Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/335194869

Lightning Impulse Withstand of Insulating Liquid in Terms of Unified Weak-


Link Theory

Conference Paper · June 2019


DOI: 10.1109/ICDL.2019.8796761

CITATIONS READS
3 66

1 author:

Mladen Marković
Končar D&ST
14 PUBLICATIONS 23 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mladen Marković on 31 January 2022.

The user has requested enhancement of the downloaded file.


Lightning Impulse Withstand of Insulating Liquid in
Terms of Unified Weak-Link Theory
Mladen Marković
Product Development Department
Končar D&ST Inc.
Zagreb, Croatia
mladen.markovic@koncar-dst.hr

Abstract—Paper investigates the applicability of newly pro- measure is mm2 . Value of SSA depends on electrode size and
posed weak-link model to lightning impulse breakdown measure- gap, and also on number of tubes of force n that exit the
ments for different electrode geometries. Applicability has been electrode surface. The number of tubes of force (represented
confirmed on both positive and negative polarity impulses, and
a functional relation of lightning impulse with AC breakdown as streamlines in a 2D axisymmetric coordinate system) can
withstand in terms of field non-uniformity has been established. be chosen arbitrarily, however, it has been shown [2] that for
n = 10 the model sufficiently represents electric flux regardless
Index Terms—streamline sweep area, impulse withstand, neg- of electrode configuration, even for highly non-uniform fields,
ative polarity, positive polarity, AC withstand, DIL factor, power because of the before-mentioned rule that the streamlines exit
transformer, mineral oil, insulating liquid, weak-link
the electrode where electric field is E ≥ 0.9Emax (and this
I. I NTRODUCTION is always at the tip of the electrode in non-uniform fields).
Therefore, the value n = 10 is used for numerical calculation
Insulating liquid breakdown can be observed from micro-
throughout this paper.
scopic and macroscopic point of view [1], [2]. Microscopic
Breakdown withstand of an electrode configuration is rep-
analysis concerns with electrical stresses between liquid and
resented as an average field stress ESSA along SSA. Field non-
electrode interface on molecular level, whereas macroscopic
uniformity is represented through factor µSSA (the higher the
analysis empirically describes how electrode and liquid char-
µSSA , the more non-uniform the field is). Finally, the break-
acteristics (e.g. number of particles in the liquid, water content,
down instance of a particular electrode configuration can  be
electrode size, electrode distance, etc.) affect breakdown. Em-
represented with a point T (x, y, z) = T SSA, ESSA , µSSA [3].
pirical influence of the size and the distance of the electrodes
It has been shown that AC breakdown instances of different
has been researched in terms of ‘weak-link’ methods, namely
electrode configurations (for uniform, quasi-uniform and non-
stressed liquid volume, stressed electrode area and cumulative 
uniform fields) defined as T SSA, ESSA , µSSA may be repre-
stress (Weidmann curves) [1]. It has been shown that these
sented in form of an ‘AC breakdown withstand plane’ in 3D
methods can be unified [2], [3] by a ‘weak-link’ quantification
logarithmic coordinate system (see [3], Fig. 48). The question
of a particular part of electric flux. Application of the unified
is whether such plane can be defined for LI breakdown
model has been shown for AC measurements, and this paper
measurements as well.
investigates if the model may be applied to lightning impulse
(LI) measurements as well. III. N UMERICAL CALCULATION
II. S TREAMLINE SWEEP AREA A. LI measurements
Unified weak-link model [2], [3] quantifies tubes of force LI measurements for 50 % probability breakdown of min-
that exit the electrode where electric stress has the greatest eral oil are taken from [4]–[18] with summarized electrode
value. The main concept of the model is that the imaginary sur- configuration in Table I. Extracted LI breakdown values are
face area (mm2 ) of tubes of force is proportional to number of shown on Fig. 1 and used as boundary values in numerical
imaginary ‘weak-links’ of an observed electrode configuration model.
(generally accepted ‘weak-link’ principle is that the greater the
B. Numerical model
number of ‘weak-links’, the higher the probability that one of
them will ‘break’, which is manifested as breakdown voltage Electrode configurations from Table I are modeled as 2D ax-
decrease). The surface area is calculated for tubes of force isymmetric FEM models. Model examples are shown on Fig.
that exit the part of electrode which ‘contains’ electric field 2, with electric field streamlines (which represent streamline
greater than 90 % of maximum electric field (E ≥ 0.9Emax ). sweep area) colored in red.
The term used for this type of surface area of tubes of C. Results
force is streamline sweep area (SSA), and the unit for this 
Calculated points T (x, y, z) = T SSA, ESSA , µSSA for break-
978-1-7281-1718-8/19/$31.00 ©2019 IEEE down instances from Fig. 1 for negative and positive polarity
TABLE I
E LECTRODE CONFIGURATIONS OF LI MEASUREMENTS

Measurement El.a Pol.b Rc gd


Liu & Wang [4] –| +, – 0.050 15...75
oo
Haegele et al. [5] ·o – 0.060...6.5 10...50
–o
Duy et al. [6],
Lesaint & Massala [7], –| +, – 0.100 20...300
Massala & Lesaint [8]
Trinh et al. [9]e B – 82.5...1279.5 2...150
Katim et al. [10] oo +, – 6.35 2.0; 3.8
Rozga et al. [11] ·o +, – 0.050 25
–|
Thien et al. [12]–[15] +, – 0.050...6.35 2.0...50
–o
–o
Denat et al. [16] +, – 0.300 10...70
–|
–|
R. Liu et al. [17]
–o
+, – 0.040 25 a) b) c)
Hosticka [18] ·| +, – 0.7, 1.6 1.3...76
a Electrodes: ‘–’ needle; ‘·’ point; ‘|’ plane; ‘o’ sphere; ‘B’ Bruce;
Fig. 2. 2D axisymmetric FEM models for different electrode geometries:
e.g. ‘oo’ is ‘sphere-sphere’, ‘·|’ is ‘point-plane’ etc. a) needle-plane b) needle-sphere (point-sphere) c) sphere-sphere. Electric field
b Polarity: ‘+’ positive, ‘–’ negative streamlines (colored in red) are swept around axis of symmetry and form
c Electrode radius (mm) ‘streamline sweep area’ or SSA (mm2 ).
d Electrode gap (mm)
e Measurements in [9] are assumed to be of negative polarity,
although this is not explicitly stated in [9]
Negative polarity

Negative polarity Positive polarity 103.0


Liu Liu
103.0 Duy 102.5 Duy
Katim Katim
2.5
103 103 10 Thien 102.0 Thien
Denat Denat
102.0 R. Liu
LI (+) breakdown voltage (kV)

R. Liu
LI (-) breakdown voltage (kV)

101.5

µSSA
Hosticka Hosticka
µSSA

101.5 Rozga Rozga


Haegele
Haegele 101.0
101.0 Trinh Trinh

10 0.5 100.5

2 Liu 2 0
10 Duy 10 10 100
Katim 107 107
Liu 6
Thien Lesaint 10 106
105

SSA
Denat Katim 105
R. Liu Thien
SS 104 104

(mm
Hosticka Denat A 103
(m 103
Rozga R. Liu m2 102 100
101 102

2)
Haegele Hosticka )
101 102 m) 101
Trinh Rozga
(kV/m 102 101 100
101 101 E SSA
100 101 102 103 100 101 102 103 ESSA (kV/mm)
Gap (mm) Gap (mm)

a) b) Fig. 3. Breakdown
 instances from Fig. 1a represented as points
T SSA, ESSA , µSSA , viewed from two different angles.
Fig. 1. LI breakdown voltages (mean values) for geometries given in Table I

are plotted in 3D coordinate system with logarithmic scale Positive polarity

and shown on Fig. 3 and 4 respectively. As visible from these Liu Liu
Lesaint
figures, LI breakdowns may be represented as planes, similarly 2.5
Katim
102.5 Lesaint
Katim
10 Thien Thien
to AC measurements in [3]. 102.0
Denat
R. Liu
102.0 Denat
R. Liu
Planes from Fig. 3 and 4 may be written in a form [19]: Hosticka Hosticka
µSSA

1.5 Rozga
10 101.5 Rozga
µSSA

101.0
Ax + By +Cz + D = 0 (1) 100.5
101.0

100 100.5
where 107
6
10
100
105
x = log SSA, (2) 104 1076
SSA

SS
A 103 10 5
(m 10 4
m2 102 100 10 3
y = log ESSA , (3)
(m m

1
) 10 10 2
101 102 10 1
2)

10 102 101 100


z = log µSSA (4) E SSA (kV/m
m)
ESSA (kV/mm)

and A, B, C and D are given in Table II. Table II also contains Fig. 4. Breakdown instances from Fig. 1b represented as points

coefficient values for AC measurements calculated in [19]. All T SSA, ESSA , µSSA , viewed from two different angles.
three planes are shown on Fig 5.
Negative polarity
TABLE II
VALUES OF COEFFICIENTS IN (1) DEPENDING ON TYPE OF MEASUREMENT

Measurement A B C D 12
LI(−) -0.3924 -2.077 -1 4.842
10
LI(+) -0.2081 -1.690 -1 3.612

DIL factor
ACa -0.2463 -1.247 -1 2.446 8
a calculated in [19]
6

3.0 AC 2
10 102.0 107
LI (-) 106
101.5 105

102.5 LI (+) 101.0 104


103
100.5 10 2
SSA (mm2 )
102.0 µSSA 100 101

a)
µSSA

101.5 Positive polarity

101.0

100.5 12
0 10
10
107

DIL factor
8
106 6
105 4
SS 104
2
A 10 3
107
(m 102.0
106
m2 102
100 101.5 105
) 101 101.0 104
101 102 103
m) 100.5
(kV /m
102
µSSA 100 101 SSA (mm2 )
E SSA
b)
Fig. 5. ‘Breakdown withstand planes’ for LI and AC measurements
Fig. 6. LI/AC ratio (DIL factor) depending on streamline sweep area (SSA)
and field non-uniformity (µSSA ) for a) negative and b) positive polarity of the
impulse
IV. A NALYSIS
Different values of coefficients A, B and D for LI(−) , LI(+)
and AC measurements from (1) in Table II should be compared will decrease 0.8/0.48 = 1.67 times ‘faster’ than LI(−) , which
in terms of gradients (i.e. slopes) of planes which they define. will lead to greater difference of AC and LI(−) withstand (in
For example, if (1) is written as kV mm−1 ) in highly non-uniform fields.
To qualitatively describe LI/AC withstand ratio vs. field
A C D
y(x, z) = − x − z − , (5) non-uniformity and SSA, the following function can be de-
B B B fined:
then for an arbitrary value of field non-uniformity (z = const.), yLI (x, z)
DIL (x, z) = , (10)
gradient yAC (x, z)
Φyx = ∂ y/∂ x = −A/B kV mm−3
 
(6)
where DIL is ’Design Insulation Level’ factor, yLI and yAC
defines ‘rate of change of electric withstand vs. streamline are calculated from (5) for LI and AC measurements, and
sweep area’; or, for a constant value of streamline sweep area x and z are streamline sweep area and field non-uniformity
(x = const.), gradient from (2) and (4), respectively. It should be noted that DIL
factor is traditionally calculated for voltages [20], and in
Φyz = ∂ y/∂ z = −C/B kV mm−1
 
(7) (10) it is calculated for average withstand fields, however,
defines ‘rate of change of electric withstand vs. field non- both calculations of DIL are mutually comparable because
uniformity’; etc. average field (kV mm−1 ) is nothing other than voltage (kV)
In case of arbitrary x = const., for LI(−) and AC measure- ‘normalized’ to 1 mm distance (which is reduced in DIL
ments, (7) is: calculation itself).
(−)
Visualization of function (10) for LI(−) and LI(+) measure-
ΦLI
yz = −0.48 kV mm−1 , (8) ments is given on Fig. 6. Positive polarity results shown on
ΦAC
yz = −0.80 kV mm . −1
(9) Fig. 6b should be taken with reserve, as LI(+) measurements
for SSA & 104 mm2 in uniform fields were not found in
(−)
The fact that ΦLIyz 6= ΦAC
yz means that the LI/AC withstand ra- available literature at the time of writing this paper (absence
tio differs for different values of field non-uniformity. Namely, of these measurements may significantly influence the gradient
for arbitrary increase of field non-uniformity, the AC withstand of the LI(+) plane). On the other hand, results from Fig. 6a
3.0
[4] Q. Liu and Z. Wang, “Streamer characteristic and breakdown in synthetic
and natural ester transformer liquids under standard lightning impulse
2.8 voltage,” IEEE Transactions on Dielectrics and Electrical Insulation,
vol. 18, no. 1, pp. 285–294, Feb. 2011.
DIL factor

2.6 [5] S. Haegele, F. Vahidi, S. Tenbohlen, K. Rapp, and A. Sbravati, “Light-


ning impulse withstand of natural ester liquid,” Energies, vol. 11, no. 8,
p. 1964, 2018.
2.4
[6] C. Duy, O. Lesaint, A. Denat, and N. Bonifaci, “Streamer propagation
and breakdown in natural ester at high voltage,” IEEE Transactions on
2.2 Dielectrics and Electrical Insulation, vol. 16, no. 6, pp. 1582–1594,
0 50 100 150 200 250 300 350 400 450 500 550 600 Dec. 2009.
Highest Voltage for equipment winding, Um (kV)
[7] O. Lesaint and G. Massala, “Positive streamer propagation in large
oil gaps: experimental characterization of propagation modes,” IEEE
Fig. 7. LI/AC ratio (DIL factor) calculated from Table 2 in IEC 60076-3 [21] Transactions on Dielectrics and Electrical Insulation, vol. 5, no. 3, pp.
360–370, Jun. 1998.
[8] G. Massala and O. Lesaint, “A comparison of negative and positive
indicate that DIL factor for LI(−) measurements ranges from streamers in mineral oil at large gaps,” Journal of Physics D: Applied
DIL ≈ 2.4 in uniform fields (µSSA ≈ 1) up to DIL ≈ 10.5 and Physics, vol. 34, no. 10, pp. 1525–1532, May 2001.
even more in highly non-uniform fields (µSSA & 100). [9] N. Trinh, C. Vincent, and J. Regis, “Statistical dielectric degradation
of large-volume oil-insulation,” IEEE Transactions on Power Apparatus
In terms of power transformer insulation design, this may and Systems, vol. PAS-101, no. 10, pp. 3712–3721, Oct. 1982.
indicate that bare electrode configurations with highly non- [10] N. Katim, M. Ishak, N. Mohamad Amin, M. Abdul Hamid, K. Amali
uniform fields (e.g. sharp edges on grounded parts like trans- Ahmad, and N. Azis, “Lightning breakdown voltage evaluation of
former core or tank) are more susceptible to breakdown when palm oil and coconut oil as transformer oil under quasi-uniform field
conditions,” Energies, vol. 11, no. 10, p. 2676, 2018.
exposed to ‘one-minute’ AC tests. Namely, IEC [21] defines
[11] P. Rozga, M. Stanek, and K. Rapp, “Lightning properties of selected
LI voltages for power transformers (Table 2 in [21]) with insulating synthetic esters and mineral oil in point-to-sphere electrode
corresponding DIL values in the range of 2.26 ≤ DIL ≤ 3.00 system,” IEEE Transactions on Dielectrics and Electrical Insulation,
(visualized on Fig. 7), which are associated with uniform and vol. 25, no. 5, pp. 1699–1705, 2018.
quasi-uniform fields (µSSA . 100.25 ≈ 1.8, Fig. 6a). Therefore, [12] Y. V. Thien, N. Azis, J. Jasni, M. Z. A. A. Kadir, R. Yunus, M. T.
if a power transformer is subjected to dielectric tests according Ishak, and Z. Yaakub, “Evaluation on the Lightning Breakdown Voltages
of Palm Oil and Coconut Oil under Non-Uniform Field at Small Gap
to IEC 60076-3 [21], its bare electrode parts with non- Distances,” Journal of Electrical Engineering and Technology, vol. 11,
uniform fields (µSSA & 1.8) should be dimensioned according no. 1, pp. 184–191, Jan. 2016.
to AC test voltages, since LI voltage levels (kV) do not [13] Y. Thien, N. Azis, J. Jasni, M. Ab Kadir, R. Yunus, M. Ishak, and
have sufficiently high values in order to be relevant in such Z. Yaakub, “The effect of polarity on the lightning breakdown voltages
of palm oil and coconut oil under a non-uniform field for transformers
case. It should be noted however that this is indicative only application,” Industrial Crops and Products, vol. 89, pp. 250–256, Oct.
for bare electrode configurations with solely insulating liquid 2016.
as an insulator in electrode gap, and not necessarily for [14] Y. V. Thien, N. Azis, J. Jasni, M. Z. A. A. Kadir, R. Yunus, M. K. M.
creepage breakdowns and electrodes with liquid-impregnated Jamil, and Z. Yaakub, “Pre-breakdown streamer propagation and break-
down characteristics of refined bleached and deodorized palm oil under
paper coatings. lightning impulse voltage,” IEEE Transactions on Dielectrics and Elec-
trical Insulation, vol. 25, no. 5, pp. 1614–1620, Oct. 2018.
V. C ONCLUSION
[15] Y. V. Thien, N. Azis, J. Jasni, R. Yunus, M. Z. A. A. Kadir, and
Calculated results indicate that lightning impulse (LI) break- Z. Yaakub, “Influence of Electrode Geometry on the Lightning Impulse
downs may be represented as ‘LI breakdown withstand plane’, Breakdown Voltage of Palm Oil,” in 2018 IEEE PES Asia-Pacific Power
and Energy Engineering Conference (APPEEC). IEEE, Oct. 2018, pp.
similarly to ‘AC withstand plane’ from previous research. 317–320.
Comparison of LI and AC breakdown planes indicates that [16] A. Denat, O. Lesaint, and F. McCluskey, “Breakdown of liquids in
uniform fields have lower values of LI/AC ratio than non- long gaps: influence of distance, impulse shape, liquid nature, and
uniform fields. For power transformers subjected to dielectric interpretation of measurements,” IEEE Transactions on Dielectrics and
Electrical Insulation, vol. 22, no. 5, pp. 2581–2591, Oct. 2015.
tests according to IEC standard, this could mean that parts of
the transformer with bare electrodes and non-uniform fields [17] R. Liu, C. Törnkvist, V. Chandramouli, O. Girlanda, and L. A. A.
Pettersson, “Geometry impact on streamer propagation in transformer
may be more susceptible to breakdown when exposed to AC insulation liquids,” in 2010 Annual Report Conference on Electrical
test voltages. Insulation and Dielectic Phenomena, Oct 2010, pp. 1–4.
[18] C. Hosticka, “Dependence of Uniform/Nonuniform Field Transformer
R EFERENCES Oil Breakdown on Oil Composition,” IEEE Transactions on Electrical
[1] A. Küchler, High Voltage Engineering, 1st ed. Berlin, Heidelberg: Insulation, vol. EI-14, no. 1, pp. 43–50, Feb. 1979.
Springer, 2018. [19] M. Marković, “Breakdown withstand of insulating liquid with paper
[2] M. Marković, “Unified method for determination of breakdown voltages insulated electrodes : A numerical approach,” in 2018 IEEE 2nd Inter-
of insulating liquid AC breakdown (in Croatian),” Ph.D. dissertation, national Conference on Dielectrics (ICD). IEEE, Jul. 2018, pp. 1–4.
Faculty of Electrical Engineering and Computing, University of Zagreb,
[20] S. V. Kulkarni and S. A. Khaparde, Transformer Engineering: Design
May 2017.
And Practice (Plant Engineering Series). Marcel Dekker Inc, 2004.
[3] M. Marković and Ž. Štih, “New method for prediction of insulating
liquid AC breakdown,” IEEE Transactions on Dielectrics and Electrical [21] IEC 60076-3:2013 Power transformers - Part 3: Insulation levels,
Insulation, vol. 24, no. 1, pp. 437–454, Feb. 2017. dielectric tests and external clearances in air, IEC Std., Jul. 2013.

View publication stats

You might also like