2. Since the analysis of those
results show that water tree can still be detected and located
even though the water tree grows at different length along the
cable insulation, thus the lengths used in this paper in the first
attempt on three insulation regions cable in Fig. 6 are taken at
10m for Ttest, 5m for Ttest2 and 15m for Ttest3.
Although the study in [2] has carried out the simulations on
three levels of degradation, this paper only concentrates on
one which is 25% of the XLPE insulation thickness is
assumed to be degraded by water tree. Therefore, all results
are analyzed based on this specific level of degradation. The
pulse reflected from the PSpice simulations due to impedance
difference as the signal travels along the set-up are recorded
and analysed to investigate whether this technique is capable
of detecting water tree within three insulation regions.
III. RESULTS AND DISCUSSIONS
The study in [3] shows the simulation of PSpice set-up
illustrated in Fig. 7 below for testing 30m long cable with un-
degraded insulation outputs the result as shown in Fig. 8. As
discussed in [3], there are four pulses detected but only the
first three pulses are valid which is due to impedance
difference of two cables in the set-up. The fourth pulse is
observed to be multiple reflections due to Ropen. The signal
routes as it travels through the set-up in Fig. 7 can be seen in
Fig. 9.
Fig. 7. PSpice Set-up for Testing Un-degraded Cable
Fig. 8. Reflected Pulses for Un-degraded Cable
Fig. 9. Signal Routes for Results in Fig. 8
In the same study, the simulation is repeated on partially
degraded cable which now has two insulation regions using
the set-up in Fig. 10 and the result is shown in Fig. 11. The
cable is now represented by two components; Ttest and Ttest2
with 15m long each. Since Ttest represents the un-degraded
portion, the pulse A1 is observed to occur at an earlier time
between P2 and P3 with a negative magnitude. Pulse A1
occurs at an earlier time because the length of Ttest is now
15m and therefore the shorter time is required for the signal to
be reflected. Since the impedance of Ttest2 which is degraded
has smaller impedance than Ttest, the subtraction term in the
reflection coefficient, ρ from equation (1) causes the pulse A1
to have a negative magnitude.
� � ��2 � �1�1 � �2�1�
where
Z1 = Impedance of current component that the signal
travels through
Z2 = Impedance of the next component that the signal
detects
As Ropen has much higher impedance compared to Ttest2
which is degraded due to water tree, the subtraction term of
the reflection coefficient, ρ gives a positive value. This
explains the positive magnitude of pulse P3. Since the
impedance of a degraded cable due to water tree is usually
lower than other cables, this makes the subtraction term in the
reflection coefficient to be in positive value, therefore, it can
be assumed that an additional pulse with positive magnitude
detected between P2 and P3 shows a possibility of water tree
degraded cable. Then, using the time delay equation, water
tree can be located as discussed in [3]. Similarly to the un-
degraded cable simulation, pulses A2, P4 and A3 are detected
due to multiple reflections at Ropen which can be neglected
when the time delay equation is used to locate the water tree.
The signal routes for the first four pulses detected in Fig. 11
can be seen in Fig. 12.
P1 = 16.955ns P3 = 417.940ns
P2 = 117.836ns P4 = 717.015ns
Time (µs)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
9.0mm, 630mm2 (33kV) (30m)
-1.0V
0V
1.0V
2.0V
3.0V
P1
P2
P3
P4
P1
P2
P3
2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June2013
174
Fig. 10. PSpice Set-up for Testing Degraded Cable (2 Insulation Regions)
Fig. 11. Reflected Pulses for Degraded Cable (2 Insulation Regions)
Fig. 12. Signal Routes for Results in Fig. 11
It is known that the same set-up is used for testing the un-
degraded and degraded cable (2 insulation regions) besides the
additional component introduced to the degraded set-up to
incorporate the water tree to the cable. Thus, comparing the
results in Fig. 11 to Fig. 8, it can be observed that the
additional pulse, A1 detected in Fig. 11 indicates the existence
of water tree region in the cable insulation which is recorded
by the pulse P3. This shows that this system is capable of
detecting water tree along the cable due to impedance
difference between the un-degraded insulation and the
degraded insulation caused by the growth of water tree. Then,
using the time delay equation, water tree can be located as
discussed in [3].
As the result in Fig. 11 shows the capability of this method
to detect and locate water tree for the two insulation regions
cable, the study in this paper extends the simulation to test on
three insulation regions cable as illustrated earlier in Fig. 6.
The PSpice set-up for this simulation is shown in Fig. 13.
The simulation of set-up in Fig. 13 gives the result in Fig.
14. Observing the result, it can be noticed that there are now
two additional pulses recorded between P2 and P3, named B1
and B2. Similarly to results in Fig. 8 and Fig. 11, the pulses
detected after the time where P3 is detected are all classified
as repeated pulses due to multiple reflections from Ropen
which are neglected. The signal routes for the first five pulses
recorded in Fig. 14 are shown in Fig. 15.
Fig. 13. PSpice Set-up for Testing Degraded Cable (3 Insulation Regions)
Fig. 14. Reflected Pulses for Degraded Cable (3 Insulation Regions)
Fig. 15. Signal Routes for Results in Fig. 14
P1
P2
A1
P3
Time (µs)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
9.0mm, 630mm2 (33kV)
(10/5/15m)
0V
-1.0V
1.0V
2.0V
3.0V
P1 = 16.955ns P3 = 416.667ns B6 = 616.667ns B8 = 816.667ns
P2 = 117.836ns B3 = 466.667ns P4 = 716.667ns B9 = 866.667ns
B1 = 216.667ns B4 = 516.667ns B7 = 766.667ns B10 = 916.667ns
B2 = 266.667ns B5 = 566.667ns
P1
P2
P3
P4
B1 B2 B3 B4 B5 B6 B7 B8 B9 B10
P1
P2
B1
B2
P3
Time (µs)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
9.0mm, 630mm2 (33kV) (15/15m)
-1.0V
0V
1.0V
2.0V
3.0V P1
P2 A1
P3
A2
P4
A3
P1 = 16.955ns A2 = 567.090ns
P2 = 117.836ns P4 = 717.015ns
A1= 267.240ns A3 = 867.097ns
P3 = 417.940ns
1.0
2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June2013
175
Referring to the time delay table from Table III in [2], the
first pulse, P1 is detected by the oscilloscope after the incident
wave travels through coaxial cable T1 and T3 as shown in Fig.
15 which is the shortest path for the wave to reach the
oscilloscope. The pulse reaches node 8 at time 16.955ns as
indicated in Fig 14. The time occurrence for P1 can be
obtained by adding the theoretical delay time or time of
propagation down the line of length T1 and T3 which is
described as:-
Total delay time = TD in T1 + TD in T3
= 10.11ns + 7.08ns
= 17.19ns
The calculated value is close to the measured value. The fall
time for the pulse is at 56.955ns, since the pulse width, PW is
set to 40ns.
From Fig. 14, the second pulse, P2 occurs at 117.836ns,
which is obtained by adding the delay time or time of
propagation down the line of length T1, 2 × T2 and T3. The
fall time for the pulse is at 157.836ns, since the pulse width,
PW is set to 40ns. P2 in Fig. 14 has a negative value which is
again due to the reflection coefficient, ρ since the impedance
of Ttest is smaller than the impedance of T2, then the
subtraction between these two impedances will give a negative
value.
Since the water tree is assumed to grow within the cable
length, the test cable which is now represented by three Test
cables causes more reflections to the oscilloscope. The
additional pulse, B1 reached the oscilloscope at time
216.667ns with negative magnitude. This negative magnitude
is also due to the reflection coefficient, ρ since the impedance
of Ttest2 is smaller than the impedance of Ttest. Therefore,
the subtraction between these two impedances will give a
negative value. The time is obtained by adding the delay time
or time of propagation down the line of length T1, 2 × T2, 2 ×
Ttest and T3. The fall time for the pulse is at 256.667ns, since
the pulse width, PW is set to 40ns.
From the simulation of the earlier PSpice set-up of degraded
cable with the two insulation regions, an assumption has been
made where if there is any additional pulse with positive
magnitude detected between P2 and P3, it shows a possibility
of water tree degraded cable. Referring to Fig. 14, the second
additional pulse, B2 reached the oscilloscope at 266.667ns
with a positive magnitude. This pulse, B2 indicates the
existence of water tree degradation in the test cable. This pulse
can also be obtained by adding the delay time or time of
propagation down the line of length T1, 2 × T2, 2 × Ttest, 2 ×
Ttest2 and T3. The fall time for the pulse is at 306.667ns,
since the pulse width, PW is set to 40ns.
Although the signal is reflected due to impedance mismatch
between Ttest2 and Ttest3, only small part of the signal is
reflected. This can be seen by evaluating the magnitude of the
pulse voltage. Due to high frequency signal, the wave still
travels through Ttests3 and meets the open end. This causes
another reflection and therefore pulse, P3 is detected by the
oscilloscope at time 417.940ns which is equivalent to the
pulse P3 recorded in Fig. 14. The time is obtained by adding
the delay time or time of propagation down the line of length
T1, 2 × T2, 2 × Ttest, 2 × Ttest2, 2 × Ttest3 and T3. The fall
time for the pulse is at 457.940ns, since the pulse width, PW is
set to 40ns.
From the discussions on the PSpice simulation result in Fig.
14, it shows that this method that implements the TDR
technique is capable of detecting the existence of water tree in
the cable insulation. To further investigate on the potential of
this technique to locate the water tree along the 30m long
cable, the time delay equation from study in [3] is used. The
time delays for all coaxial cables in the set-up from Fig. 13 are
again obtained from time delay table in [2]. In earlier
discussion, it has been explained that pulse B1 is resulted from
the signal travelling through coaxial cables T1, T2, Ttest and
reflected due to impedance mismatch between Ttest and
Ttest2, and back to Ttest, T2 and finally T3 (illustrated in Fig.
15). Therefore, time delay of the Ttest can be found by taking
the total time occurrence of B1 minus the total time delay of
the other coaxial cables. Hence, the length of the Ttest can be
calculated from B1 as follows:-
TD for (2 × Ttest) = Time occurrence for B1 from
Fig. 14 – [TD in T1 +
(2 × TD in T2) + TD in T3]
= 216.667ns – [10.11ns +
(2 × 50.55ns) + 7.08ns]
= 98.377ns
∴ TDforTtest � 98.377��2
= 49.1885��
Substituting the TD for Ttest into time delay equation:-
ℓ !"#! =$% !"#! × '(
√*+,+=
$% !"#!5.0553��
= 49.1885��
5.0553��
= 9.73m ≈ 10m
The calculated length of 9.73m is quite closed to the actual
length of 10m. Since the magnitude of B1 is negative due to
the reflection coefficient, therefore, it can be concluded that
the first 10m of this test cable is still in good condition as the
impedance of Ttest2 is lower than Ttest (indicating Ttest2 has
deteriorated).
B2 from Fig. 14 is resulted from the signal travelling
through coaxial cables T1, T2, Ttest, Ttest2 and reflected due
to Ttest3 and back to Ttest2, Ttest, T2 and finally T3
(illustrated in Fig. 15). Therefore, time delay of the Ttest2 can
be found by taking the total time occurrence of B2 minus the
total time delay of Ttest and the other coaxial cables. Take
note that the time delay for Ttest used in this calculation is
2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June2013
176
from the calculated value not from the time delay table.
Hence, the length of the Ttest2 can be calculated from B2 as
follows:-
TD for (2 × Ttest2) = Time occurrence for B2 from
Fig. 14 – [TD in T1 +
(2 × TD in T2) + (2 × TD in Ttest) +
TD in T3]
= 266.667ns – [10.11ns + (2 × 50.55ns) +
(2 × 50.55ns) + 7.08ns]
= 47.277ns
∴ TDforTtest2 = 47.277��
2= 23.6385��
Substituting the TD for Ttest2 into time delay equation:-
ℓ !"#!/ =$% !"#!/ × '(
√*+,+=
$% !"#!/5.0553��
= 23.6385��
5.0553��
= 4.676m ≈ 5m
From the calculation, it was found that the length of the
Ttest2 is 4.676m which is quite close to 5m. Since this pulse
gives a positive magnitude, it can be concluded that the test
cable is degraded about 5m long. This is because of the
positive magnitude from the reflection coefficient which
indicates that the Ttest3 has higher impedance than Ttest2.
Now, it is known that 5m of the total cable length is degraded
due to water tree and is located from the 10th meter until the
15th
meter of the cable.
P3 in Fig. 14 is resulted from the signal travelling through
coaxial cables T1, T2, Ttest, Ttest2, Ttest3 and reflected back
to Ttest3, Ttest2, Ttest, T2 and finally T3 due to the open end.
Therefore, time delay of the Ttest3 can be found by taking the
total time occurrence of P3 minus the total time delay of Ttest,
Ttest2 and the other coaxial cables. Take note again that the
time delay for Ttest used in this calculation is from the
calculated value not from time delay table. Hence, the length
of the Ttest3 can be calculated from P3 as follows:-
TD for (2 × Ttest3) = Time occurrence for P3 from
Figure 14 – [TD in T1 +
(2 × TD in T2) + (2 × TD in Ttest) +
(2 × TD in Ttest2) + TD in T3]
= 417.940ns – [10.11ns + (2 × 50.55ns) +
(2 × 50.55ns) + (2 × 25.28ns) + 7.08ns]
= 147.99ns
∴ TDforTtest3 = 147.99��
2= 73.995��
Substituting the TD for Ttest3 into time delay equation:-
ℓ !"#!/ =$% !"#!0 × '(
√*+,+=
$% !"#!05.0553��
= 73.995��
5.0553��
= 14.637m ≈ 15m
From the calculation, it was found that the length of the
Ttest3 is 14.637m which is close to 15m. Although this pulse
gives a positive magnitude, it is not degraded because Ttest3
has higher impedance than Ttest2. The pulse has a positive
value because the impedance of the open end is very much
larger than the impedance of Ttest3. Thus, it is now known
that another 15m of the total cable length is un-degraded
which is located from the 15th meter until the 30
th meter of the
cable.
The newly found cable conditions with their respective
lengths are quite closed to the cable conditions and lengths
used earlier in the set-up for simulation. This again shows that
this technique from analysing the pulse to calculating the
lengths of both un-degraded and degraded cable conditions
and finally locating the water tree along the cable length is
able to produce quite accurate results. Thus, this technique is
applicable even on cable degraded with three insulation
regions.
IV. CONCLUSION
This study investigated the potential of using time domain reflectometry (TDR) technique to detect and locate water tree in cable degraded with three insulation regions as previous study have shown that this technique is capable of detecting and locating the water tree in cable degraded with two insulation regions. From the analysis of the result, it was found that this technique is still capable of detecting the existence of water tree even with three insulation regions. The additional pulse detected with positive magnitude between pulses P2 and P3 are crucial in determining the existence of the water tree. Upon detecting the existence of water tree, its location can be found through calculations using the time delay equation. The potential use of this technique can be further investigated to test on cable with higher number of insulation regions; more than one region is degraded due to water treeing.
REFERENCES
[1] S. D. Grigorescu, M. Plopeanu, P. V. Notingher and C. Stancu,
“Equipment for Fast Water Trees Resistance Measurement of Power
Cable Insulations,” 8th International Conference on Insulated Power
Cables, E.5.2.14, June 2011.
[2] A. Mohd. Ariffin, T. M. Kuan, S. Sulaiman and H. A. Illias,
“Application of Time Domain Reflectometry Technique in Detecting
Water Tree Degradation within Polymeric-Insulated Cable,” 2012
International Conference on Condition Monitoring and Diagnosis, pp.
1163-1166, 23-27 September 2012, Bali, Indonesia.
[3] Tze Mei Kuan, Azrul Mohd Ariffin and Suhaila Sulaiman, “Signal Analysis to Detect Water Tree Location in Polymeric Underground Cables,” The 9th IEEE Student Conference on Research and Development, pp. 359-365, 19-20 December 2011, Cyberjaya, Malaysia.
[4] T. M. Kuan, A. Mohd. Ariffin, S. Sulaiman and Y. H. Md Thayoob,
“Wave Propagation Characteristics of Polymeric Underground Cables,”
The 5th International Power Engineering and Optimization Conference,
pp. 410-415, 6-7 June 2011, Shah Alam, Selangor, Malaysia.
2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June2013
177