Professional Documents
Culture Documents
1, January 1981
431
E . Schopper
ELIN-Union AG, Weiz/Austria
G. Preininger
S. Wenger
HI
INTRODUCTION
Hi
High vltag terminal
lightning impulse wave and is used for the
computation Qf the so-called initial voltage
distribution . The inductive equivalent netcomwork diagram defines the response to
paratively slow voltage changes as in the tail
of impulse or switching surge waves#with it
the quasi steady-state voltage distribution,
is
i.e. the so-called final distribution
developed. The oscillatory equivalent circuit
network serves for the computation of the
transient voltage distribution.The parameters
(
for
this
inductances, mutual
network
inductances, capacitances ) can be determined
from the geometric dimensions of the trasfamer4.
Fig.1 shows the basic construction of a 500kV
455 MVA single-phase generator step-up transformer and in Fig. 2 for this transformer the
oscillatory equivalent circuit network as used
for the computation of the
internal voltages
in Fig.9, 11 and 16 is shown.
For the calculation of the inductances
and capaci ances and of the transient oscildigital computers normally
-are
lations 3*
used.
in
As variously shown, for instance
references3 and 5,the voltage u (t) appearing
at a free node of the equivalent network in
- 1 applied
response to a step voltage of U
to a terminal is described by:
a
80 SM 519-9
u(t)
Lst
a -
0018-9510/81/0100-0431$00.751981
k=l
bk.e
n =
.cos(w kt +..k)
(1
432
U
50
150
100
200
As
A * b'/b . damp-
0(t
iuJ
-t/TT
(e -teT
,,
It/TF
3
(3)
1
H2
H2
Tank
u(t)
-{bi.
cos( co
)]
(2)
.I
200
MIKROSEC
100
80
I
/u
60
4(t)
40
20
0
100
5a)
o0
100
200
[MIKROSEC
1o0
200
fMIKROSEC
433
The amplitude factor defined as v =
maximum voltage with any particular wave
shape/maximum voltage with square wave can
be plotted in diagrams Fig. 6 for various
f- T*/Tt
ratios of b/a over T'= 2 T*/T or
(T = Front Time,T, = Half value Time). The
diagrams are combined to show the factor v
to
separately as a function of Ct or T and
share a common frequency scale in the range
of 1 to 100 kHz relevant for transformers,
assuming a time to half valueT#T of 50 microseconds and a front time TF
F of 100 microseconds.
exp(c
114
0
- 1-_ Tos.EdS E .-f.gi ij ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~iij
VI
I .9
9 t
- I
8
.8
7
04
.2 1 IIII
tIS
1 101+11010
4'
luptilse wae
Naialue ti
wav
fiuls
00t )
cos (
(4)
u(t)
-.00
CcosCo1t)
.%
,^
(2.1 )
^%
0-1.
__,o
-2
2r/6
o(t)
-.01
.5
-00
e.-.-0100 1-01-.
>
.4
.3
.1
--.
.3
.4
.5
.6
.7
.8
.91
434
wave, b) the test connection and c)
voltage measured on the transformer.
U
100 %
BIL = 1675kV
the
100 %
BIL = 11675kV
100
80
9b)
HV
D0
TW
60
40
ILV
20
0
- I - I -_M_
.,
1dov
77-l
43Os
~~~~1
0bl)
__
K.
9c),....
e_-44
.I
4---
4-+--
.---
: ..L_
.++
, _
. .+.....
100V
_ __
1c_
ioov~
~~~~~~~~~HH
Fig 10: Voltages ceasured on a full size codel of the 500kV vindings of
transforeer according to Fig. 1: a) Test set.op, b) and c) test
caci legra.s
h.-
-~~
...
1,
N---
435
100
Hl
1 la2/5) ls
-BIT,
80
167-5ikv
HV
.60
40
TW
1 la)
20
0
Is
40
20
60
LV(
80
100
80
60
~~SS
139QkV
40
Df
20
0
1100
9
20
40
60
60
40
20
t
0s
16%
20
100
40
60
80
t ld)
-50
1* 0
18%
10
20
-100.
40
60
Fig. 12: Neasured(a) and calculated (b) resonant ovrvoltages across the tap
winding of transformr according to Fig. 1.
80
100
80
60
kHz
20
10 %
100
R=50
-R=0 I,VN
,R=no
50
40
20
10
switchi.ng surge
60
40
I
____
TW
100
Fast front
.1
80
20
20
li1c )
I
HV
60
2/5)Ls
80-
80
100
80.
ru
D/U 0
11lb)
. /
Eample of a switcb-ing
ove rvro1t age fovind in
~field tests
. lie)
-6
20
40
L
I
60
80
Fig. 11: Voltage D across tap winding of transforer according to Fig.1 for
various test voltage shapes and connections.
a) and b) current standard wave shapes and connections.
l/TL =
fL=
t =
Vw=
vW/4t
(5)
wave
propagation velocity
Vw
Vw
100
m/ls
m/,us
lines
300
for cables
for overhead
436
Fig.14a represents the basic connection
of the HV winding at the time of energizing
phase V first ( Connected line simulated by
300Q surge impedance ). Fig.14b compares the
voltage within the HV winding for various
exciting voltages.
_30
14a)
'HP-sine voltage
<
f = 10,5kHz
13b)
30* /
14b)
200 _-/
T_ooQ
~~~T300s2
lightning impulse
wave
\_
1 *,. t 21
1,2/50ps
switching impulse
Fig. 14 OvervOltage
X m 6.0.kz....
..*******
437
winding of the transformer according to Fig.1
as measured and as calculated for various
winding damping factors and constant damp ing for the exciting voltage. A comparison
of the oscillograms 17 c and f confirms the
strong damping of this particular winding
due to the location adjacent to the core.
225 . .
200
175
150
125
100
75
50
25
0
-25
-50
400
.,
300
200
100
0
-100.
-200.
-300
500
400
300
200
100
0
-2001
-300..
-400.
15d)
-500
438
250
200
UT gH1
HV
t7a)
150
100 -
TW
17b)
50
LV
MIKROSEC
-50
-1 00
50.
-50 50
0
17c) D (A = 05)
AA AA\ \-
oV
17d) 0
- V VlW
150
MIKR0SEC
(A =0.7)
P
-A
aLA
IA
VIVVVVV
_
A A~
V 158 [I-MIKROSEC
,-
,-,
REFERENCES
-50Q
100 .4
[1]
[2]
[3]
"Die Berechnung
der
Stofspannungsbeanspruchung von Transmittels
formatoren
Digitalrechner"
ELIN-Zeitschrift 19/1967, Heft
3 / 4
" Programmsysteme fur
Gerstl A.:
rotationssymmetrische Streufeld- und
Induktivitatenberechnung von Tr:ansformatoren" ELIN - Zeitschrift 1976,
Heft 2.
Mohseni H.:"Berechnung der elektrischen
Beanspruchungen bei StoBspannung und
Schaltuberspannung" ELIN - Zeitschrift
1976, Heft 2.
Bewley L.V.:"Travelling Waves onTransmission Systems"Chapman & Hall,London,
[4]
[5]
[6]
[7]
Fig. 17: Voltage 0 across tap winding as a functiso of the winding damping A
a) transferer according to Fig. 1, b) Caculated exciting voltage UT, c) throvgh a) calculated excited voltage 0, f) easured
voltages.
1974.
Preininger G.:
1951.
439
Richard
(SM
J.
Austria,
in
'70)
Vienna,
in
born
was
1928.
He is
has
associated
been
with ELIN-UNION
Vienna,
Austria, and since
1956
with the activities of ELIN
He has been
in the U.S.
primarily involved in the design,
testing
and installation of large electric power
equipment. He currently holds the position
of Senior Vice President of American ELIN
Corp.
Mr. Musil serves on ASTM Committee D 27
and the IEEE Transformer Committee.
Weiz,
AG.
Austria,
where
he
Schopper
was
born
in
University,
Technical
the
Graz,
Austria
in 1962 and
1970.
Since
with
Siegfried Wenger
was
born
Klagenfurt,
Austria, in
He
the
received
1948.
in
Ing.
Dipl.
degree
in
Electrical
Engineering
Graz,
He
Austria
joined
from
University,
Technical
the
in
the
1975.
High
volt-
Weiz,
His
main
are
research
Austria,
fields
overvoltages
ers
of
in
1975.
activity
on
transient
transformin
Discussion
appreciated, namely:
1. The significant damping role played by external lines or loads
connected to windings other than that one receiving the transient
voltage excitation.
2. The fact that switching transients tend to excite relatively low
natural frequencies within the transformer windings, involving a
large percentage of winding turns (usually 20% or more).
3. The internal response to oscillatory voltages can be greater than that
for any form of aperiodic voltage. As a consequence of this fact,
windings can be overstressed by transient voltages below the arrester protective level if they are oscillatory in nature.
This last point is of extreme significance to a Working Group of the
IEEE Transformers Committee which has been investigating internal
winding resonance effects with respect to switching overvoltages. The
work to date has concentrated on fast front, long tail pulse voltages, but
it is now evident that oscillatory voltages must also be considered. Such
applied voltages are obviously not represented by present dielectric test
waves, nor are they representable. If suitable oscillatory voltage tests
could be devised, it would be technically possible to design transformers
which would be totally immune to applied waves of any frequency, but
it probably would not be economically desirable to do so. Our speculation on why this form of switching voltage has not produced more service failures is that it requires a very unique set of conditions to generate
an applied voltage exactly in tune with an internal winding resonance in
a region of the winding where there is dielectric sensitivity. We would
also invite the authors' speculation on this point.
440
One final question relates to the fashion in which the authors derive
their internal damping factor, A, which should be a function of frequency. Is this derived analytically or emperically?
We wish to commend the authors for a fine contribution to the
technical literature and we encourage their continued participation in
the Working Group activities aimed at introducing recognition of these
effects in industry standards.
Manuscript received July 31, 1980.
H. B. Margolis and A. J. McElroy (American Electric Power Service
Corporation, NY): In 1972 our colleagues and we described a number
of transformer failures experienced on the AEP system attributed to
part winding resonance. Since then a working group in the IEEE
Transformer Committee has been engaged in studying the
phenomenon.
The members of this working group are generally agreed that transients with fast fronts (5-40 us) and sustained tails can and do occur in
service and can give rise to part winding resonance which may damage
the transformer. The objective of the working group is to prescribe a
dielectric test to forestall field failures. This requires specifying the exact form of the wave to be imposed, its magnitudes and the conditions
at the terminals. The basis for such a test must be an understanding of
system behavior and the transformer's response, which is frequency
dependent.
The authors have made a convincing case for the need to deal with
this phenomenon so as to enhance the reliability of large transformers
whose natural frequencies often fall into the critical area where they can
resonate with the system. The authors' analyses and tests confirm
notable aspects which we have encountered in the 6 failures on the AEP
system attributed to part winding resonance.
1. Five of the failures occurred coincident with the line faults which
can produce fast front long tail switching surges; while one failure
occurred on back energizing a generator step-up transformer.
2. No arrester operations occurred with any of these failures which
indicated that the voltage was below the sparkover level of the arrester.
3. The failures occurred at tap leads or in the winding some distance
from the line leads.
4. Subsequent to the failure of 3 units, surge arresters were installed
through an added bushing and connected to a crossover connection close to the center of the common winding of the 200-MVA,
500/345-kV autotransformer. In the 5 years that elapsed since the
arresters were added, no failures have occurred while 6 arrester
operations were counted at the 500-kV terminals, zero operations
at the 345-kV terminals and 43 operations on the crossover ar-
resters.
5. On 3 bands which were suspect after one of the transformers failed, the solution was to add a capacitor bank to the tertiary winding increased the damping characteristic. This has proven effective to date.
The authors' investigations explain many of the conditions noted
above, viz:
1. The overvoltages generated by a fast front wave with a sustained
tail (Figure 6 and Ild).
2. The vulnerability of the tap winding in particular (Figure lId).
3. The failures at midwinding and at crossover connections (Figure
10).
4. The influence of terminal connections in establishing the damaping characteristics (Figure 11) which, in turn, determines the selection of the overvoltage factor (Figure 8).
To understand the factors involved, AEP has undertaken transmission
line model studies; sponsored factory tests; developed, with the
assistance of transformer designers, mathematical and physical
transformer models; and made staged fault field tests:
1. Transmission line model studies have shown that at critical frequencis significant changes in voltage can be generated directly on
the faulted phase and indirectly through coupling, on the unfaulted phases. The location of the fault with respect to the
transformer is critical.
2. Factory tests have confirmed the existence of amplification factors as high as 37:1 across parts of a winding at its resonant frequency.
3. Factory tests have further confirmed the critical role of
transformer termination i.e. grounded, ungrounded or terminated in surge impedance in establishing the damping
characteristic of the transformer.
R. J. Musil and G. Preininger: It is gratifying to note that our paper attracted the attention of such distinguished discussors, which proves that
the subject of internal "resonant" overvoltages is of considerable current interest.
We appreciate the comments of Mr. Margolis and Mr. McElroy
because of their interest in and contributions to the investigation of
overvoltage phenomena. A principal purpose of writing this paper was
to encourage further system studies and tests in order to obtain conclusive data on actual transients which may form the basis for revised
dielectric test procedures. As manufacturers we can pretty well establish
the response of our transformers to various kinds of overvoltages but
cannot determine the reality or probability of such events. The effect of
the double grounded LV winding in Fig. 1 la is not considered damping
in the strictest sense but a change in the inductive coupling and thus a
shift in the steady-state distribution.
We realize that rapid changes in transformer terminal voltage similar
to a square wave can occur. If such steep-front waves do indeed approach twice normal crest line to ground voltage, this may lead to a fast
front switching surge test at about the level of the traditional slow rise
wave.
However, we would like to caution that a drastic reduction in front
time should be carefully considered and imposed only on the basis of
compelling evidence, as it may require on occasion more insulation and
thus a more expensive product.
Figure 4 and 5 show in an idealized form the response of a typical
oscillatory circuit to various exciting wave shapes. They do not represent voltages at a particular point of the transformers discussed in the
paper.
We sympathize with the discussors concern about the validity of full
wave tests with tails shorter than in the standards. Even with high
capacity impulse generators we often get shorter tails particularly for
high current windings. Hopefully, such a low surge impedance winding
441
will also cause a shorter wave tail for a system generated surge.
We generally agree with the comments and observations of Mr.
Phillips with some exceptions.
A full wave applied to the HV terminal of the transformer in Fig. 11
would not have overstressed the LV winding if open-circuited. This
condition - not unrealistic in service - was considered in the design. The
transformer in Fig, 14 was installed in an ungrounded system.
Therefore, de-tuning by grounding the neutral - as was quite correctly
observed - was not feasible.
The relationship between the stresses imposed by the single oscillation
peak of full wave and by the multiple though lower peaks of a resonant
condition is not too well defined and the solution to the questions of
equivalency elusive. As much as we would like to stay with established
procedures, it does not appear to be readily possible to simulate all resonant voltage stresses by the application of the present package of tests;
even with a higher BIL which would be undesirable because of the unnecessary increase in transformer cost. The user normally is not in the