Professional Documents
Culture Documents
I HC BCH KHOA H NI
B MN H THNG IN
PHN I
CAO P V VT LIU IN
TM TT
Vic tch cc phn t c tr s in dn khc nhau c ng dng rt rng ri
trong cc lnh vc cng nghip, nng nghip, x l cht thi in t Bi bo trnh
by m hnh thit b v kt qu nghin cu s dng cng ngh phn tch cc ht
bng k thut cao p tnh in. Cng ngh ng dng k thut cao p tnh in
c nghin cu nhiu nm ti cc nc trn th gii do c rt nhiu u im nh
tiu tn t nng lng, c hiu sut cao v khng gy nhim mi trng. Tuy nhin
ti Vit Nam, vic nghin cu su cng ngh v nh hng ca cc yu t khc
nhau nh in trng, mi trng ti hiu qu thit b cn cha c u t tha
ng.
ABSTRACT
Sa khong
in tr sut, .cm
Phn loi
Ilmenite
4,7
1-10-3
Dn in
Zircon
4,6-4,7
1013-1015
in mi
Rutile
4,2-5,2
1-102
Dn in
Thch anh
2,5-2,8
1012-1017
in mi
Pirit
4,9-5,2
10-5-10-1
Dn in
ny da trn mt s c s sau:
Trong thc t cc ht khong sn c
im so vi cc thit b ang c s
sau [2,3,4]:
nng lng.
rtd
0, 62 3 abc
(1)
Trong a, b v c tng ng l
chiu di, rng v cao ca ht.
Cc ht o c c ly ngu
nhin vi s lng mu l 400 ht.
T kt qu o c v tnh ton c th
nghing; 3 - in cc trn; 4 - in cc
v tr khc nhau.
Qu o chuyn ng ca cc ht
Fg
mug
4
.S.rtd3 .J. g.cos D
3
(2)
in trng. in trng tc ng ln cc
ht ny v lm thay i qu o chuyn
3
Fc
0,5.cx .J.v 2 .S
Thay i tr s in p t ln in cc tc
l thay i cng in trng, khi c
th thy rt r v ghi li s thay i qu
Vi cx - H s kh ng hc ca mi
trng, v - vn tc chuyn ng ca ht, S
- tit din ca ht.
o ca cc ht.
Phng trnh chuyn ng ca cc
ht sa khong theo nh lut 2 Newton
nh sau:
JG
F
(3)
Cc lc tc ng ln ht sa khong
bao gm nhng thnh phn sau:
G
dv
m.
dt
- Trng lc:
Fg
4
.S.rtd3 .J. g.cos D
3
mug
- Lc tc ng ca in trng:
Fe
E .q
u:
Fp
H 1
4.S.H0 .r
.E . gradE
H2
3. KT LUN:
- Lc cn ca mi trng:
cn cc khay gn pha di in cc c
mu sng hn (Zircon v thch anh). Cc
mu sn phm nhn c trong khay thu
hi qung c tin hnh nh gi
nhm hon thin thit b.
Cn c vo kt qu th nghim trn m
nghing ca in cc, v tr cc in
a ch lin lc:
S 1, i C Vit, H Ni.
O. Lesaint
T.V. Top
I.
INTRODUCTION
EXPERIMENTAL TECHNIQUES
a few pC, duration <1s), and very large ones at high voltage
with large rods (charge > 1C, duration > 100s). Detection
must be very sensitive in order to detect all streamers. During
propagation of a long streamer, a large number a fast current
pulses is detected. Conventional PD measurement systems
based on pulse detection are unable to record properly
streamers: counting of all current pulses leads to considerably
overestimate the streamer number actually generated. The
detection system must also have a very low level of spurious
noise, typically less than 1 shot per hour (in some experiments,
a very low number of streamers can appear, typically 1 per
hour). The system must be able to count properly such rare
events. In this study, inception was detected by the streamer
light emission using a photomultiplier (PM). This provided a
very sensitive detection. A dead time of 200 s was fixed
after each detection to avoid overcounting streamers. The
detection threshold was fixed above the background noise of
the PM, and the test cell was placed in a totally dark room.
Pollution inlet
HV
electrode
III.
PM
-1
10
stirrers
Oil
sampling
10
Drying and
filtration unit
Negative Streamer
-1
10
Positive Streamer
Figure 1.
Test cell
-3
10
100
Particle
Filtered
+ 0,7 mg/l
+ 18 mg/l
+ 70 mg/l
size (m)
oil
cellulose
cellulose
cellulose
2 -5
252
748
9690
16316
5-10
20
34
6060
806
10-15
16
56
76
15-25
18
18
25-50
50-100
100-150
>150
150
200
250
10
10
r =0.5mm
r =1mm
r =2.5mm
10
10
10
r =5mm
p
-1
r =10mm
10
-2
10
50
150
250
350
450
10
-1
10
10m
0.3mm
40m
10
1mm
10
2.5mm
2
10
10
8mm
1
10
-1
10
Filtered oil
2mg/l cellulose
13mg/l cellulose
25mg/l cellulose
10
-1
10
-2
10
100
200
300
400
500
-2
10
-1
10
10
10
calculated tip field (MV/cm)
10
10
voltage (kV) @ 1 streamer/minute
500
impulse voltage
400
II
10
17 ppm
300
35 ppm
200
-1
10
100
r =8mm
p
-2
g (
150
10
15
20
25
10
30
-3
10
)@
17 ppm
100
r =10m
p
0
0
b)
10
15
20
25
30
-1
10
10
10
V.
35 ppm
50
-2
10
CONCLUSIONS
[2]
[3]
[4]
[5]
11
a)
12
I.
INTRODUCTION
RUSK MODEL
e (r , z, t ) = (r , z, t )
A(r , z, t )
t
(1)
dz
u ( x, t ) = u ( x, t ) + z
t
0
(2)
u ( x , t )
i ( x, t )
+ i ( x, t ).R + L
=0
t
x
(3)
i ( x, t )
u ( x, t )
( x, h, t )
+C
=C
x
t
t
(4)
(5)
A( x, t )
t
(6)
13
1
.x
2
r
(vt + z)2 +
P
22/0,4
Load
(7)
RL
y
280m
z0 I0
1
1
I (x, t) =
+
v
4
2
2
r
r
(vt z)2 +
(vt + z)2 +
700m
(8)
N
1,05m
P
100m
Flashing point
III.
SIMULATION
A. System configuration
1.05m
0.35m
A
B
h = 6.5m
C
N
Figure 2.
Configuration
of adistribution
typical low voltage
A typical
section
of 0.4kV
line in system
rural area of
Vietnam as shown in figure 2 was chosen to investigate. The
system consists of 3 phase conductors and 1 neutral conductor
B. Modeling method
Distribution line: A length of 700m of single
phase of the line was simulated. In order to
simulate the maximum induced overvoltage
across the load (between phase and neutral), the
latter is powered by phase A and N (figure 3).
Coupling effect from other phase conductor was
neglected for the sake of simplicity. The line is
assumed to be lossless for the worst case.
Lighting flashes to a point on the ground in the
vicinity of phase A. The line was divided into 10
sections of 70m.
Load: An inductance was used to model the
where S and U are the rated power and rated
voltage of the distribution transformer. In this
case, S=160kVA, U=380V and L=17.89H.
14
q
1
0
Ie (x,t ) =
+
4 Z
2
0 c
r
(vt z)2 +
C. Simulation result
1) Voltage profile along the line
15
CONCLUSION
REFERENCES
[1]
[2]
[3]
[4]
[5]
[7]
[8]
[9]
16
EB-5
INTRODUCTION
Although the use of silicone rubber composite
insulators has been increased significantly in recent
years, porcelain and glass insulators are still
manufactured and remain predominant in distribution
and transmission lines of Vietnam. Except for
hydrophilic property, porcelain and glass insulators are
widely used because they offer many advantages: low
cost, flexible maintenance and high strength. When
energized in polluted area such as coal industry zone
and coastal area, the insulators are easily contaminated
[1], dry band will be formed and leading to flashover
[2].
Other
3%
Pollution
26%
Other
49%
Lightning
71%
Pollution
33%
Lightning
18%
Exploring 7 million tones of coal per year, Quang Ninha northern province of Vietnam faces not only
environment problem arising from coal dust but also the
outages of power distribution and transmission lines.
Field data has recorded about 20% outages in 110kV
transmission line of Quang Ninh province is due to
flashover [3]. Power Company 3 which manages
transmission network of 12 coastal provinces in the
central Vietnam has reported that 33% of service
interruption is also due to flashover [4] (Figure 1).
The paper presents the results of finite element (FE)
calculation of the electrical field distribution along a
string of cap and pin porcelain insulator using in current
transmission lines in Vietnam. Pollution level and dry
band width are varied in order to investigate their
effects on field distribution along the creepage path.
530
B mn H thng in - i hc Bch Khoa H Ni
17
insulator string
531
B mn H thng in - i hc Bch Khoa H Ni
18
PL end
pin
2250
E(V/mm)
1500
750
0
0
14.5
20
conductivity (S)
532
B mn H thng in - i hc Bch Khoa H Ni
19
CONCLUSION
REFERENCES
8000
cap
bandgap
PL end
pin
E(V/mm)
6000
4000
2000
0
0
0.25
0.5
533
B mn H thng in - i hc Bch Khoa H Ni
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
,
,1752'8&7,21
(;3(5,0(17$/
$ &RQGXFWLQJ FRPSRVLWH
$ )UHHVWDQGLQJ 3DQL38 FRPSRVLWH
7KH IUHHVWDQGLQJ 3DQL38 ILOPV ZHUH SUHSDUHG DV GHVFULEHG
LQ >@ 7KH SHUPLWWLYLW\ DQG FRQGXFWLYLW\ RI WKH VDPSOHV ZHUH
PHDVXUHG E\ XVLQJ WKH RSHQ HQGHG FRD[LDO SUREH IRU GLIIHUHQW
ZHLJKW UDWLR RI 3DQL 3DQL38 3DQL38 DQG 3DQL38
ILOPV >@ (OHFWULFDO SURSHUWLHV DW *+] RI WKH FRPSRVLWHV ZHUH
SUHVHQWHG LQ WKH WDEOH ,
7$%/( ,
GP
6P
366
35
3$QL38
ILOP
+
(
3
6( = ORJ W = ORJ W = ORJ W
+L
(L
3L
[
\
Q
Q
(7
(5
]
G
G
G1
0
0
0
0
(,
G
L
=L =
M
+
L
L
NL =
(0 = 0 ) = (0 0 = )
(0 = 0 ) + = (0 0 = )
7=
6( = ORJ 7
36
G
(3)
[( L
+ L M ) F ] DQG LV
WKH ZDYHOHQJWK LQ WKH DLU DQG F LV WKH FXWRII ZDYHOHQJWK RI WKH
7( PRGH LQ WKH ZDYHJXLGH
:KHUH
3$QL38
ILOP
G
M= L VLQN L G L
FRV N L G L
M
VLQ
FRV NL G L
N
G
L L
=
L
[0 L ] =
$W ) *+]
3$QL38
HSV
6( G%
VLJ6P
PP
6 ( G%
G PP
G PP I *+]
6LJP D 6P
HSV
HSV
6 ( G %
HSV
0HDVXUHPHQW
0RGHOLQJ
G P
VW OD\HU
6P
XP
6P
XP
QG OD\HU
6P
G XP
.DSWRQ
G XP
UG OD\HU
6P
G XP
37
,9
46
9/,5+87
,QVWLWXWLRQDO
3URJUDP
LV
JUDWHIXOO\
DFNQRZOHGJHG IRU WKH ILQDQFLDO VXSSRUW SURYLGHG GXULQJ WKH
FRXUVH RI WKLV VWXG\
44
Two layer
43
Three layer
42
41
40
39
0
2E+09
4E+09
6E+09
8E+09
1E+10 1.2E+10
Frequency (Hz)
,,, &21&/86,21
,Q WKLV ZRUN QXPHURXV SDUDPHWHUV ZHUH FKDQJHG LQ RUGHU WR
REWDLQ DQ RSWLPDO VKLHOGLQJ PDWHULDO 7KH VKLHOGLQJ
HIIHFWLYHQHVV LQFUHDVHV LQ H[SRQHQWLDO VFDOH ZLWK WKH WKLFNQHVV
RI PDWHULDO GHVSLWH D UHVRQDQFH SHDN DSSHDUV DW ORZ
FRQFHQWUDWLRQ RI 3DQL 3HUPLWWLYLW\ KDV QHJOLJLEOH HIIHFW RQ
VKLHOGLQJ HIIHFWLYHQHVV LQ WKH ZKROH UDQJH RI WKLFNQHVV ,Q WKH
JRDO RI HODERUDWLQJ D FRQGXFWLQJ PDWHULDO ZLWK GHVLUHG
VKLHOGLQJ HIIHFWLYHQHVV EDVHG RQ 3DQL FRQFHQWUDWLRQ D FXUYH
ZKLFK LOOXVWUDWHV WKH GHSHQGHQFH RI VKLHOGLQJ WR WKH
FRQGXFWLYLW\ ZDV EXLOW 7KH WKUHHOD\HUHG FRQGXFWLQJ VWUXFWXUH
ZDV IRXQG WR KDYH EHWWHU HOHFWULFDO DQG PHFKDQLFDO SURSHUWLHV
WKDQ WKH WZR OD\HUHG RQH HVSHFLDOO\ DW KLJK IUHTXHQF\ %HFDXVH
RI YHU\ WKLQ GLPHQVLRQ WKH PDWHULDO FDQ EH FRPSDWLEOH ZLWK
DHURQDXWLF DSSOLFDWLRQV
9
5()(5(1&(
>@
38
45
SE (dB)
$&.12:/('*0(17
MVP1-1-5
INTRODUCTION
1-y
Material
Pani0.2/PU
Pani0.5/PU
Pani1/PU
Pani4.7/PU
Pani8.8/PU
Pani16/PU
Pani44/PU
EXPERIMENTAL
Shielding electromagnetic material
d(m)
150
150
150
160
155
145
130
V(S/m)
0.2
0.3
4
235
792
2456
11500
75
B mn H thng in - i hc Bch Khoa H Ni
39
f 1 ( x)
(10
SE Fix
20
T)
Minimum
(4)
or:
f1 ( x)
20 log( 10
SE Fix
20
T ) Maximum (5)
Kapton
10
PAni/P
PAni/P
d1
percolation law
d2
d3
Measurment
-6
Material
-10
PAni-PU
Kapton
PAni-PU
3.1
-
-5
-4
-3
-2
-1
0
ln(p-pc)
Conductivity
(S/m)
From 30 to 104
0
From 30 to 104
Thickness (Pm)
From 0 to 300
125
From 0 to 300
Genetic algorithm
Objectif funtion
ln Sigma (S/m)
76
B mn H thng in - i hc Bch Khoa H Ni
40
(m)
2516
328
179
101
47
V(S/m)
4
235
792
2456
11500
Optimization results
The shielding effectiveness of three-layered material was
calculated from a combination of four parameters:
conductivity and thickness of the first and the third layers.
The optimization purpose is to find four parameters that
define a SE best suited for the specific application with the
least weight of PAni and the least total thickness. Knowing
that the required SE is usually greater than 40dB for many
commercial shielding applications, and the SE should be
significantly higher than 80dB for military applications.
For commercial shielding application (SE>40dB) from
50MHz
Table 4: The optimized parameters for commercial application
from 50MHz
Material
1st-layer: PAni /PU
2nd-layer: Kapton
3rd-layer: PAni /PU
H
3.1
-
V (S/m)
2215.3
|0
5888.7
d (m)
78.4
125
59.6
Sample
Pani1/PU
Pani4.7/PU
Pani8.8/PU
Pani16/PU
Pani44/PU
d(m)
150
160
155
145
130
80
70
70
60
50
SE (dB)
SE(dB)
40
40
30
30
20
20
Measurement
10
10
0
8.0E+09
Modeling
50
0
5.0E+06
1.0E+10
1.2E+10
1.4E+10
1.6E+10
1.0E+10
F( Hz)
1.5E+10
2.0E+10
1.8E+10
F(Hz)
Figure 4: Shielding effectiveness of Pani/PU films in X and
Ku bands (Dots are experimental data and solid lines are
modeling results)
5.0E+09
77
B mn H thng in - i hc Bch Khoa H Ni
41
120
Measurement
100
70
d (m)
196
125
180.1
60
8.0E+09
1.0E+10
1.2E+10
1.4E+10
1.6E+10
1.8E+10
2.0
F(Hz)
CONCLUSION
A high level of electromagnetic shielding effectiveness was
achieved by using the Pani/PU blends and the composites
with Kapton film. The physical properties of the Pani/PU
blends were varied from insulating (10-11S/m) to
conducting states (11500 S/m), and showed a very low
percolation threshold. The PAni16/PU film with only 145
m of thickness have an attenuation of 38dB in X and Ku
bands. The structure optimization was performed by using
the genetic algorithm to realize multilayered composites
with a desired SE level. The SE higher than 40dB from
50MHz and 80dB from 8GHz were found with a total
thickness below 1mm, which are compatible with many
commercial and military shielding applications.
ACKNOWLEDGMENT
VLIR-HUT
Institutional
Program
is
gratefully
acknowledged for the financial support provided during the
course of this study
V (S/m)
5803.7
|0
5972.7
6dB
modeling of freestanding film
H
3.1
-
90
80
Material
1st-layer: PAni /PU
2nd-layer: Kapton
3rd-layer: PAni /PU
Thee-layered SE modeling
partition of X and Ku band
110
SE (dB)
REFERENCES
[1] Basudam Adhikari and Sarmishtha Majumdar, Polymers in
sensor application, Prog. Polym. Sci. 29-2004, pp. 699-766.
[2] Angelopoulos M., Conducting polymers in
microelectronics, IMB. J. RES. DEV. Vol. 45, No. 1,
January 2001.
[3] Colaneri Nick F. and Shacklette Laurence W., EMI
shielding measurements of conductive polymer blends,
IEEE Transaction on Instrumentation and Measurement,
Vol.4, No.2, pp. 291-297, 1992.
[4] Krishna Naishadham, "Shielding effectiveness of
conductive
polymers",
IEEE
Transaction
on
Electromagnetic Compatibility, Vol. 34, No. 1, pp. 47-50,
1992.
[5] Cao Y., Qiu J. and Smith P., Effect of Solvents and
Co-solvents on the Processibility of Polyaniline: I.
Solubility and Conductivity studies Synthetic Metals 69,
pp. 187-190, 1995.
[6] Hoang Ngoc Nhan, Pham Hong Thinh and Nguyen Thi
Lan Huong, Electromagnetic characterization of Pani/PU
in multilayered structure, application for EMI protection at
microwave frequency, Proceeding of 2th International
Conference on Communication and Electronics, pp.
366-369, June 4-6, 2008
[7] Holland J., Adaptation in Natural and Artificial Systems,
The University of Michigan Press, Ann Arbor, 1975.
[8] Satheesh Kumar KK., Geetha S. and Trivedi D.C.,
Freestanding conducting polyaniline film for the control
of electromagnetic radiations, Current Applied Physics,
Volume 5, Issue 6, pp. 603-608. September 2005.
78
B mn H thng in - i hc Bch Khoa H Ni
42
Tng hp cc bi bo2009
khoa hc
giai on
2007-2012
Annual
Report
Conference
Fig. 1. Glass and silicone rubber insulators using in 35kV distribution network
I. INTRODUCTION
With the low height of towers, the insulator of distribution
network located in industrial and populated areas including the
coastal one is especially polluted by metallic oxide and metal
in form of dusts or powders, and also by soluble ionic and
inert layers. When energizing, the conductivity layer could be
formed with or without water on the insulator and leads to
flashover. The development of the leakage current (LC) just
prior to flashover leads to the damage of insulator surface [1],
and it was usually used as a tool for predicting the
performance of the insulator [2,3]. However, each type of
insulator has the particularity on their LC waveform
depending on the material in use, the rated voltage, and also
the condition of environment.
In this work, a system for LC measurement is specifically
developed for insulator in distribution network with the rated
voltage of 35kV. Frequency analysis of the LC has been
carried out for two types of insulator in glass and in silicone
rubber. The solid layer method has been used to simulate four
degree of pollution severity. The relationship between
different harmonic content of the LC and the pollution degree,
the electric stress and the resting time has been also discussed.
TABLE I
TECHNICAL DATA FOR GLASS AND SILICONE RUBBER INSULATORS
Shed diameter
Unit spacing
Leakage distance
Number of unit or shed
per string
TABLE II
THE SALINITY AND EQUIVALENT SDD OF THE TESTING PROCEDURE
Severity
A. Insulator parameters
Different type of insulator is currently used in Vietnam
distribution network. Standard cap & pin glass and porcelain
insulators are widely used in most of the old system. The
silicone rubber insulator is recently put into operation in new
Silicone rubber
95mm
580mm
880mm
8
B. Testing procedure
The solid layer method (clean fog method) [4] has been used
to simulate the pollution severity of the testing insulator.
While using the tap water, kaolin and a suitable amount of
NaCl, the desired volume conductivity of the suspension on
the insulator under test could be obtained. From the relation
between the volume conductivity of the suspension and the
amount of NaCl and the salt deposit density (SDD), the
surface conductivity of the insulator under test could be
inferred as described in [4]. In this work, the amount of NaCl
dissolved in 1000ml of tap water were 14g, 25g, 55 and 90g,
which correspond to 4 degrees of pollution as defined in [5]
(Table II)
Glass
225mm
146mm
303mm/unit
3
Light
Medium
Heavy
Very heavy
462
Salinity
(g/l)
14
25
55
90
Surface conductivity
(S/m)
2.2
3.8
7.6
10.9
SDD
(mg/cm2)
0.05
0.09
0.20
0.32
43
100 AI 2 2 AI 3 2 ... AI N 2
(1)
AI1
where, A(I1) is the amplitude of the fundamental component,
A(IN) is the amplitude of the Nth harmonic and N is the
number of harmonics.
To PC
Fig.5. Harmonics of leakage current and THD of glass insulator versus the
salinity at 35kV
Data
acquisition
board
To OSC
Transformer 220V/12V
Voltage signal
from shunts
Shielded
box
D. Signal analysis
The interface communication and the analysis were
performed by means of LabView package [6]. For the purpose
of storing and analyzing LC data, two main programs have
been written in this package. The first one is used to transfer
voltage waveforms from the microcontroller to the PC and
save them as data files. The second one displays the data files
in time domain and analyzes them in frequency domain using
the fast Fourier transformation (FFT). The total harmonic
463
44
Fig. 9a. LC waveform of glass insulator at 5kV and under very heavy
pollution
Fig. 9b. LC waveform of glass insulator at 15kV and under very heavy
pollution
464
45
Fig. 9c. LC waveform of glass insulator at 25kV and under very heavy
pollution
Fig. 9d. LC waveform of glass insulator at 35kV and under very heavy
pollution
ACKNOWLEDGMENT
This article was funded in part by a grant from the Vietnam
Education Foundation (VEF). The opinions, findings, and
conclusions stated herein are those of the authors and do not
necessarily reflect those of VEF.
REFERENCES
[1] El-Hag A. H. Ayman, H. El-Hag, Shesha H. Jayaram and Edward A.
Cherney Fundamental and low frequency harmonics components of
leakage current as a diagnostic tool to study aging of RTV and HTV
silicone rubber in salt-fog, IEEE Trans. Dielectr. Elec. Insulation, vol.
10, pp. 655-664, August 2003.
[2] T. Suda, Frequency Characteristics of Leakage Current Waveforms of a
String of Suspension Insulators, IEEE Trans. Power Delivery, vol. 20,
pp. 481-487, January 2005.
[3] M. Fernando, Performace of Non-ceramic Insulators in Tropical
Environments, PhD thesis of Chalmers University of Technology, 1999
[4] International Electrochemical Commission, Artificial Pollution Tests on
HV Insulators to be Used on AC Systems, IEC Publications No. 507,
Second Edition, 1991
[5] International Electrochemical Commission, Guide for the Selection of
Insulators in Respect of Polluted Conditions, IEC Publications No. 815,
1979
[6] LabVIEW User Manual, National Instruments, 1998
IV. CONCLUSION
465
46
Tng hp cc bi bo2009
khoa hc
giai on
2007-2012
Annual
Report
Conference
INTRODUCTION
(1)
(2)
Where:
J: current density
V: electrical conductivity
E: electric field
B. Electrostatic Analysis
Another method used to calculate the grounding resistance
in FEM is to know the stored energy by the electric field in the
ground:
(3)
(4)
278
47
I.
300
250
200
SIMULATION RESULTS
R,
150
100
50
0
10
12
14
Electrode length, m
Fig 3. Grounding resistance of horizontal rod versus the length
20000
Perpendicular to
the rod
15000
10000
5000
12000
0
0
10
15
20
25
Distance from the electrode end, m
10000
30
8000
Fig 2. Step voltage distribution on the soil surface, along and perpendicular to
the horizontal rod
Perpendicular to the
horizontal rod
6000
4000
2000
0
0
10
15
20
25
30
279
48
III.
18
9000
16
8000
14
7000
20mx20m
12
6000
10mx10m
R, :10
V 5000
4000
3000
2000
1000
4mx4m
0
0
4
6
Distance between two vertical rods, m
10
20
30
40
50
X(m)
9000
8000
7000
4mx4m
6000
10mx10m
5000
20mx20m
4000
3000
2000
10
20
30
40
50
Y(m)
Fig. 9. Y-axis potential profile for 16 mesh grid without ground rod
TABLE I
PARAMETERS AND SIMULATION RESULTS OF REDUCED SCALE MODEL IN GRID
WITHOUT GROUND RODS
Scale
Grid
Grid
Buried
factor
dimension
conductor
depth
diameter
Soil
Ground
resistivity resistance
.m
Max. step
voltage,
1251.9
cm
20x20
0.6
400
7.85
10x10
0.3
200
8.24
1266
1/5
4x4
0.4
0.12
80
8.61
1268.7
1
1/2
280
49
1000
9000
8000
20mx20m
6000
10mx10m
5000
4mx4m
ACKNOWLEDGMENT
4000
3000
2000
1000
0
10
20
30
40
50
X(m)
Fig. 10. X-axis potential profile for 16 mesh grid with 16 ground rods
installed in surrounding junction
REFERENCES
9000
8000
[2]
[3]
10mx10m
6000
[1]
20mx20m
7000
4mx4m
5000
4000
[4]
3000
2000
[5]
1000
0
0
10
20
30
40
50
[6]
Y( m)
Fig. 11. Y-axis potential profile for 16 mesh grid with 16 ground rods
installed in surrounding junction
[7]
TABLE II
PARAMETERS AND SIMULATION RESULTS OF REDUCED SCALE MODEL IN GRID WITH
GROUND RODS
[8]
Scale
Grid
Grid
Radius of
Buried
factor
dimension
conductor
ground
depth
diameter
rod
cm
Soil
Ground
resistivity resistance
.m
Max. step
voltage,
[9]
cm
20x20
0.6
400
7.08
1175.1
1/2
10x10
0.5
0.3
200
7.63
1181.3
1/5
4x4
0.4
0.2
0.12
80
7.90
1179
[10]
[11]
CONCLUSION
281
50
7000
I.
INTRODUCTION
II.
FLASHOVER MODELS
A. Volt-time curve
The voltage-time curve has been determined experimentally
for the specific gap of arcing horn using in 220kV transmission
line insulator (Figure 1). This curve is determined using the
standard lightning impulse waveform applied across the arcing
horn. During the simulation, the voltage across the insulator is
compared with the volt-time curve. If simulated voltage
exceeds volt-time curve, flashover occurs in the simulation (an
ideal switch is closed through the inductance of the arc. As
recommended by IEC [7], the inductance of the arc is 1H/m.
The initial time of the V-t curve must be synchronized to the
instant that the voltage reaches the position of the insulator in
question, which is differs for each tower in a transmission line.
When the simulation is performed, this time lag must be
included in the model.
4000
3500
3000
2500
2000
1500
0.0
0.5
1.0
1.5
time (s)
2.0
2.5
51
Voltage (kV)
In ATP/EMTP simulation, the first computation is performed with SW1 and SW2 open. The close times of these
switches are determined by measuring the current through the
arcing horn. By repeating this procedure from the insulator
which is the most prone to breakdown to those which are less
so until no flashover occurs, the simulation can be completed.
polarity
polarity
for positive
SIMULATION RESULTS
12
[kA]
10
8
for negative
6
4
2
0
12
16
[us] 20
V-t
No flashover
No flashover
No flashover
No flashover
No flashover
3.2E-6
2.9E-6
9.4E-7
5.5E-7
3.9E-7
3.1E-7
Kind
No flashover
No flashover
No flashover
No flashover
No flashover
6.3E-6
4E-6
8.7E-7
4.4E-7
2.95E-7
2.2E-7
Motoyama
No flashover
No flashover
2.9E-6
2.7E-6
2.5E-6
2.16E-6
2E-6
1.03E-6
6.7E-7
5.3E-7
4.3E-7
Ueda
No flashover
No flashover
3E-6
2.48E-6
2.41E-6
2.27E-6
2.2E-6
1.35E-6
9.25E-7
7.55E-7
6.5E-7
52
III.
496m
560m
#2
#3
533m
#4
237m
#5
740m
#6
#7
53
200
V-t
No flashover
No flashover
No flashover
1A, 2A, 4A,
7A
1A, 2A, 4A,
5A, 6A, 7A
1A, 3A, 4A,
5A, 6A, 7A
Kind
No flashover
No flashover
4A
4A
Motoyama
No flashover
4A
4A, 4B
4A, 4B
Ueda
No flashover
4A
4A
4A, 4B
4A
4A, 4B
4A, 4B
4A
[2]
[3]
[4]
[5]
[6]
CONCLUSION
[7]
[8]
[9]
Fast Front Transient Task Force of the IEEE, Modeling guidelines for
fast front transients, IEEE Trans. Power Delivery, vol. 11, No. 1,
January 1996
T. Shindo, Y. Aoshima, I. Kishizima, T. Harada, A study of
predischarge current characteristic of long air gaps, IEEE Trans. on
Power Apparatus and Systems, vol. PAS-104, No. 11, November 1985.
A. Pigini, G. Rizzi, E. Garbagnati, A. Porrino, G. Baldo, G. Pessavento,
Performance of large air gaps under lightning overvoltages :
Experimental study and analysis of accuracy of predetermination
methods , IEEE Trans. on Power Delivery, vol. 4, no. 2, April 1999.
H. Motoyama, Experimental study and analysis of breakdown
characteristics of long air gaps with short tail lightning impulse, IEEE
Trans. on Power Delivery, vol. 11, no. 2, April 1996.
CIGRE WG 33-01, Guide to procedure for estimating the lightning
procedures the lightning performance of transmission lines, October
1991.
O. Ueda, S. Neo, T. Funabashi, T. Hagiwara and H. Wanatabe,
Flashover model for arcing horns and transmission line arresters,
International Conference on Power Systems Transients, Lisbon, 3-7
September 1995.
IEC Technical Report 60071-4, Insulation co-ordination. Part 4:
Computational guide to insulation co-ordination and modeling of
electrical networks, 2004-06.
CIGRE WG 33-02, Guidelines for representation of network elements
when calculating transients, No. 39, 1990.
A. Ametani and T. Kawamura, A method of a lightning surge analysis
recommended in Japan using EMTP, IEEE Trans. on Power Delivery,
vol. 20, no. 2, April 2005.
54
1991
I. INTRODUCTION
IGHTNING performance is one of the most important aspects of transmission line design, as most transmission
line outages are caused by lightning. Lightning can strike the
shielding wire (mid-span or tower top) to create back ashovers
or strike a phase conductor as a result of shielding failure.
Shield wires, in combination low footing resistance, improve
the lightning performance of a transmission line, and the application of line arresters provides an additional increment of protection. Such arresters come in two forms, gapless [1], [2] and
externally gapped (EGLA) [3], [9], [10]. The main advantage of
an EGLA is reduced size and weight, as clearing of the series
gap after a lightning event removes the arrester from the line so
that thermal stability of the metaloxide varistors (MOVs) after
the lightning event is not an issue. In some cases, transmission
Manuscript received September 15, 2011; accepted February 04, 2012. Date
of publication September 07, 2012; date of current version September 19, 2012.
This work was supported in part by the Vietnam Education Foundation (VEF).
Paper no. TPWRD-00783-2011.
T. Pham is with Hanoi University of Science and Technology, Hanoi, Vietnam
(e-mail: thinh.pham@ims.uconn.edu).
S. A. Boggs is with the Electrical Insulation Research Center, University of
Connecticut, Storrs, CT 06269 USA (e-mail: steven.boggs@ieee.org).
H. Suzuki and T. Imai are with the Surge Arrester Department,
Hamakawasaki Operations, Toshiba Corporation, Kawasaki-ku 210-0862,
Japan (e-mail: hironori.suzuki@toshiba.co.jp; toshiya.imai@toshiba.co.jp).
Color versions of one or more of the gures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identier 10.1109/TPWRD.2012.2205729
B mn H thng in - i hc Bch Khoa H Ni
(2)
55
Thinh H. Pham, Steven A. Boggs, Fellow, IEEE, Hironori Suzuki, Member, IEEE, and Toshiya Imai
Fig. 1. Tower conguration (left) and its equivalent EGM model to determine expose width of shielding wire and phase conductors. The parameters r (I), r (I),
and r (I) are computed from (1) and are used to compute the values of D and D , which are the exposure widths which enter into ((2) and (3)). The indices c, s,
and e stand for conductor, shielding wire, and earth. Figure adapted from [8].
For a line protected by two shielding wires which are separated by a distance , (2) [8] becomes
(3)
where
is the ground ash density, 10 per km -year in this
study,B mn
is the
length of the line,
is maximum lightning
H thng in - i hc Bch Khoa H Ni
where
1.33 and
61.1 for
20 kA, and
0.605
and
33.3 for
20 kA.
Fig. 2 shows the frequency of strokes to shielding wires and
phase conductors per 100 km-year for a 60 m high tower as
computed using EGM. For large lightning current, the strike distance increases to the point that a strike can only terminate on
shielding wires or earth. In the present tower conguration, the
probability of shielding failure is relatively small and can occur
only for relatively low currents, that is, 28 kA for the56upper
1993
Fig. 4. Transmission line to be modeled. In the simulation, spans from tower 12 and 67 are included along with their matching impedances which avoid reections from the ends. The phase conductor is ACSR 330 2 (aluminum conductor steel reinforced), the shielding wire (SW) is TK-70 and OPGW-82 (optical
ground wire).
Fig. 5. Potential difference across the upper phase insulator of the stricken
tower when lightning strikes the top of tower 4 for various values of adjacent
tower footing resistance. The footing resistance of stricken tower is 20
, and
the lightning current is 100 kA with linear ramp waveform (2/70 s).
and the upper phase conductor than for the middle and lower
phases, which results in greater probability of the ashover of
the lower and middle phase insulators than the upper phase insulator. However, this difference only changes the order in which
the phases ash over but not the total number of phases which
ash in a given simulation, which is the most important parameter when implementing EGLAs in a transmission line.
The tower is represented by four lossless transmission lines,
and the footing impedance of each tower is a simple resistance
as suggested in [6]. The ashover of the EGLA series gap or
the arcing horn gap is simulated by the Motoyama model for
reasons discussed in [5], [7].
Tower footing resistance varies along a transmission line depending on the soil resistivity. When lightning strikes to the
tower top or the phase conductor, the overvoltage across an insulator depends on both the footing resistance of the stricken
tower and that of the adjacent towers. Fig. 5 shows the dependence of overvoltage across an insulator of the stricken tower
on
57
TABLE I
INSULATORS WHICH FLASHOVER AS A FUNCTION OF 2/70 S LIGHTNING CURRENT AMPLITUDE FOR LIGHTNING
TO THE TOP OF TOWER 4 WITH 60 M TOWER HEIGHT, 30
FOOTING
1995
TABLE II
FLASHOVER OF INSULATORS AND OPERATION OF SURGE ARRESTERS AS A FUNCTION OF 2/70 S LIGHTNING CURRENT AMPLITUDE FOR LIGHTNING TO THE TOP
OF TOWER 4 WITH 60 m TOWER HEIGHT, 30
FOOTING RESISTANCE, AND EGLAS INSTALLED ON THE UPPER PHASE OF CIRCUIT 1
TABLE IV
FLASHOVER OF INSULATORS AND OPERATION OF SURGE ARRESTERS AS A FUNCTION OF 2/70 S LIGHTNING CURRENT AMPLITUDE FOR LIGHTNING TO THE TOP
OF TOWER 4 WITH 60 m TOWER HEIGHT, 30
FOOTING RESISTANCE, AND EGLAS INSTALLED ON ALL THREE PHASES OF CIRCUIT 1
Installing EGLAs on upper and middle phases of one circuit protects the insulators of both circuits from ashover up to
120 kA to the tower top. Flashover of the lower phase, which
is not protected by an EGLA, occurs only after ashover of
two phases of second, unprotected circuit for lighting currents
greater than 130 kA. Thus installation of EGLAs on the upper
and middle phases of one circuit prevents a double circuit outage
up to 130 kA. Where the probability of lightning currents above
120 kA is negligible, installation of EGLAs on only the upper
two phases of one circuit may be justied. For the transmission
line in question, the frequency of greater than 120 kA lightning
current to a shielding wire is about 7 per 100 km-year (Fig. 2),
which means 0.8 times per year for a 10 km transmission line,
which is relatively low given that most of towers on an actual
220 kV transmission line are likely to be closer to 40 m high
rather than to 60 m assumed before, with only a small fraction
of the towers near 60 m. Above 150 kA, ashover occurs across
the lower phase insulators of circuit 1 and all three phases of
circuitB
2 mn
(Table
III).
H thng in - i hc Bch Khoa H Ni
TABLE III
FLASHOVER OF INSULATORS AND OPERATION OF SURGE ARRESTERS AS A FUNCTION OF 2/70 S LIGHTNING CURRENT AMPLITUDE FOR LIGHTNING TO THE TOP
OF TOWER 4 WITH 60 m TOWER HEIGHT, 30
FOOTING RESISTANCE, AND EGLAS INSTALLED ON UPPER AND MIDDLE PHASES OF CIRCUIT 1
TABLE V
FLASHOVER OF INSULATORS AND OPERATION OF SURGE ARRESTERS AS A FUNCTION OF 2/70 S LIGHTNING CURRENT AMPLITUDE FOR A LIGHTNING STRIKE TO
THE UPPER PHASE OF TOWER 4 WITH 60 m TOWER HEIGHT, 30
FOOTING RESISTANCE, AND EGLAS INSTALLED ON ALL THREE PHASES OF CIRCUIT 1
V. CONCLUSION
Simulations were performed for a twin circuit, 220 kV
transmission line to estimate the effectiveness of EGLAs as a
function of the number of EGLAs installed, increasing from
0 (without EGLAs) to 3 (EGLAs on all three phases of one
circuit). The EGM model was used to estimate the lightning
frequency to the tower top (or shielding wire) and to phase
conductors.
In case of lightning to a tower top, installing EGLAs on one
or two phases of one circuit improves the lightning performance
of the transmission line substantially. However a double circuit outage can still occur at very high lightning current, and/or
with high footing resistance. Installing EGLAs with appropriate
rating on all three phases of one circuit reduces the likelihood of
a double circuit outage to near zero for lightning to tower top,
but insulator ashover on the unprotected parallel circuit still
occurs for very high lightning current and/or high footing resistance. Installing EGLAs on one or more phases of the second
circuit should be considered at towers with high footing resistance ( 10 ), which is unusual for most transmission lines.
The number of EGLAs installed on one circuit has a small effect
on the maximum duty of a single EGLA (discharge current and
energy absorption) which suggests that the same rating EGLA
should be selected over the entire transmission line no matter
how many EGLAs are installed per circuit.
The lightning current and strike frequency for shielding failures (strikes to a conductor) is much lower than for strikes to the
shielding wires and tower top. However, a lightning strike to a
phase conductor can cause insulator ashover on the stricken
phase if it is not protected by an EGLA, although a double circuit outage is unlikely to occur if EGLAs are installed on each
phase of one circuit. Given the very low probability of shielding
failure, the risk of a single circuit outage may be worth taking.
REFERENCES
[1] T. Wakai, N. Itamoto, T. Sakai, and M. Ishii, Evaluation of transmission line arresters against winter lightning, IEEE Trans. Power Del.,
vol. 15, no. 2, pp. 684690, Apr. 2000.
[2] Y. Matsumoto, O. Sakuma, K. Shinjo, M. Saiki, T. Wakai, T. Sakai, H.
Nagasaka, and H. Motoyama, Measurement of lightning surges on test
transmission line equipped with arresters struck by natural and trigged
lightning, IEEE Trans. Power Del., vol. 11, no. 2, pp. 9961001, Apr.
1996.
[3] S. Furukawa, O. Usuda, T. Isozaki, and T. Irie, Development and
application of lightning arresters for transmission lines, IEEE Trans.
Power Del., vol. 4, no. 4, pp. 21212128, Oct. 1989.
[4] H. R. Amstrong and E. R. Whitehead, Field and analytical studies
of transmission line shielding, IEEE Trans. Power App. Syst., vol.
PAS-87, no. 1, pp. 270281, Jan. 1968.
Thinh H. Pham received the B.S. degree in power system engineering from
Hanoi University of Science and Technology, Hanoi, Vietnam, in 1996, the
M.Sc. degree in electrical engineering from Grenoble Institute of Technology,
Grenoble, France, in 2001, and the Ph.D. degree in electrical engineering from
University of Joseph Fourier, Grenoble, in 2005.
Since 2006, he has been a faculty member in the Department of Power Systems, Hanoi University of Science and Technology. From 2009 to 2011, he
was with the Electrical Insulation Research Center (EIRC), University of Connecticut, as a Visiting Professor and then a Postdoctoral Fellow. His research
interests focus on transient phenomena and insulation coordination in power
systems and dielectric properties of insulation subjected to high elds.
Steven A. Boggs (F92) was graduated with a B.A. degree from Reed College,
Portland, OR, in 1968, and the Ph.D. and MBA degrees from the University of
Toronto, Toronto, ON, Canada, in 1972 and 1987, respectively.
He spent 12 years with the Research Division of Ontario Hydro. He was
elected an IEEE Fellow for his contributions to understanding of SF -insulated
systems. From 1987 to 1993, he was Director of Research and Engineering at
Underground Systems, Inc. He is presently Director of the Electrical Insulation
Research Center and the Research Professor of Materials Science, Electrical
Engineering, and Physics at the University of Connecticut, Storrs, and Adjunct
Professor of Electrical Engineering at the University of Toronto.
Hironori Suzuki (M12) was born on June 29, 1959. He received the B.S. and
M.S. degrees in physics from the University of Tsukuba, Ibaraki, Japan, in 1982
and 1984, respectively.
In 1984, he joined Toshiba Corporation, Kawasaki, Japan. Since then, he
has been engaged in the development of zincoxide elements and surge arresters. Currently, he is an Assistant Secretary of the standard committee for
metaloxide surge arresters with the Institute of Electrical Engineers of Japan.
Toshiya Imai was born in Hyogo Prefecture, Kansai, Japan, in 1966. He received the B.S. degree in inorganic material engineering from Kyoto Institute
of Technology, Kyoto, Japan, in 1989.
He joined Toshiba Corporation, Kawasaki, Japan, in 1989. Since then, he has
been engaged in the development of zincoxide elements and metal-oxide surge
arresters.
61
1997
SATIE, ENS de Cachan, CNRS, UniverSud, 61, av President Wilson F-94230-Cachan, France.
Department of Energy, Suplec, Campus de Gif-sur_Yvette, 91192 GIF SUR YVETTE, France.
3
Department of Power System, Hanoi University of Technology, Hanoi, Vietnam.
4
SATIE-Brittany Branch, ENS de Cachan-Ker Lann Campus, 35170 BRUZ, France.
E-mail: xdang@satie.ens-cachan.fr
Key words
Fig. 1. Overview circuit of the PV array system
1. Introduction
In recent years, the development on the installed capacity
and the improvement on the efficiency of the renewable
energy sources are implemented in many countries around
the world due to the exhaustion and pollution of fossil
fuels. One of the most attractive renewable energy sources
is photovoltaic array, because PV is the most direct way to
convert solar radiation into electricity and is based on the
photovoltaic effect [2]. Otherwise, there is no mechanical
moving parts, no noise, no pollution and PV panels have a
long lifetime. Moreover, the energy source is the sun
which is free, ubiquitous and inexhaustible. In some
popular appliances, it is usually used in charging battery,
hybrid vehicles, water pumping and connecting into
Distributed Energy Resources (DER) systems.
As a power source generated by solar array depending so
much on the solar radiation, temperature and array voltage,
it is essential to control the operating points to yield the
qVPV I PV RS (VPV I PV RS )
1
I PH I S exp
kTC A
RSH
(1)
62
qVPV N S I PV RS N P
1
N P I PH N P I S exp
kTC A
(2)
iD
qVPV N S I PV RS N P
1 1 D
N P I PH N P I S exp
kTC A
VOUT 1 D
VIN
(7)
(6)
(5)
VBatt1 D
(8)
qVBatt1 D N S I PV RS N P
1 1 D
N P I PH N P I S exp
kTC A
(9)
From the Fig.2 above, when the switch IGBT is off, the
diode is connected to the inductor. Using the small-ripple
approximation, the inductor current is equal to the output
current of PV array, so the current through the diode iD(t)
is obtained:
2.9
2.8
2.7
2.6
2.5
0
0.2
0.4
0.6
0.8
1.2
1.4
1.6
1.8
1.2
1.4
1.6
1.8
time (s)
0.25
i D t i PV t
(3)
Duty cycle
0.2
0.15
0.1
0.05
0
0
0.2
0.4
0.6
0.8
time (s)
63
3
2
O = 0.9 (kW/m )
2.5
Tc=15oC
2.5
O = 0.6 (kW/m2)
Tc=35oC
Tc=55oC
O = 0.3 (kW/m )
<ID>(A)
1.5
1.5
0.5
0.5
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0
0
0.1
0.2
0.3
0.4
0.7
0.8
0.9
modulating signal
1
Tc=15oC
0.5
Tc=35oC
Tc=55oC
0
-0.5
-1
0
O = 0.9 (kW/m )
O = 0.3 (kW/m2)
O = 0.6 (kW/m2)
2
Tc=15oC
Tc=35oC
0
0
time (s)
60
40
20
0
0
-0.5
time (s)
Power (W)
1
0.5
Tc=55oC
<ID> (A)
time (s)
0
0
<ID> (A)
modulating signal
0.6
time (s)
-1
0
time (s)
3
<ID> (A)
0.5
O = 0.9 (kW/m2)
O = 0.6 (kW/m2)
2
O = 0.3 (kW/m )
0
0
time (s)
64
<ID> (A)
Reference
[1]
Solar radiation
(kW/m2)
0.3
0.6
0.9
MAC
Simulation
0.93
1.91
2.87
0.88
1.82
2.74
Error
(%)
5.38
4.71
4.53
[2]
[3]
[4]
Table I (b). Comparison results between MAC and Simulation
models with different temperatures
Temperature
(oC)
15
35
55
MAC
Simulation
2.87
2.42
1.95
2.74
2.28
1.84
Error
(%)
4.53
5.79
5.64
[5]
5. Conclusion
[6]
The proposed MAC method allows optimizing the electric
power which is generated from the PV array close to the
maximum power point of the array under changing
atmospheric conditions.
In this paper, the difference between the MPPT methods
and the proposed MAC method were proven by searching
the maximum average current passing through the diode.
Furthermore, the simulation model takes into account the
conducting losses in the inductor, diode and switch IGBT.
The losses generate the fluctuation during the first few
seconds of the holding maximum value period.
[7]
[8]
65
1 INTRODUCTION
THE influence of water uptake on dielectric properties of
epoxy-glass-mica composite materials has been previously
reported in several papers. In such materials, a silane
treatment of glass is usually performed to reduce the
degradation occurring at interfaces with the epoxy matrix. The
degradations induced by water mostly appear to be localised at
interfaces between basic components (epoxy, glass and mica),
due to the enhancement of chemical reactions [1]. Electrical
measurements show that the presence of moisture produces an
increase of permittivity and loss tangent [2, 3], growth of trees
[4, 5], and can reduce the breakdown strength [2].
Water impregnation of epoxy composites was used to study
the ageing processes induced by humidity [6]. In large
generators, epoxy composites are used for the insulation of
stator bars. In service conditions, many different degradation
processes may occur and interact [7]. Thermal, electrical and
mechanical stresses mainly contribute to the ageing of bar
insulation. In the case of water-cooled stator bars, water leaks
although very rare may also constitute a factor of premature
ageing, in addition to the causes mentioned above. Analysis of
water-permeated bars indicate local water contents in the
insulation higher than 1% close to copper strands, associated
to a large reduction of insulation life [8]. For practical
purposes, it is very important to characterize and understand
Manuscript received on 2 June 2008, in final form 17 September 2008.
B mn H thng in - i hc Bch Khoa H Ni
2 WATER ABSORPTION
2.1 MATERIAL AND SAMPLE PREPARATION
The epoxy composite studied here is currently used in stator
bar insulation of very large generators (some 100 MW). The
material has a laminated structure consisting of alternating
layers of glass cloth and mica, binded together by an epoxy
resin (diglycidyl ether of bisphenol A DGEBA - cured with
a polyester acid). Figure 1a shows a SEM (Secondary Electron
Microscopy) picture of the composite cross-section.
Thicknesses of the mica and glass cloth layers are about 100
m and 50 m respectively. After burning out the resin at 600
C, the residual mass corresponding to mica and glass was
measured. It accounts for about 67% of the total composite
Authorized licensed use limited to: La Van Ut. Downloaded on February 23, 2009 at 01:30 from IEEE Xplore. Restrictions apply.
66
ABSTRACT
Water absorption in an epoxy composite used for insulation of stator bars in large
generators is investigated in order to study both diffusion kinetics and effects on
dielectric properties. Water absorption can be described by the superposition of two
Fickian mechanisms, the first occurring within epoxy and the other at interfaces
between epoxy and fillers. A large decrease of breakdown strength versus water uptake
is observed. It points to the large deleterious effect of water on insulation reliability.
Dielectric spectroscopy is performed under various conditions of water impregnation,
electric field and temperature. A very large increase of permittivity and losses is
recorded in the presence of water, especially at low frequency. At power frequency,
dielectric properties do not show a non-linearity when the field is increased up to
service values. At very low frequency, a decrease of losses when the field is increased is
observed. It is due to the Garton effect. Most of results show that the wet material
behaves as a low frequency dispersion system (LFD).
glass
mica
m m0
.100(%)
m0
(1)
glass cloth
mica layer
c
Figure 1. a: SEM images of the composite cross-section, b: glass cloth and c:
mica layers obtained after burning the resin.
This insulation material was machined to provide diskshaped samples of 50 mm diameter with thickness from 1 to 4
mm. The faceplate of disks was parallel to glass cloth layers.
Samples were first dried at 120 C in a vacuum oven for more
than 1 week. The efficiency of the drying procedure was
monitored by weighing samples with a high precision balance
(0.1 mg resolution). The sample weight first dropped, and then
became constant. No variation was recorded during the last 3
days of the drying procedure. Taking into account the
h
S
4m f
m 2 m1
t t
2
1
(2)
D0 exp(
A
)
kT
(3)
67
1.4
1.2
M(%)
1.0
0.8
0.6
60 C-measured
60 C-Fick's law
90 C-measured
90 C-Fick's law
110 C-measured
110 C-Fick's law
0.4
0.2
0.0
10
20
1/2
t(h )
30
40
1.0
M(%)
0.8
0.6
0.4
0.2
0.0
60 C
90 C
110 C
4
1/2
t(h )
Figure 3. Initial stages of water absorption versus the square root of time (1
mm thick samples).
-2
D=1090exp(-4809/T)
-1
D(mm .h )
10
-3
10
-4
10
2.6
2.7
2.8
-1
1000/T(K )
2.9
3.0
0.75
1 exp 7.3 Dt m
f
2
(4)
m1f D1 m2 f D2
(5)
mf D
68
1.5
60C-measured
M P model
M ode1
M ode2
1.0
M(%)
3.00
2.25
10
20
1/2
t(h )
30
40
1.50
0.75
1 mm
2 mm
4 mm
0.00
T(C)
60
90
110
10
20
1/2
t(h )
30
40
50
Figure 6. Water absorption versus the square root of time for 1, 2 and 4 mm
sample thickness.
69
0.0
M(%)
0.5
3. BREAKDOWN MEASUREMENTS
80
60
PD inception
40
20
0
0
0.4
0.8
1.2
Water content (%)
1.6
4. DIELECTRIC SPECTROSCOPY
4.1. EFFECT OF WATER CONTENT
Water absorption does not only reduce breakdown strength,
but also considerably modifies dielectric properties. Dielectric
responses of samples at different water contents were
measured over the frequency ranges 10-3 to 103 Hz at different
thicknesses, temperatures and electric fields.
These measurements were carried out with a dielectric
spectrometer equipped with a 30 kV amplifier, offering the
possibility to study the influence of electric field up to some
kV/mm.
Dielectric properties were measured on the same sample at
different times during the impregnation process. The following
70
10
dry
0.05%
0.1%
0.18%
0.5%
1%
saturated
De-ionized water
10
10
10
H'
10
1/(Z)
10
10
10 -3
10
HV ELECTRODE
GUARD
ELECTRODE
10
f(Hz)
10
10
10
MEASUREMENT
ELECTRODE
10
10
10
10
10
-1
10
De-ionized
water
n=-0.93
dry
0.05%
0.1%
0.18%
0.5%
1%
saturated
n=-0.95
H"
n=-0,95
10
n=-0.93
n=-1
-2
10
-3
10
-2
10
-1
10
f(Hz)
10
10
10
Figure 11: Loss factor of a 4 mm thick sample for different water uptake (10
V/mm, room temperature). Data for de-ionized water come from [21].
71
SAMPLE
-2
10
H"
10
10 V/mm
100 V/mm
1 kV/mm
M=0%
-1
10
1 mm
4 mm
-2
10 -3
10
10
-1
10
f(Hz)
10
10
10
10
0.4
0.6
M(%)
0.8
10
10
M=0.1%
10 V/mm
100 V/mm
1 kV/mm
1.0
H"
0.2
10
H'
-2
10
0.0
10
10
-1
10 -3
10
H''
10
10
-1
10
1 mm
4 mm
-2
0.0
-2
10
-1
10
f(Hz)
10
10
10
Figure 15: Field dependence of loss factor versus frequency in a dry sample
(1 mm, room temperature).
10
10
0.2
0.4
0.6
M(%)
0.8
1.0
72
10
M=0.1%
RT
30 C
40 C
50 C
60 C
70 C
80 C
H'
H'
10
100
E(V/mm)
Figure 16: Field dependence of dielectric constant for different water content
at 50 Hz (1 mm sample, room temperature).
100
H''
10
-1
10
f(Hz)
10
10
10
dry
0.05%
0.1%
0.2%
0.5%
1%
10
H"
-2
10
1000
10
10
10
10
M=0.1%
1
10
dry
0.05%
0.1%
0.2%
0.5%
1%
RT
30 C
40 C
50 C
60 C
70 C
80 C
0.1
10
0.01
10
-1
10
100
E(V/mm)
-2
10
-1
1000
Figure 17: Field dependence of loss factor for different water content at 50
Hz (1 mm sample, room temperature).
10
f(Hz)
10
10
10
73
M=1%
4
10
10
10
10
10
10
H"
H'
10
RT
30 C
40 C
50 C
60 C
70 C
80 C
10
-1
10
-2
10
-1
10
f(Hz)
10
10
10
H''
10
10
10
10
10
10
M=1%
RT
30 C
40 C
50 C
60 C
70 C
80 C
10
-1
10
f(Hz)
10
10
10
H'
10
10
10
dry
0.1%
0.2%
0.5%
1%
saturated
30
40
50
T(C)
60
70
30
40
50
T(C)
60
70
80
5. CONCLUSION
10 -2
10
-2
10
80
74
10
dry
0.1%
0.2%
0.5%
1%
saturated
ACKNOWLEDGMENTS
The authors wish to acknowledge Electricit de France
company (EDF) for its financial support to this study. One
author (P.H. Thinh) was supported by a grant from the French
Ministry of Education, which is also greatly acknowledged.
The authors wish to thank warmly Mr. H. Debruyne and A.
Petit (EDF) for their technical assistance and helpful advices,
and L. Saragossi for the building of experimental systems.
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
75
10
11
In this paper, the water concentration profile within the insulation of a stator bar in the
presence of a water leak is calculated by resolving the Ficks equation. The
temperature strongly activates diffusion processes, but the shapes of concentration
profiles remain similar. From the measured dielectric properties versus water content,
an electrical model is developed in order to calculate the field and potential distribution
within the insulation. The field is reduced in wet zones due to the high local permittivity
and conductivity compared to dry zones. This effect is favorable since the breakdown
field is lowered in wet zones. On the other hand, the field in dry zones is reinforced, but
the value attained remains lower than the short-term breakdown strength. This
explains why stator bars in the presence of waters leaks may survive for long times, in
spite of high local water concentrations in excess of 1%. The high sensitivity of
dielectric properties at low frequency to the presence of water suggests non-destructive
methods to detect wet bars in a real power generator.
Index Terms Epoxy composite, water absorption, rotating machine, dielectric
spectroscopy, diagnostic method.
1 INTRODUCTION
IN a previous paper [1], water uptake in glass-mica-epoxy
samples was studied, together with its influence on dielectric
properties.
In this paper, water absorption is calculated in conditions
close those existing in water-cooled bars, using data obtained
in [1]. Water concentration profiles are calculated for two
temperatures: at room temperature (corresponding to
conditions usually found during testing of generators), and at
60 C (representative of typical operating conditions). The
change of local dielectric properties versus water content are
known from the preceding study. When the water
concentration profile and local dielectric properties are
known, it is possible to calculate the field and potential
distribution by developing an electrical equivalent model of
the insulation. From these calculations, field distributions in
various conditions of temperature and frequency can be
obtained. A diagnostic method for the on-site detection of wet
bars is then proposed.
insulation
Copper strands
and cooling water
x(mm)
Authorized licensed use limited to: La Van Ut. Downloaded on February 23, 2009 at 01:37 from IEEE Xplore. Restrictions apply.
76
T. P. Hong et al.: Water Absorption in a Glass-mica-epoxy Composite II: Field Distribution and Diagnostic in a Stator Bar Geometry
w2m
wx 2
(1)
0<x<l
and
l=4mm
f 2m
mf
nSx
n 2S2 t
f
x mf
sin
exp D 2 , (2)
l
l
n 1 nS
l
U0
(3)
x x c t
77
12
13
with:
Hequivalent
Zi
10
Z total
10
0,0
0,5
1,0
M(%)
1,5
2,0
2,5
Uequivalent(:.mm)
10
14
10
13
10
12
10
11
10
10
10
10
60C
25C
10
0,0
0,5
1,0
M(%)
1,5
2,0
(6)
Zi
i 1
60C
25C
(5)
1
2
ZC i
R i2
2,5
4 FIELD CALCULATIONS
In a dry bar, the field within the insulation is constant
across its thickness. Once the water starts to leak into the
insulation, the water concentration distribution from wet to
dry sides is very inhomogeneous, as showed in Figures 1 and
2. The local dielectric properties depend upon water
concentration as illustrated in Figures 4 and 5. As a result, the
field distribution is strongly modified within the insulation.
To calculate the field distribution across the insulation of a
wet bar, we consider an equivalent electrical model in which
the insulation is divided into n discrete RC elements in series.
The Figure 6 shows the electrical scheme of the insulation
where V0 is the nominal voltage of the bar (copper strand
side) and Vn is the voltage of ground side (=0).
dry
(4)
78
10
14
T. P. Hong et al.: Water Absorption in a Glass-mica-epoxy Composite II: Field Distribution and Diagnostic in a Stator Bar Geometry
5 EFFECT OF FREQUENCY
Figure 10 shows the potential distribution at steady
state of diffusion (t=20 000 h at 60 C), corresponding to
6 different frequencies with an average field of 10 V/mm.
At this field, the data are available for all water contents
up to saturation [1]. The lower the frequency, the higher
the field distortion. When the frequency is less than 0.01
Hz, more than 90% of applied voltage is applied on 1/10
of the sample thickness close to outer layers.
dry
79
dry
7 CONCLUSION
Calculated water concentration profiles within the insulation at
room and service temperature in the presence of a water leak show
the same tendency, although water diffusion is considerably slower
at room temperature. The non-homogeneity of water concentration
leads to variable dielectric properties versus position. As a
consequence, the field distribution within the material is strongly
modified.
We have developed an equivalent electrical model by
considering the insulation as a network of RC parallel circuits in
series, in order to calculate the potential and field distribution. The
field distribution in the presence of water becomes frequencydependent. At 50 Hz, the field is enhanced in the dry zones of the
material, while it is reduced in wet zones. This arrangement is
favorable since the field is reduced in wet areas where the
breakdown field is reduced. At very low frequency, this tendency is
accentuated and the risk of breakdown in dry zones becomes high.
Based on these calculations, non-destructive methods to detect
wet bars in a real machine are proposed. These methods have been
tested successfully on full scale bars. Finally, calculations show that
the risk of breakdown is high when low frequency or dc high
voltage is applied on a wet bar.
ACKNOWLEDGMENTS
The authors wish to acknowledge Electricit de France company
(EDF) for its financial support to this study. One author (P.H.
Thinh) was supported by a grant from the French Ministry of
Education, which is also gratefully acknowledged. The authors
wish to thank warmly Mr. H. Debruyne and A. Petit (EDF) for
their technical assistance and helpful advices, and L. Saragossi for
the building of experimental systems.
Internal floating
electrode
REFERENCES
[1]
Copper strands
LF Generator
[2]
[3]
15
[5]
[6]
[7]
[8]
80
16
T. P. Hong et al.: Water Absorption in a Glass-mica-epoxy Composite II: Field Distribution and Diagnostic in a Stator Bar Geometry
81