Page 1 of 17

Analysis of Integral Snubber Circuit Design for

Transformers in Urban High Rise Office Building
Peter E. Sutherland, Ph.D. , PE, Fellow, IEEE
GE Energy Services
Industrial Solutions
1 80 Rotterdam Industrial Park
Schenectady, NY 1 2306
Peter. Sutherland @ ieee.org

Abstract- Transformer failures have in recent years transient or the transient recovery voltage (TRV) of
led to the development of Resistor-Capacitor snubber a switching device, is within the capabilities of the
circuits for the protection of the transformer and transformer insulation system to withstand. This
winding insulation from the damaging effects of high­ insulation capability is defined as the Basic
voltage high-frequency transients. Transformer Insulating Level, ElL, and is part of the
insulation may be damaged if the Basic Insulation manufacturer's specifications for the transformer,
Level (BIL) is exceeded, turn-to-turn insulation when
and the subject of factory tests. For these
there is excessive rate of change of voltage with time
transformers, typical transient voltages may be 1 0
(dv/dt), and to switching devices by restrikes when
kV higher than the system voltage peak, thus, with
the Transient Recovery Voltage (TRV) is exceeded.
a 1 3 . 8 kV system, at a possible 5 % overvoltage, the
These transients are most often observed when dry­
peak line-ground transient voltage would be
type transformers are close coupled to vacuum
switching devices. Some manufacturers are now
approximately 1 3 . 8 x 1 .05 x + 1 0 = 2 1 . 8 kV. This
including snubbers in their transformer designs.
is well below the 95 kV ElL. The transient
This paper provides a thorough review of the causes overvoltages caused by switching events may,
of the transients, methods of analysis, and mitigation however, be much higher, and may damage the
of the effects of these transients. An example is transformer insulation.
provided of transformers to be installed in the
The problem of switching transients, primarily from
basement of an urban high rise office building, where
vacuum interrupters, damaging dry-type (and other)
the space is limited and the available fault current is
high, where the transformer enclosure includes built­
transformers with low insulation strength has been
in snubber circuits. The strengths and weaknesses of
known since the invention of vacuum interrupters,
current methods are examined. Recommendations as described in the late Professor Greenwood's
are made for improvements in snubber circuit design book Vacuum Switchgear [3] . The analysis of the
and analysis. transient overvoltages which have resulted in the
application of snubber circuits has been given in
Index Terms-Snubbers, Transformer snubber detail in Greenwood's classical textbook Electrical
circuits, Transformer protection, Transient analysis, Transients in Power Systems [4] . The ailed analysis
Transient overvoltages, Transient rate of rise of of the internal windings and insulation systems of
voltage, Transformer failure. transformers is generally only possible with
information on the construction of the transformer
I. INTRODUCTION
which is usually manufacturer's proprietary
The purpose of a snubber study is to determine information. This makes some of the techniques
the need for and ratings of snubbers necessary to discussed in the draft IEEE standard and in
protect the subject transformers from over-voltages Greenwood's Transients book difficult to apply.
due to switching transients. [ 1 ] Snubber studies are
One technique which has been successfully applied
often conducted in accordance with the
is to perform "black-box" sweep frequency
recommendations of the draft standard IEEE
impedance measurements of transformers and
PC57. 142 [2]
cables [5] [6] . Measurements over a range of 50 Hz
When a transformer is subjected to high-frequency to 2 MHz were performed on a de-energized
voltages from switching transients, there is a transformer using a network analyzer, which
possibility that these frequencies may excite a provided a frequency sweep and .
resonance in the transformer windings. This
A snubber study was performed for four 2500/3333
resonance may cause a high turn-turn voltage to be
kVA 1 3 . 8 kV - 0.48 kV transformers at a
developed internally to the transformer. This may
Manhattan office tower. The purpose of the
be true even in the situation where the transient
Snubber Study was to evaluate the design of the
voltage itself, whether from a cable energization
integral snubber circuit as shown in the drawings

$
978-1-4673-0651-5/12/ 31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
 Page 2 of 17

for the transformers. The study was performed for UTILITY

�ABLEC1
a typical transformer, with the longest cable length,
but is applicable to all fou.r transformers.

[--��L���-�:�M���---'
II. SYSTEM DATA
A. System Configuration

CABLEC2
A simplified one-line diagram is shown in
Figure 1 . Four three-phase 750 kcmil utility
service cables (Cable Cl), of which one is shown,
enter the first basement (B 1) of the building
through a set of steel conduits . Out of the four
cables, Cl, two each enter one of two Splice
Chambers, of which one is shown. In the Splice
Chambers, each single 750 kcmil conductor is
spliced to a pair of 750 kcmil conductors (Cable
C2), so that four (4) cables leave each Splice
Chamber. These leave the Splice Chambers in
paired steel conduits which pass down to the third
basement (B3) level. The Cables C2 enter the
service equipment vault, where there is a bank of
vacuum switchgear units called primary
CABLEC3
TRANSI
BOXTION
distribution centers (PDCs).
The 1 3 . 8 kV, 1 200 A PDCs have a main service
breaker (52-S) and four feeder breakers (52-Fl
through 52-F4). Electrical power from the PDCs is
CABLEC4
SPOT:RK n
then fed (Cable C3) from the four feeder breakers

NETW
to Transition Boxes. From here cables (Cable C4)
go to an array of Spot Networks in basement level
2 (B2) and electrical rooms in upper floors . Each
Spot Network contains four (4) 2500/3325 kVA

LL 2500KVA
transformers. The Cables C4 are connected to the
2500/3325 kVA 1 3 . 8-277Y1480V spot network
transformer primary windings through fused
grounding switches. The spot network transformer
u
-i 'r 13.8KV-480V
secondary windings are connected together through

�
4000 A Network Protectors and Network Fuses at
an 8000 A Spot Network Bus. However, this paper
is only concerned with the system up to the

8000A BUS
primary windings of the Spot Network
transformers. ------------- ---------------------

B. Utility Data
I
The Utility contribution for three-phase, first­ Figure 1 Simplified one line diagram.
cycle symmetrical short-circuits is used in this
study, shown in Table 1 . There are eight 1 3 . 8 kV
services, divided up between the two Splice
Chambers, any one of which could be used to Table 113.8 kV Utility Short Circuit Data

supply the load in question, so the average value of

I
Spl Util Fault R X L Z
fault current is used. The inductance of the Ch Svc IA) In) In) ImH) In)
reactive component is calculated for use in the Average 31,553 0.0206 0.2450 0.6499 0.2459
resonance calculations.

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
Page 3 of 17

C. Cable Data required, as in cable C2, then parallel conduits with

identical sets of cables are used.
All cables are 1 3 . 8 kV, 20 ft. long, EPR If three-conductor cables are used, the three
insulation, with copper conductors. They are in phase conductors (along with any ground or neutral
steel conduits, except for the interlocked armor conductors which may be present) are contained
cables, which are steel. All are one conductor per within a single outer protective jacket; the cable is
phase, except for C2 which is two parallel designated "3/C."
conductors per phase. The 60 Hz impedance data is Both "lIC" and "3/C" cables are normally
calculated using a commercially available power required to be installed in a conduit or raceway.
systems analysis program. The cables shown in the The material used, and its magnetic properties
one-line diagram are listed in Table 2. affect both the resistance and inductance of the
cables within. Cables built with an integral conduit
Table 2 13.8 k V Cable Data
of flexible material wound around the insulated
10 Type Cond/ Size Length R 60Hz Z inner conductors are called "Interlocked Armor" or
cjI (kcmil) (m) (n) 60Hz
IA cables. A third letter designates the material of
(n)
the armor, such as "lAS" for steel and "IAA" for
C1 l/ C 1 750 6.096 0.000389 0.000944
C2 l/ C 2 750 6.096 0.000195 0.000472
aluminum. An insulating jacket normally covers
C3 l/ C 1 500 6.096 0.000535 0.000996 the armor, on which the conductor size and other
C4 lAS 1 500 6.096 0.000535 0.000747 information may be marked.
For the purposes of this simulation, the cables
Medium-voltage single-conductor power cables from the utility to the vacuum circuit breakers in the
are constructed in a series of layers, as shown in PDC will be modeled at 60 Hz and lumped into the
Figure 2. source impedance. The results are listed in Table 3 .
The cables from the PDC t o the transformers will
be modeled for high frequency transients.

1. Conductor Table 3 Source Impedance Values

2. Extruded semiconducting Element R(n) X(n) L(mH) Z(n)

Utility 0.02056 0.24500 0.64987 0.24586
conductor shield
Cl 0.00039 0.00094 0.00250 0.00102
3. Insulation (such as EPR) C2 0.00019 0.00047 0.00125 0.00051

Total 0.02115 0.24642 0.65362 0.24732

4. Extruded semiconducting

insulation shield Cable constants are calculated in a different

manner than for power frequency operation when
5. Copper tape shield
cables are being evaluated for the effects of high­
6. Protective jacket
frequency transients. High frequency travels
through a power cable in the same manner as it
travels through any other coaxial cable. The return
current from each central conductor flows through
Figure 2 Single conductor medium voltage cable the shield.
construction.
a) Cable Resistance
The cable type "lIC" used for cables Cl, C2 and The resistance values are the dc resistance of the
C3 represents three single-conductor cables for a conductor adjusted for skin effect at the frequency
three-phase three-wire circuit. Each phase consists of the transients. [7] Resistance per unit length for a
of a separate and complete single-conductor cable conductor with a circular cross-section can be
including a protective jacket. The three cables calculated in the ideal case as :
(along with any ground or neutral conductors which
p I n
may be present) are pulled together through a single Tdcw = A = arrr2
w
� (1)
conduit. If multiple conductors per phase are
Where :

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
 Page 4 of 17

rdew is the dc resistance of the conductor in Q/m, 1E+07 2.06E-OS

2
A is the cross-sectional area of the conductor in m ,
rw is the radius of the conductor in m The frequency for a given skin depth is
p = 1 .673 X 1 0-8 Q·m is the resistivity of copper at
f= 1 Hz
8(f)21f/lOIT
(3)
20C.
cr = lip = 5 .977 X 1 0
7 Slm is the conductivity of
The exact formula for the ratio of rae conductor ac
copper at 20C.
resistance to rdc conductor dc resistance is a solution
The appropriate calculated or table value of dc
for the diffusion equation involving the use of
resistance, rdew, should be used for the particular
Bessel functions, which is beyond the scope of this
cable in use, taking into account such factors as
paper. However, the following approximation may
type of copper, stranding and coating. [8] Ambient
be made:
temperature correction factors should be applied as
necessary.
The skin depth is: (4)

oCt) = J �
1tf OIT m (1)
For the second case, where the skin effect is
Where: operative, the ac resistance is a function of the
Ii(f) is the skin depth, in m square root of the frequency. The factor ran can ;>!
f = frequency, in Hz, be used to calculate the ac resistance:
7
flo = 41t X 1 0- HIm i s the permeability o f free space.
racw _ rw.jff/lOIT
The skin depth of a conductor is inversely - rdc
w 2 �or r
w
> 20 (5)
11
proportional to the square root of frequency. For a
copper conductor: The breakpoint in equation (5) occurs when the
radius of the conductor equals twice the skin depth:
oct) = 0.0655
m (2)
11 rw = USC!) m, (7)
as shown in Figure 3 and Table 4 . Combining (2) and (7), the conductor may be said
to be at the skin depth transition frequency:
1.E-Ol
4
F = -- Hz •
Jtw 1',21[1l (j (8)
1.£-02 ....... .......
W ,.. 0

E
:;;
"
�I'-.. The skin depth transition frequency for copper
! 1.£-03 conductors with a circular cross-section is:
�
1.£-04
�....... F
Jtw
- 0.01695 HZ
- 2 (9)
rw
'� Skin depth transition frequencies, dc resistance and
1.£-05
lEtOI lEt02 tE+03 lEt04 tEfOS lEt06 lEt07 >!
the factor rae,; are shown in Table 5 for some
Frequency (HI)
common sizes of copper conductors.
Figure 3 Skin depth, B, of a copper conductor This approximation is not used in calculating
Table 4 Skin depth, B, of a copper conductor the power frequency resistance of conductors for
power system studies such as short circuit and load
Frequency.! Skin Depth, B Frequency, f Skin Depth, B
flow analysis. However, it can be used for transient
(Hz) (m) (Hz) ( m)
1E+01 2.06E-02 1E+04 6.S1E-04
studies above the transition frequency.
2E+01 1.46E-02 2E+04 4.60E-04 For snubber studies, the frequency of interest
SE+01 9.21E-03 SE+04 2.91E-04 may vary between 10 kHz and several MHz,
1E+02 6.S1E-03 1E+OS 2.06E-04 depending upon the mode of the oscillation. Once
2E+02 4.60E-03 2E+OS 1.46E-04 the frequency of oscillation is determined, then the
SE+02 2.91E-03 SE+OS 9.21E-OS resistance for the simulation may be calculated
1E+03 2.06E-03 1E+06 6.S1E-OS
using (5). This is plotted in Figure 4 (a) .
2E+03 1.46E-03 2E+06 4.60E-OS
SE+03 9.21E-04 SE+06 2.91E-OS

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
Page 5 of 17

Table 5 Resistance of copper conductors.

( 1 0)
Size
(AWGI 'w flw 'dew 'acwl";' Where:
(m) (Hz) nlm nlm
Kcmil) T, is the average radius of the shield, in m

2 0.0036 1311.7 5.31E-04 1.47E-05 ts is the thickness of the shield, in m

1 0.0041 1013.4 4.23E-04 1.33E-05 For higher frequencies, current flows only within
1/0 0.0046 802.9 3.35E-04 1.18E-05 the inner surface of the shield. The dc resistance
2/0 0.0051 640.4 2.66E-04 1.05E-05 equals the ac resistance for all frequencies below
3/0 0.0058 505.7 2.11E-04 9.38E-06 where the skin depth equals the thickness of the
4/0 0.0065 401.2 1.67E-04 8.34E-06 shield:
250 0.0071 337.7 1.41E-04 7.67E-06
350 0.0084 240.2 1.01E-04 6.52E-06 ts = ocn m, (1 1)
500 0.0100 168.3 7.08E-05 5.46E-06 The skin depth transition frequency for the shield is:
750 0.0123 112.1 4.72E-05 4.46E-06
F
l
1000 0.0142 84.2 3.54E-05 3.86E-06 Jts = ts2rr/1 0 (J
__Hz ( 1 2)

l.E+OO The skin depth transitIOn frequency for copper

shield conductors is:
l.E.Ql

0.004238
F = HZ
E >Of . Jts t2
.
' ( 1 3)
1.E.Q2
s
£
�
e
For a typical copper shield of 5 mils thickness or
1;;
� 1.E.Q3

.� ""...l'� '!'-,' 0 . 1 27 mm, its = 263 kHz. When the transition to ac

'" .................. :.:.3'.,- , �.::- .
. .:, .�
1.E-04
resistance is made, the resistance changes from
1.E.Q5 +-�""4��+-�""""'��""I-��+-�4
being fixed by the thickness of the shield to varying
1E+01 1£+02 1E+03 1E+04

Frequency (Hz)
1E+05 1E+06 1E+07
with the skin depth by /1:
- . -750

{
······· 250 ----350 - - 500
rdes if ts < 0 n
raes=
rdes ts y�
( 14)
m
fJrJioa if ts >
-Shield 250 -Shield 350-Shield SOO_Shield 750
0
(a) Resistance of conductors and shields
1.E+OO
Then:
l.E.Ql racs
11 = rdestsyJr JioaI;-::; . f ts >
r;;:-; J:
u (15)
....""'101
�
�
1.E.Q2

w"'�'11 ...-:10-.... ...�-'L....... r"�'''': ""....�...,

..,.
The factor raes /.J7
which can be used to calculate
e

1l
1.E-03

.�
the ac resistance is shown in Table 6 for the shields
'"
1.E-04
of some common sizes of copper conductors.
1.E-05 Table 6 Resistance o f 5 mil (0.127 mm) copper shield.
lE+Ol lE+02 1E+03 lE+04 lE+05 lE+06 1E+07

Frequency (Hz)
- - -
Size
······· • rs fls rdcs Tacs/fl
(AWGI
250 ----350 500 -750

(m) (Hz) nlm nlm

kcmil)
(b) Conductor plus shield resistance
2 0.01051 2.63E+05 1.98E-03 3.87E-06
Figure 4 Resistance vs frequency of conductors and shields 1 0.01102 2.63E+05 1.89E-03 3.69E-06
1/0 0.01153 2.63E+05 1.81E-03 3.53E-06
for several common cable sizes (in kcmil)

At high frequencies, the cable acts as a coaxial 2/0 0.01204 2.63E+05 1.73E-03 3.38E-06
cable and the return current flows through the 3/0 0.01267 2.63E+05 1.65E-03 3.21E-06
shield. The currents being considered are 4/0 0.01337 2.63E+05 1.56E-03 3.05E-06
capacitive ground currents through the cable and 250 0.01407 2.63E+05 1.48E-03 2.90E-06
transformer capacitances, and not the load currents, 350 0.01540 2.63E+05 1.36E-03 2.65E-06
which are phase to phase and do not involve 500 0.01699 2.63E+05 1.23E-03 2.40E-06
ground. If the shield consists of a conducting tape 750 0.01937 2.63E+05 1.08E-03 2.10E-06
of fixed thickness, its dc resistance may be 1000 0.02121 2.63E+05 9.86E-04 1.92E-06

calculated:

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
 Page 6 of 17

I.E-05
The shield resistances are plotted in Figure 4(a),
and are seen to be constant up to the shield skin l.E-06 .p._-I__"__"_�__"_...-j
depth transitIOn frequency. The conductor
resistance goes to the load, and the shield resistance
is the ground return, so that the total resistance is
the sum of the two. The total resistance is plotted in
Figure 4(b), and is dominated by the shield
resistance.
1.E·ll +-����--I-��4---�....-j����..j

Frequencv(Hz)
b) Cable Inductance lE+Ol lE+02 lE+03 lE+04 lE+05 lE+06 lE+07

- - 500lnl - -7501nl
-500E)(1 _7SOExt
Inductance is the sum of the internal inductance ·•·· .. ·2SOlnt ----350lnl

.. ---- 350Shld
- - 500Shld - -7SOshld
-250E)(1 -350Et K
·•·
of the wire, the external inductance between the ··2SOShld

(a) Internal and external inductance of wire and internal

wire and the shield, and the internal inductance of
inductance of shield.
the shield. For the capacitive ground currents being 2.0E-06

considered, the external inductance, In is calculated

+-----+--+-----1
from the dimensions of the cable:
-
1.5E-06
.............. ....... . ............. .... ......... . ...... ......... ...............
[ = J.lo [n (rs) �
------ - ----- ------ ---
. . . . . .

- ---- -------
.

f --------
-

-- -
-

--------
--- --

e 2n: rw m
( 1 6) ��� � - -- -- f------
For normal phase currents, the distance between
I 5.0E-07 +-----+--+-----1
conductors would be substituted for the radius of
the shield. The dc internal inductance of the wire O.DE+DO +-����--I-��+-����....j

Frequency (HZ)
1.0E+01 1.0E+02 1.0E+03 1.0E+04 1.0E+OS 1.0E+06 1.0E+07
for low frequencies is constant:
_
[dew - J.lo � ·······250 ----350 - - 500 - -750
( 1 7)
Sn: m (b) Total internal and external inductance of wire and
shield.
At higher frequencies, the internal inductance liw
has characteristics similar to resistance, as shown in Figure 5 Internal and external inductance vs frequency for
several common cable sizes (in kcmil)
Figure 5 (a), exhibiting a skin effect also. [3]
[dew if rw < 28 The third inductance in the cable is that of the

I
H
[.tw "'" [ 1 1 - internal inductance of the shield. Paul [9] has
- dew � ( 1 8)
2rw n:J.lo(J v17t if rw > 28 m
given the calculation of this inductance by both the
energy method and the method of flux linkages.
The internal inductance is generally a small The shield transitIOn frequency, its, was
percentage of the external inductance, and is often calculated using ( 1 2) to be 263 kHz. For
ignored. frequencies above it" the effective thickness of the
shield goes from ts to is.
The internal dc inductance and the internal ac
inductance of the shield above it" are approximated
as :

( 1 9)

(20)

Below the transition frequency, where the skin inductance is plotted in Figure 5, and some of the
depth is greater than the thickness of the shield, factors used in the calculations are listed in Error!
current distribution in the shield is uniform, and the Not a valid bookmark self-reference.. Because
dc shield inductance may be used. The shield the internal inductances of the wire and shield are

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
Page 7 of 17

100
over an order of magnitude smaller than the
90 t----+--�--_+--+_--�
external inductance of the wire, le in equation ( 1 6), 80 ............ ............ ............ ............ ............ ............
-----
that can be used as a good approximation for the £ 70
-----
- ----- ------ ----- ------
� ------ ---1-------
cable inductance. -ll ...... ---- r- ------
60

x.
-
50

� 40 t----+--�--_+--+_--�
l t == l e = 110 In 2i.:!. r T----+----�--r_--_+--+_--�
30
(2 1 )
2rr Tw m 20 T----+----�--r_--_+--+_--�
10 t----+----�--_+--+_--�

lE+Ol lE+02 1E+03 lE+04 1E+05 lE+o6 lE+07

Frequency (Hz)
Table 7 Inductance of 5 mil (0.127 mm) copper shield.
...... ·250 ---·350 - - 500 - -750

Size
rs rs+ts fts Ides
(AWGI 6
(m) (m) (Hz) Him
Figure Surge impedance vs frequency of for several
kcmil) common cable sizes (in kcmil).

2 0.0105 0.0106 2.628E+05 8.056E-l0 d) Surge impedance

1 0.0110 0.0111 2.628E+05 7.686E-l0
The surge impedance of a cable is defined as :
1/0 0.0115 0.0117 2.628E+05 7.346E-l0
2/0 0.0120 0.0122 2.628E+05 7.035E-l0
3/0 0.0127 0.0128 2.628E+05 6.682E-l0 ( 1 9)
4/0 0.0134 0.0135 2.628E+05 6.335E-l0
250 0.0141 0.0142 2.628E+05 6.020E-l0
350 0.0154 0.0155 2.628E+05 5.498E-l0
Cable resistance, inductance, capacitance and
500 0.0170 0.0171 2.628E+05 4.985E-l0
surge impedance at 20 kHz are shown in Table 9.
750 0.0194 0.0195 2.628E+05 4.371E-l0
Calculated values for the combined series cables
1000 0.0212 0.0213 2.628E+05 3.992E-l0 C3+C4 are shown in Table 10, with the equivalent
circuit in Figure 7 .
Table 9 Typical Cable Resistance, Inductance, Capacitance
and Surge Impedance at 20 kHz .
c) Cable Capacitance
Capacitance, c, is calculated as: Size
Zo
27r£0£, (AWGI R(!l/m) L(H/m) C(F/m)
c= Fl m (2 1 )
Kcmil) (!l)
In rJrw
2 4.068E-03 2.130E-06 1.296E-l0 128.2
Where
12 1 3.783E-03 1.967E-06 1.404E-l0 118.4
EO = 8 . 854x1O- F/m
1/0 3.491E-03 1.825E-06 1.512E-l0 109.9
E f (unitless) i s the relative permittivity (dielectric
2/0 3.229E-03 1.686E-06 1.637E-l0 101.5
constant) of the insulation. Values of Ef for some 3/0 2.982E-03 1.554E-06 1.776E-l0 93.5
typical cable insulating materials are shown in 4/0 2.748E-03 1.430E-06 1.929E-l0 86.1
Table 8. More detailed simulations will also 250 2.576E-03 1.360E-06 2.028E-l0 81.9
include the permittivity and conductivity of the 350 2.283E-03 1.202E-06 2.295E-l0 72.4
semiconducting shields. 500 2.006E-03 1.043E-06 2.643E-l0 62.8
Table 8 Dielectric constants o f some cable insulation
750 1.713E-03 9.010E-07 3.060E-l0 54.3
materials [11] 1000 1.534E-03 7.973E-07 3.457E-l0 48.0

Material Description Maximum Dielectric

Service Constant Table 10 Cable Resistance, Inductive Reactance expressed
Temperature as 60 Hz, Capacitance at 20 kHz for 500 kcmil cable.
(0C)
PVC Polyvinyl Chloride 105 3.4 Cable R(!l/m) XL (!lIm) C (IlF/m)
C3+C4 2.006E-03 3.932E-4 2.643E-4
XLPE Cross-linked 90 5.0
Polyethylene
MVXLPE Medium Voltage 90 2.3
Cross-linked
Polyethylene
EPR Ethylene- 90 2.5
Propylene Rubber

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
 Page 8 of 17

converted to a 10 kV base. That is, if the measured

L R current is multiplied by 10 kV instead of 2 kV and
by the power factor, the tabulated power is the
result. From this can be calculated:

--r.---C'-- 2 � •

•
CHL: The high side to low side capacitance:
CHL = CLH = 2060 pF

I
The high side input capacitance:
(. + CHLXCLG
ID
= C HG = 2448 F
CHL +CLG P
Table 13 Manufacturer's No Load Loss
Measurements Converted to Primary Side.
These measurements are made on the low side, with
the high side open circuited. The three-phase
Figure 7 Equivalent Circuit of a Single Phase of Cable.
voltages shown are applied in turn from X" X2 and
D. Transformer Data X3 to Xo and the three-phase power measured. The
The four delta-wye grounded 2500/3333 kVA measured no-load losses on the low side are
AAfFA 1 3 . 8kV-480YJ277V, 7 . 1 5 % IZ cast coil converted to three equal phase currents on the high
transformers are all of the same manufacturer and side.
construction, so that only one transformer model is PL-3</>measured
IL-NLL-line -
- (20)
required. These transformers have integral surge 3VL-line-line
arresters and snubber circuits mounted within the
Because the high side is connected in delta, line
transformer enclosure. The winding configuration
current values must be converted to winding
is shown in Figure 9. Transformer data is listed in
currents :
the following tables:
lL-NLL-line VL
IH-NLL-winding =
VH
(2 1 )
Table 11 Transformer Data. This Table lists the .,f3
transformer data as obtained from the nameplate This high side winding current i s modeled a s a
and the manufacturer's drawings. nonlinear resistive branch Re, which accounts for
Table 12 Manufacturer's Capacitance the fR, hysteresis and eddy current losses. The
Measurements and Calculated Capacitances. current through the resistor Re is modeled as a
The table lists the results of the manufacturer's piecewise linear approximation, shown in Figure
capacitance measurements. These are as follows : 11.
CHL+CHG: The capacitance from the high side to the
low side in parallel with the capacitance from the
high side to ground is obtained by shorting and Table 14 Transformer Saturation Current
grounding the low side windings and injecting a test Measurements Converted to Primary Side.
signal from the high side to ground. These measurements are made on the low side, with
CHG: The capacitance from the high side to ground the high side open circuited. The single phase
is obtained by open circuiting the low side windings voltages shown are applied in turn from Xl to Xo,
and injecting a test signal from the high side to X2 to Xo and X3 to Xo and the currents are measured

ground. (not shown) . The average of the three measured

CLH + CLG: The capacitance from the low side to the low side phase currents is plotted as the rms
high side in parallel with the capacitance from the saturation curve in Figure 1 2 . In order to
low side to ground is obtained by shorting and determine values for the equivalent circuit, the
grounding the high side windings and injecting a average low side current is then converted to
test signal from the low side to ground. equivalent high side winding currents. Since the
CLG: The capacitance from the low side to ground is primary winding is delta connected:
I L-measured VL
. .
IH-wmdmg -
obtained by open circuiting the high side windings - .,f3 VH
(22)
and injecting a test signal from the low side to
ground. In order to obtain inductive current, the no load
The measurements taken were current, power and losses must be subtracted:
power factor (single phase) with an applied voltage
of 2.0 kV at 60 Hz. The tabulated results were

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
Page 9 of 17

To make the vertical axis "- as used in EMTP,

IH-e-winding = l� -Winding - 1� _NLL winding remember that the flux linkages of a coil with N
(23) turns are :

The transformer saturation data is measured as il = Net> = NBA Wb· turn ( 2 8)

root mean square (rms) values of voltage and
Which relates B and "- by the two constants of
current. In a time domain analysis, the saturation
cross-sectional area and number of turns. Neither
data must be in an instantaneous form. The EMTP
N nor A are known for a transformer inside an
program uses flux linkages ("-) as the vertical axis
enclosure which cannot be disassembled. This data
and peak current (IE) as the horizontal axis. [12],
is usually proprietary to the manufacturer. Neither
[13]. The magnetic saturation curve used in
textbooks is the B-H
curve. The magnetic flux
is it convenient to measure either B or et> in an
density vector is given by B,
measured in Teslas .
energized transformer. To make the vertical axis in
volts, as in measurement results, remember that the
The magnetic flux <1>, measured in Webers, over a
voltage across a coil is
surface S is the integral of the dot product of the
flux density and the differential area (Figure 8 ) : dil di
v =- = L-- V
dt dt
<1>
= Is B· dS Wb (24) ( 29)
z Thus for a sinusoidal voltage

vet) = vlzVr s sin(wt) (30)

��__���__�__�__�b____ y m
and

vlz s
il(t) = - Vrm cos(wt)
--

w
x
(31)
Figure 8 Magnetic Flux Density Across a Surface.
The vertical axis quantities can be easily calculated
for plotting saturation according to peak voltage or
Consider the area A, which is the cross-section flux. However, the horizontal axis quantities must
of the magnetic core of a transformer winding: be determined for peak current considering the rms

Is Ioa I: dx . dy
and peak values of the distorted wave shape of
A = dS = m2 (25) saturation current. This may be done by a support
If the flux is perpendicular to a flat surface S routine within EMTP [12], or by other means of
with area A, and the flux density is uniform calculation.
throughout the area, then the magnitudes relate as : The current through the saturable inductor Le,
et> = BA Wb Figure 10, is modeled as a piecewise linear
(26)
approximation, shown in Figure 12.
This is the situation which occurs in a magnetic Table 15 Transformer Series Impedance The
core to which flux is confined, as in a transformer, resistance RT and inductance 4 of the series
where A is the cross-sectional area of the core. The branch, Figure 10, are calculated for the high side
magnetic flux is produced by the current flowing in from the percent impedance, kVA rating and XIR.
the windings wrapped around the transformer core. X2

---\
The magnetic field H,
is a vector in the same H2
B,
B
direction as but with a magnitude which
depends upon the medium through which the flux X1 -XO
is flowing. Here enters the familiar relation:

B= /1H = /1o/1rH T (27) H1 H3 X3

where I-lr (unitless) is the relative permeability of
the medium. If B is plotted against H, the result Figure 9 Transformer Winding Configuration
will be a straight line with slope equal to I-l.

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
 Page 10 of 17

CHL-A
10 71 0.493 0.017 0.010
CHL-B 20 225 0.781 0.027 0.016
CHL-C
30 430 0.995 0.035 0.020
40 722 1.253 0.044 0.025
SO
RT-ALT-A

1072 1.489 0.052 0.030

�r Re-AB
�Le-AB �.n:
"
60 1481 1.714 0.060 0.034
RT-8LT-8
70 1972 1.956 0.068 0.039
80
RT-CLT-C
�� Re-BC \� Le-BC �. "

90
2575
3398
2.235
2.622
0.078
0.091
0.045
0.053
100
�Le-cA ."'
4245 2.948 0.103 0.059
� Re-CA 110 5722 3.612 0.126 0.073
120
'T....f-J CLG-C
CHG-A
CHG-C 15048 8.708 0.303 0.175
CLG-A

CHG-B
CLG-8

20.0
Figure 10 Transformer Equivalent Circuit.

Table 11 Transformer Data

15.0

Property Value 10.0

KVA 2500/3333 kVA
Voltage 13800-480Y/277 V 5.0
Connections L'.-YG

0.0
BIL 95HV/lOLV kV
LV Amp 3007/4009 A

-5.0
HV Amp 104.6/139.4 A
HV Taps 0, ±2.5%,±5.0%

-10.0
Avg'/Max. Amb. 30/40 C
Class AA/FA
Type CAST COIL
-15.0
Windings Copper
Phase 3
Impedance 7.15%
-20.'4--'---'--'----''--
X/R Ratio 9.96 -2 .0-150.0-100.0 -50.0 0.0 50.0 100.0 150.0 200.0
Frequency 60 Hz
Temp. Rise 80/80 C Figure 11 No Load Losses Modeled as a Nonlinear
Resistance Converted to Primary Side.
Table 12 Manufacturer's Capacitance Measurements and
Calculated Capacitances
80.0
Power
Test Current Power Capacitance
ID (rnA) (W)
Factor
(pF) 60.0
(%)
CHL+ CHG 11.808 0.83 0.703 3132 40.0

CHG 4.044 0.846 2.093 1072

20.0
CLH + CLG 23.383 0.919 0.393 6203

CLG 15.619 0.938 0.601 4143 0.0

CHL(calc) -- -- -- 2060 -20.0

C;n (calc) -- -- -- 2448
-40.0
Note: All measurements at 2.0 kY, Ref. to 10 kY, 60 Hz.
-60.0

Table 13 Manufacturer's No Load Loss Measurements I A]

Converted to Primary Side. -80."--1_----'_--'-_-'-_-'--
-1 .0 -9.0 -6.0 -3.0 0.0 3.0 6.0 9.0 12.0
V 3-Phase Calculated Calculated Calculated
(%) Power Current Current Current
Figure 12 Transformer Saturation Curve as converted to
LV Side LV Line HV Line HV Wind EMTP parameters.
(W) (A) (A) (A)

10

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
Page 11 of 17

Figure 1 4 . The peak voltage is 1 6 . 696 kV at t =

Table 14 Transformer Saturation Current Measurements 1 0.022 ms, rising from a value of 0.0 kV at t = 1 0 . 0
Converted to Primary Side. m s for a 11V o f 1 6 . 696 kV and a rise time of 22 Ils.
lavg-rms lavg-peak
The frequency of the oscillation is approximately
A.peak
lavg-rms lavg-rms 25 . 1 kHz, as shown in Figure 1 5 .
V HV HV HV Wind
LV Line HV Line
(%) (A) (A)
Wind Wind (Wb-t)
(A) (A)
The second case i s with the snubber switch closed.
10 1.30 0.045 0.026 0.0342 5.177 If a snubber circuit is applied, the oscillation is
20 1.80 0.063 0.036 0.0558 10.354 lowered in frequency to approximately 5 .4 kHz,
30 2.13 0.074 0.043 0.0663 15.530 and highly damped (Figure 16 and Figure 1 7). The
40 2.62 0.091 0.053 0.0833 20.707 peak voltage is 1 8 . 3 1 6 kV at t = 1 0.086 ms, rising
50 3.09 0.107 0.062 0.0999 25.884
from a value of 0.0 kV at t = 1 0 . 0 ms for a 11V of
60 3.63 0.126 0.073 0.1200 31.061
1 8 . 3 1 6 kV and a rise time of 86lls.
70 4.45 0.155 0.089 0.1531 36.238
80 6.06 0.211 0.122 0.2223 41.415
90 13.88 0.483 0.279 0.5513 46.591
100 21.73 0.756 0.436 0.8998 51.768
110 60.57 2.107 1.216 2.5769 56.945
120 273.66 9.519 5.496 11.6870 62.122

Table 15 Transformer Series Impedance Converted to

Primary Side.

Rr Lr
(n) (mH)
(a) overview
0.54 14.38

III. SWITCHING TRANSIENTS

A. Energization Voltage
Energization voltage occurs when an unloaded or
lightly loaded transformer is switched on. An
.2�+'O. O-�--"�''-�-�-�-�-� -�-� 'mI-I -1-l IO'
0
example of a transient overvoltage in the 1 3 . 8 kV
(lIosm�.� .._� "StulA
circuit was simulated using EMTP/ATP. These "StulB �StulC

simulations are performed assuming the (b) detail

transformer is lightly loaded/unloaded ( 1 000 Q L­ Figure 14 Energization Transient without snubber.

N @ 480V) . Switching conditions consist of
25o-,--------------,-----------;
closing the breaker 52 Bl at t = 0 . 0 1 s. The first
[kV]
case is with the snubber switch open.
200
CHL

150

100

I 50
SNUBBER

Figure 13 EMTP/ATP model of system. Lumped

transformer model is shown for compactness, but discrete
transformer model was used as shown in other figure.
�*.0-�2�.5-�3.�0 -�3.5C -�4.0�-�4 �==5�.0--�5�.5-C
(file WTC2nosnubber.pI4; x-var t) II :)f)OOlA
[S[�6.0 �
Snubber is shown switched in model to simulate response
Figure 15 Plot of voltage at Transformer primary without
with and without snubber.
snubber showing resonance peak at 25.1 kHz. The

Simulation results show the energization transient horizontal frequency axis is in values of x, where f = 10'.

when the transformer primary breaker is closed,

11

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
 Page 12 of 17

.' The series inductance o f the transformer i s 1 4 . 3 8

mHo The input capacitance Cin o f the transformer
is 2448 pF (Table 1 2) . The capacitance of the cable
between the circuit breaker and the transformer is
3200 pF. The total capacitance is thus 5648 pF.
The resistance Re is calculated from the No
Load amps (5 . 1 1 A at 1 00% voltage) . First,
convert the current to the primary voltage,
o
....
(.,. ...... ""'" pI4;.- � vSfUl�
•
vSfUlB �Sf.l.OC
(mo1 20

Ie = -INL
VL
= --
480V
- X 5.11A = 0.178A
(a) overview VH 13.8kV
.' Then calculate the resistance:
VLN 13.8kV
Re = -=
Ie ..J3 X 0.178A
= 44.8 kf!
The magnitude of the current chopped is
dependent upon the properties of the cathode
(negative contact) material. Materials with a high
vapor pressure and low thermal conductivity tend
to produce lower chopping currents [3] . The
9.9 10,0
1�1o IIrlJbbor_pIoI; ,.\Of � VS� �SfUlB
10.1
�S�
10.5 [mol 10.6
duration of the arc instability which may result in a
current chop is a statistical phenomenon, with a
(b) detail
consistent average value.
Figure 16 Energization Transient with snubber.
Suppose, for example, that the magnitude of a
70
[kVI current chop is 2.0A. The energy in the inductance
60 from the chopped current is:

50 EL = "21 LeIo = "21 28H

2
x (2.0A) 2 = 56 J
40 Given that the energy in the inductor equals the
energy in the capacitor during resonance:
30
1 1
L 2 _

-
2
"2 eIo "2Cop Vc

� =IoZo
20

�
Then:
10

0
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 [sl 6.0
Vc = 10 C
e
(file snubbeLpl4; x-var I) v :SNUBA
op
The surge impedance of the transformer input is
Figure 17 Plot of voltage at Transformer primary with
5.4 kHz. 28H
Zo= � = 5648pF = 70 kf!
snubber showing resonance peak at The horizontal
fre quency axis is in values of x, where f = lOx.

B.
Current Chopping �c;
The voltage produced is:
Vc = IeZo= 2.0A x 70 kf! = 140 kV
Current chopping occurs when the interrupter
breaks the current flow in an inductive circuit
If the transformer series inductance is
before a current zero . In certain types of vacuum
considered instead :
and SF6 circuit breakers, the current can be
14mH
Zo= � = 3200pF = 1.57 f!
interrupted before the current zero. The magnitude
of the current chopping transient may be estimated
in a number of ways. [4] First, a few constants �c;
The voltage produced is:
Vc = IeZo= 2.0A x 1.57 = 140 kV
need to be calculated:
The transformer excitation current was
approximately 22A at 480V. On the primary side, This is an excessive theoretical value which
the equivalent current is Ie = 0.765A. The does not correspond to observed voltages. In
inductance is: actuality, the voltage is limited by two factors [4] :

=--=
VLN 13.8kV
= 28 H
1 . Due to hysteresis, only a fraction of the energy
Le in the magnetic core is transferred to the
2rr e fI ..J3 x 2rr x 60 x 0.765A

12

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
Page 13 of 17

capacitance. The remainder is trapped within from a value o f 5 .923 kV at t = 14.038 ms, for a
the core as residual magnetism. For grain f't.Vof 5 .753 kV and a rise time of 1 9 /lS. The
oriented steel: overvoltages shown in the simulation are less than
Vc = 0.551eZo = 0.55 x 140 kV = 39 kV the peak voltage of the sinusoidal wave, much less
2. Due to resistances in the circuit damping the BIL.
reduces the magnitude of the voltage peak.
35,------,
The damping constant is: [A[
R e 44.8 kD.
= = = 0.64
25

TJ Zo 70 kD.
As is well known, the circuit response varies with 15

the damping factor as follows :

• If 11 < Y2, the circuit is overdamped.
• If 11 = Y2, the circuit is critically damped. ·5

• If 11 > Y2, the circuit is underdamped.

In this case, the circuit is underdamped, and has an ·15

oscillatory decay.
14.3 [ms] 14.4
·25+--�-�--�-�-�-��-�-_J
The time constants of the circuit must be
calculated. The parallel time constant is:
13.6 13.7 13.8 13.9 14.0
(file snubber.pI4; x-val II c:PDCBA S A c:PDCBB S B c:PDCBC S C
14.1 14.2
- W - W - W

Tp = R eCap = 44.8 kD. x 5670 pF = 254 [1.S Figure 19 Chopping of Current at 5 Amperes

The series time constant is: 12 ,----------------.-------,

Le 28H
Ts = R =
44.8 kD.
= 625 [1.S
e
The period of the undamped circuit is:
T = .jTpTs = .j254[1.s x 625[1.s = 398 [1.S
Using the solutions curves of the second order
differential equations in [4] , the peak voltage is 0.4
per unit or 15.6 kV. ·4

These simulations may be performed assuming

·8
the transformer is fully loaded. In this case, the
load was modeled to produce approximately 1 00 A
(rms) primary current (0. 1 Q L-N @ 480V) . A 4 8 12 16 [ms] 20
(fileWTC2snubber.pI4; x-varl) v:SNUBA v:SNUBB v:SNUBC

nominal resistor ( 1 000 Q L-N @ 1 3 . 8 kV) was Figure 20 Transformer voltage with current chopping.
placed on the source to prevent the voltage value
from floating by providing a complete circuit for A. Transient Recovery Voltage (TRV)
the source after breaker opening. This is shown in Transient Recovery Voltage (TRV) occurs when
Figure 1 8 . the transformer is being de-energized, and the
CHL
voltage across the just-opening contacts of the
breaker exceeds the breakdown voltage of the
interrupting medium, resulting in a high current
transient.

r The TRV can occur when the primary breaker

SNUBBER
interrupts a fault on the transformer secondary,
causing oscillations between the cable and
Figure 18 EMTP/ATP model of system for current
transformer winding capacitances and inductances.
chopping.
If the switch has not opened sufficiently, reignition
Current chopping in the system under study was can occur, resulting in a high frequency oscillation.
simulated and the results shown in Figure 19 and
The TRV of the breaker is shown in Figure 2 1
Figure 20. The interrupted current causes a voltage
without a snubber. The peak transient voltage is
transient, as expressed by the well-known
5 .727 kV, occurring at t = 1 4 . 0 2 1 ms, rising from a
relationship V = L di/dt. The peak transient voltage
is 1 1 .767 kV, occurring at t = 1 4 .057 ms, rising

13

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
 Page 14 of 17

value of 0.0 kV at t = 14.011 ms, for a L'lV of 5.727 6 .----------------��-____,

[kV[
kV and a rise time of 11 f!s.

Figure 22 shows the reduction of current chopping

transients when a snubber is used. The peak value ·3
is 6.779 kV at t = 14.066 ms, rising from a value of
·6
5.847 kV at t =13.988 ms, for a L'lV of 0.932 kV
and a rise time of 78 f!s. ·9

Figure 23 shows the reduction of TRV when a ·12

snubber is used. The peak transient voltage is 2.485 ·15

kV, occurring at t = 13.995 ms, rising from a value . , 8 +-�-�-�- - - - - - -:l

---.. ---.. ---.
, 2. ---.. ---.
, 6. . -:-
---.,--
of 0.0 kV at t = 13.989 ms, for a L'lV of 2.485 kV 0 8 [ms l
(fileWTC2snubber-opening.pI4; x-var t) v:PDCBA -SWA
20
v:PDCBB -SWB v:PDCBC -SWC

and a rise time of 6 f!S.

Figure 23 Reduction of Transient Recovery Voltages with
8. o ,-------, snubber.
[kV[
Table 16 TRV ratings for 15 kV circuit breakers [14]
2.8
TRV
C37.06 Rated
C37.06 Peak Time
Table: Maximum Test
·2.4 Table Value, t3
Line Voltage Duty
No. u, IlS
No. (kV)
kV
·7.6
Terminal
1 5 15 66 10
Fault
·12.8 Out-of-
1 6 15 132 20
phase

. , 8.0 +-�-�---;----;----;-- - - --..". ,..--:!

2 9 15 TWO' 25.7 66
-�
,2 � -r-
,6 -
0 [ ms l 20 2 10 15 T60' 27.5 29
(fileWTC2snubber.pI4; )(-var t) v :PDCBA -SWA v :PDCBB -SWB v :PDCBC -SWC
2 11 15 T30' 29.4 15
no'
Figure 21 Transient Recovery Voltages without Snubber
2 12 15 31.2 15
* =
12 ,-------,
Percent of rated interrupting capability.

Table 17 Summary of Results

Vpk(kV) dV/dt kV/l1ls

Case Voltage Snubber Snubber B I Ll Snubber Snubber Fwl

TRV
Out In Out In t3
pk
·4 Closing Pri. 17.435 18.296 95 0.793 0.203 79.200

Opening Pri. 11.767 6.779 95 0.308 0.012 79.200

·8

Opening TRV 5.727 2.485 31.2 0.521 0.414 15.000

o
·12 +--.....--.----1
12 16 [ms[ 20

B.
(file snubber.pI4; x-vart) v:SNUBA v:SNUBB v:SNUBC
Summary of Simulation Results

Figure 22 Reduction of current chopping transients with The withstand capabilities of the transformer
snubber. under investigation consist of:
1. Basic Lightning Impulse Level (BIL) for the
HV winding from transformer nameplate.
This is determined from the 1.2 x 50 f!S Full
Wave Test Value in kV performed by the
manufacturer.
2. dv/dt or rate of rise of the voltage at the
transformer terminal.

14

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
Page 15 of 17

a. Front of Wave (FOW) Test Value in kV

from IEEE Std. C57.98-1993, Table CABLE
1. [15]
b. FOW Time-to-crest, from IEEE Std.
[
c.
C62.11-2005, Table 8. [16]
Full Wave (FW) Test Value in kV from
IEEE Std. C57.98-1993, Table 1[15]
FUSE
�• S U RG E
ARRESTER
TRANSFORMER

d. FW rise time from IEEE Std. C57.98- RESISTOR

1993, Table 1[15].
For the circuit breaker that switches the CAPAC ITOR
transformer, the concern is the TRV rating.
The TRV ratings standard is in IEEE C37.06-
2009, Table 1 [14]. The ratings for the lowest
I
Figure 24 Snubber circuit (per phase)
current were used because the switching is done
under loaded, not fault conditions.
The summary of results, Table 17, shows that The component values are as follows :
the snubber substantially reduces both peak voltage
and dV/dt for all cases. The calculated values both A. Surge capacitor.
with and without the snubber are all well below the The capacitance should reduce the oscillatory
limits shown in the applicable standards. frequency by approximately an order of magnitude,
so that it is below the first peak of the transformer
resonance characteristic (25 kHz). This can be
IV. SNUBBER DESIGN
accomplished by a capacitance of 0.25 /IF.
The snubber circuit consists of four basic The capacitors should have a BIL to match that
components (one of each per phase) : of the transformer, 95 kV in this case, and should
1. Surge capacitors be a surge protection type, Figure 25. The current
2. Damping resistors drawn by the capacitor (neglecting the resistor) can
3. Fuses (optional) be calculated from the capacitive reactance:
4. Surge arresters 1 1
Xc = -- = 6
= 1 0 .6 kQ
For this transformer, a snubber circuit was already 2nfC 2n x 60 x 0.25 x 1 0 -
included and shown in the drawings. The design The current drawn is thus:
values for that snubber are shown here. The
snubber arrangement for a single phase is shown in
_ VL_N 1 3.8 kV
Figure 24, with the component values used shown Ic - = O .75 A
in Table 18. Xc .J3 x 1 0.6 kQ

Table 18 Snubber Circuits

Component Value Type

Fuse 15.5 kV, 6 A Current-limiting
Capacitor 13.8 kV, 0.251lF Surge capacitor
Resistor 20 0 +10% Non-inductive

Figure 25 Typical surge capacitor, three-pole, medium

voltage.

15

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
 Page 16 of 17

B. Damping resistor. Should the fuses blow for an arcing fault where the
The resistor should be non-inductive. Typical snubber circuit is not destroyed, then the protection
high power, high voltage non-inductive resistors provided by the snubber circuit will be lost.
(Figure 26) are constructed of a thick film of
resistive material coated on a ceramic tube,
typically 0.3 - 0.6 m in length and 0.025-0.05 m in
diameter, with a conductive band at each end to
make connection with. The desired RC time
constant is 1 - 1 0 f!s. Thus with a 0.25 capacitor, the
resistance should be between 4 and 400. The
resistance should also attempt to match the surge
impedance of the cable. The surge impedance was
Figure 27 Typical fuse for snubber circuit.
calculated to be approximately 240 at 60 Hz, but
140 at 25 kHz. The existing 20 0 resistor is an D. Arrester.
acceptable value. This resistor has a time constant The characteristics of the existing arresters are
with the 0.25 f!F capacitor of listed in Table 19. The maximum available fault
current is 32.9 kA. The available fault current
T = RC = 20x 0.25 X 1 0- = 5.0 J1S .
6
should be no more than approximately SO% of the
The resistor should have a continuous power
pressure relief rating of the arrester to allow for
rating sufficient for twice the capacitive current (in
inaccuracies in calculating fault currents and for
case of harmonics and overvoltages) . With the
future system expansion. Using the 80% criterion,
capacitive current of 0.75 A, the resistor should be
the pressure relief rating should be 4 1 kA. The
able to dissipate:
existing Distribution Class arresters, having a
P= / 2 R = (0.75AY 200 = 11.25W pressure relief rating 20 kA, should be replaced
If higher current is present due to harmonics and/or with Station Class Arresters, Figure 28, having a
overvoltages, the resistor may dissipate up to : pressure relief rating of at least 40 kA.

Table 19 Surge Arresters

Property Value
Rated Voltage 18 kV rms
MCOV 15.3 kV rms
Class Distribution,
heavy-duty
Material Polymer
Pressure relief 20 kA, rms for 12
rating cycles.
Figure 26 Non inductive power resistor.

P= / 2 R = (2 x O.75A) 2 200 = 45W

If the capacitor should short, the resistor will pass
400A, until the fuse opens. Assuming that the fuse
clears in 0.004 seconds or less, the resistor will
have to absorb the following energy:
W = Pt = 4002 20 x 0.004 = 12.SkJ
The existing 200 ± 1 0 % non-inductive resistors
are rated for 750 Watts and 15 kJ. With the
protection of the fuses, the resistors should survi ve
a short circuit.
C. Fuse.
The fuses are optional. The existing fuses are 6A,
15.5 kV high-speed full-range current limiting Figure 28 Polymer Station Class Arrester.

fuses, similar to that shown in Figure 27 . If the

fuses are not used, the resistors will serve as fuses.

16

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.
Page 17 of 17

V. CONCLUSIONS [ 15] IEEE Std. C57 .98-1993, IEEE Guide for Transformer
Impulse Tests
A snubber study was performed for the four (4) [ 16] IEEE Std. C62. 11-2005, IEEE Standard for Metal-Oxide
2500/3333 kVA 1 3 . 8 kV - 0.48 kV transformers. Surge Arrestersfor AC Power Circuits (> 1 kV)
The purpose of the Snubber Study was to evaluate
the snubber design as shown in the drawings for
VII. B IOGRAPHY
the transformers. Only one transformer was
analyzed because the all data, including the cable
Peter E. Sutherland (Fellow, IEEE) received the B . S .
lengths are the same for all transformers.
degree in Electrical Engineering from the University o f Maine,
The snubber substantially reduces both peak Orono, and the Ph.D. degree in Electric Power Engineering at
voltage and dVIdt for all cases. The calculated Rensselaer Polytechnic Institute, Troy, NY.
values both with and without the snubber are all In 1987, he joined General Electric Company, Schenectady,
NY, and held a variety of positions, becoming a Senior
well below the limits shown in the applicable Engineer in the GE Power Systems Energy Consulting
standards. Department. In 200 1, he joined SuperPower, Inc., Schenectady,
One change was recommended in the snubber N.Y., where he worked on applications of superconductivity to
electric power systems. Dr. Sutherland then joined EPRI PEAC
design, to replace the distribution class surge Corporation' s (now EPRI Solutions, Inc.), Schenectady, NY
arresters with station class. office as a Consulting Engineer. He is currently a Lead
Consultant with GE Energy Services in Schenectady NY.
VI. REFERENCES Author of numerous technical papers, he is active in the IEEE
Industry Applications Society, and in the IEEE Schenectady
[ 1] Young, B .W., " Snubbers: Protecting medium-voltage Section. Dr. Sutherland is a member of CIGRE and the lET
transformers from electrical transients," Consulting­ (formerly lEE). He is a Registered Professional Engineer in
Specifying Engineer, March, 2009, pp.4 1-44. Pennsylvania, Maine, New Jersey and New York and is a
[2] IEEE Draft Standard PC57. 142/D8 .2, May 2, 2010. Guide Chartered Engineer in the UK.
to Describe the Occurrence and Mitigation of Switching
Transients Induced by Transformer, Switching Device and
System Interaction.

[3] Greenwood, A., Vacuum Switchgear, London, UK: The

lET, 1994. pp. 8 1-89 and 216-218.
[4] Greenwood, A., Electrical Transients in Power Systems,
2"d Edition, Hoboken, NJ: John Wiley & Sons, 199 1. pp.
92- 100.
[5] Gustavsen, B . , "Wide B and Modeling of Power
Transformers," IEEE Transactions on Power Delivery,
Vol. 19, No. 1, January 2004, pp. 414-422.
[6] Gustavsen, B . , "Study of Transformer Resonant
Overvoltages Caused by Cable-Transformer High­
Frequency Interaction," IEEE Transactions on Power
Delivery, Vol. 25 , No. 2, April 20 L O , pp. 770-778 .

[7] Paul, C.R., Analysis of Multiconductor Transmission

Lines, 2"d Edition, Hoboken, NJ: John Wiley & Sons,
2008, Section 4.2.3, pp. 130- 144.
[8] Power Cable Manual, 4th Edition, Southwire® , CaroUton,
GA, 2005. Table 2-6, pp. 2- 15.
[9] Paul, c.R., Inductance Loop and Partial, Hoboken, NJ:
John Wiley & Sons, 2010, pp. 157-158 and 165-166.
[ 10] Gustavsen, B . , Martinez, J.A. and Durbak, D. "Parameter
Determination for Modeling System Transients-Part II:
Insulated Cables" IEEE Transactions on Power Delivery,
Vol. 20, No. 3 , July 2005, pp. 2045-2050.
[ 1 1] Power Cable Manual, 4th Edition, Southwire® , Carollton,
GA, 2005. Table 3 - 1 , p. 3 -4.
[ 12] Dommel, H.W., Electro-Magnetic Transients Program
(EMTP) Theory Book, Portland, OR: Can-Am EMTP
Users Group, 1987.
[ 13] Can-Am EMTP Users Group, A TP Rule Book, Portland,
OR: Can-Am EMTP Users Group, 1995. Section XIX-G.
[ 14] IEEE C37.06-2009, A C High- Voltage Circuit Breakers
Rated on a Symmetrical Current Basis- Preferred
Ratings and Related Required Capabilities

17

978-1-4673-0651-5/12/$31.00 © 2012 IEEE

Authorized licensed use limited to: UNIVERSIDADE FEDERAL DE ENGENHARIA DE ITAJUBA. Downloaded on April 29,2021 at 21:41:35 UTC from IEEE Xplore. Restrictions apply.