Anda di halaman 1dari 15

studies will stimulate others to make similar among different investigators is for all to use at the time such

among different investigators is for all to use at the time such tests were made. Therefore,
measurements so that the over-all fund of standard procedures and equipment. care should be taken to select for such meas-
knowledge of carrier-frequency noise phe- The procedures described by Mr. Carter urements a time at which the noise level is
nomena can be increased. The only possi- give results predicated upon the existence of known to be unusually high, for example
ble way to assure consistency of results a noise level no greater than that prevailing during a thunderstorm.

The Duty
ne Luty on LIghtning
Irre.
igntning
* eAs or
orresters tor drawn in this paper are logical in the
presIent state of knowledge.
Estimates of lightning arrester dis-
A-C Systems charge oncurrent
made magnitudes
a theoretical and been
basis3have data
have been accumulated in the field by
EDWARD BECK recording currents discharged through
MEMBER AIEE lightning arresters in actual service. It is
of interest to note the rather good agree-
ment between the conclusions of the
THE duty of lightning arresters is to surfaces. However, there are unusual early paper by A. M. Opsahl3 and the
protect the electric equipment against conditions of installation which may re- field data. Considerable infornation has
damage and the service against outage quire special consideration. Present been accumulated on the expected fre-
from transient overvoltages. The duty standards cover this phase.2 quency of discharge and on the crest
on lightning arresters is the continuous magnitudes of lightning arrester dis-
normal system voltage to which they are General Discussion of Lightning charge currents. A limited amount of
subjected; the frequency with which Arrester Discharge Currents data is available on wave shapes of the
they discharge; the character, that is, discharge currents. For descriptions of
crest value and wave shape of the surge The characteristics of lightning arrester the recording devices used in the field
currents and the power follow current discharge currents are not definite, fixed researches the reader is referred to the
which they are called upon to handle. quantities, but have a statistical distri- list of references given at the end of this
This paper concerns itself not with the bution. The selection of a current for paper.
protective characteristics of lightning laboratory or routine tests must bear
arresters but with the duty on them and this in mind. The conclusions reached Frequency of Lightning Arrester
suggests tests which will demonstrate in this paper are based on field research, Discharge
their ability to perform their operating theory, and experience. As in many
duty cycle under conditions which are matters involving probability, there may Data in statistical curve form from two
expected in service. The test procedures be differences of opinion regarding the sources are shown in Figures 1 and 2.
suggested pertain to valve-type lightning conclusions, but it is believed that those Additional data on distribution lightning
arresters. Significant duty cycle tests arresters on distribution systems are
on expulsion-type lightning arresters are published in references 6 and 7. The
under study. A paper' by 0. Ackermann En OO
bU) - data in Figures 1 and 2, gathered prin-
discusses that subject. <- cipally on distribution and highly in-
' < A sulated unshielded medium voltage lines,
Normal System Voltage <80 are representative of the available in-
z
Z ffi _ formation.
Lightning arresters are subjected to 60- > The majority of the records have been
cycle voltage continuously, for all prac- < 60 8 obtained by means of devices with a lower
tical purposes. Unless discharged by a limit of recording ability of 50 or several
surge, they are expected to be insulators, G O- C hundred amperes. It has been found by
or very high resistances, depending on cn 40 t X_ the use of more sensitive instruments in
their construction. Their insulating ' >_ limited number that under some condi-
qualities are expected to remain as con- C. tions lightning arresters apparently dis-
stant as those of other insulation in wet a 20 - charge surge of quite low magnitude nd
as well as dry weather and under reason- Z _ishort duration more frequently.5 Some
able degrees of contamination of the outer _= in addition to those published have
Lrecords
- 2 4 6 8 given evidence of a large number of short
-- ~~~~~~~~NUMBER
OF DISCHARGES PER INDIVIDUAL duration discharges of less than 50-
Paper 51-190, recommended by the AIEE Protec- ARRESTER PHASE PER YEARiuarsatodr
tive Devices Committee and approved by the Tech- aprsceti atclrsaindr
nical Program Committee for presentation at the Figure 1. Distribution of discharges through ing certain periods. These, it was con-
AIEE Summer General Meeting, Toronto, Ont., individual lightning arresters4 cluded, were not caused by lightning, but
Canada, June 25-29, 1951; Manuscript suhmitted
January 31. 1951; made available for printing A. 34 single-phase lightning arrester years4 by surges produced by some system dis-
i wthth WetighuseElctic
EDWRDBEK B. 1,020 lightning arrester years9 turbances. These small discharges
Corporation, East Pittsburgh, Pa. C. 73 lightning drrester years23 through lightning arresters are sometimes
1134 Beck-The Duty on Lightning Arresters for A-C Systems AJEE TRANSACTIONS

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
z l00o discharge for a higher percentage of dis- lines shielded adequately against direct
(n
< @ - A turbances. Apparently this factor is not strokes for at least a half to one mile out
a: 80 [ l\ < uas weighty as the density of lightning from the station, the frequency of opera-
TicI) 80 IM eD l ° \arresters per mile of circuit. A similar tion is less than in the more exposed sta-
c _ B situation was found in an investigation of tions. This is indicated by field investi-
o60
cr
ww <60 _
discharges through expulsion lightning gations9"l and by calculation.12
zD _ W Figure 2. Average number Crest Magnitudes of Lightning
, 40 - - of single-pole lightning ar- Arrester Discharge Currents
Iw-
rester discharges per year6
x
L - There exists a large volume of data on
<20 -< i20,.-
--
n a

stations, 133.5 lightning the crest magnitudes of lightning arrester

cr Lt
Z<r

<
e >>

t 1 iI L 2E j
B darrester years

B. Distribution lightning
discharge currents, recorded principally
by the surge crest ammeter.'3 Although
0. a - I 25 3 6 arresters, 160 lightning ar- this device has certain limitations, they
NUMBER OF DISCHARGES PER ARRESTER PER YEAR rester years are not serious. The lower limit of sensi-
tivity is usually 50 or more amperes. In
the case of several surges of the same
tp.O polarity, the maximum is recorded. If
accompanied by power follow current arresters of the transmission line type. several res occur inspections,
and sometimes not. It requires a surge of Such lightning arresters at isolated loca- theyan be dsiuished. Howv,
some minimum magnitude and duration tions disearged about once per year, theyprobabe oishingu areenot
timed in the proper relation to the phase of whereas lightning arresters of the same great itis onsid unikl at
the system voltage to start power follow. type distributed along the lines at every posib. It hs aosigniicant e tho the
The records of small magnitude were second pole discharged on the average of data.
obtained on valve lightning arresters on a 0.35 times per year. F st t t
subtransmission system. On these the In high-voltage stations connected to tion rof msues of cresti current
ratio of lightning arrester sparkover to
reported by a number of investigators.
triutonsystemvlag
tribution systemss rethe rativy
ratiolow.
of Otndis
lightning L-l t100o
80 The data comprise records of generally
arrester sparkover to system voltage is 60
relatively high in most cases, and it is D-1 Figure 3. Statistical distribution of records
believed that most distribution lightning 40 D-2 of crest currents obtained on various in-
arresters would not be discharged by such eG Vstigations
small surges, even if present. The low D-3
current, short discharges are not con- en 20
sidered significant as far as duty on (no t\\\ \\ >(cD-4
lightning arresters or effect on apparatus < - V_ _- /
is concerned. * .Hereafter ,
they are neg- < ~~~~~~~~~~~10
8 - WN I'M V X7 -
= -
lected. . - -- -
It is concluded from the available data -
that in isokeraunic levels of 30 to 40, 4 e - - L-2
lightning arresters on rural distribution .e
systems may be expected to discharge, on
the average, about once per year. On 2 - - - - -I
urban distribution systems the expected AK
frequency of operation is less because of s-I l
the more effective shielding from build- w - - - - - - -
ings. In stations -
. . . connected to wood pole , l l
i|l
the number of a-hl~~~~~~~~~~~~ 0.6._
. T
lines without shield wires,
discharges per lightning arrester pole is 0.4 --2-_ - - - -- -
estimated at about 1.5 per year. The l .l
_
figure may be affected by the number of
lines entering a station and the number of 0.21 0 44 83
-~~12 - - 1 6 2
20 - 24 - - 28- - 32 - 36
lightning arresters in the station.9 Light-
ning arresters in stations such as those KILOAMPERES
mentioned discharge more frequently than D-1. Wilbur and McMorris. Lightning arresters on 4,800 volt delta distribution system,25
lightning arresters on distribution sys- 500 records
tems. This may be explained by the fact D-2. McEachron and McMorris, distribution lightning arresters,6 1,608 records
that, in general, there is alower density of D-3. McCann and Beck. Distribution lightning arresters,5 114 records
suc sttin pe mil of lin thnditi D-4. McCarthy, Stann, Edge and McKinley. Distribution lightning arresters,7 1,685 records
G. Grunewald. Lightning arresters in Germany
butonranforerswit lihtnng rret- L-1. Andrews and McCann. Expulsion lightning arresters on 33-kv wood pole lines,"°
ers. Distribution lightning arresters are 73 records
lower in voltage ratings and sparkover L-2. Collins. Lightning arresters on 24-ky wood pole lines.'6 499 records
than lightning arresters on subtransmis- S-I. McCann and Beck. Lightning arresters in stations,5 266 records
sion systems, and might be expected to S-2. Gross and McMorris. Lightning arresters in stations,9 459 records

1951, VOLUME 70 Beck-The Duty on Lightning Arresters for A-C Systems 1135

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
not less than 50 to several hundred l o _
l ll
l- _
amperes. The surges of small magnitude 80C\ t -
mentioned under Frequency of Discharge 60 -x - = - =
==
are not included. Possibly differences in 4C - - - _ -
the limits of measurement affect the _ _ s -
curves in some degree so that they may
not all actually be on the same basis. It 5 -- _ - -
is believed that the probable effect of
this on the over-all picture is small. 1c - - - .-
In Figure 4 the data of Figure 3 are V 8l - - -| | - -| - -
compiled. The curves of Figure 3 indi- - -
cate that there may be a difference be- 0) 4 - -
tween the data for distribution lightning - - - - - _
arresters and those in stations. There- 2 - .
fore three curves are shown in Figure 4 as Aj
explained in its caption. The maximum C
current indicated in Figure 4 is 35,000 - = _ v =
amperes. Examination of the spark gap z 0.8 =_ - _
electrodes of lightning arresters that had X 0.6 z |\ z i ---
been in service on rural distribution lines o. - - _ -- X
has given some evidence of occasional
higher currents.'4'5 The dashed curves - - - - - - - - - - - - -
are the crest currents recorded in an 0.2.
investigation of direct strokes to masts
and rods." These have been plotted for 0.0 8 12 6 2 4 b3 -
2
comparison. The crest currents of strokes KILOAMPERES
to lines are probably of the same magni-
tudes as those to masts and rods of Figure 4. Statistical distribution of discharge records compiled from data of Figure 3
moderate heights. It thus appears that A. Currents in distribution lightning arresters, 3,907 records. Wilbur and McMorris,
the currents discharged through lightning McEachron and McMorris, McCann and Beck, McCarthy et al
arresters are less than those in the stroke B. Currents in lightning arresters in stations, 1,184 records. Collins, Gross dnd McCann,
terminal. The ratio, roughly speaking, is McCdnn and Beck
about one to ten. C. Compilation of all records except Grunewald, 5,164 records
To investigate the nature of the statis- S. Direct stroke records.16 Maximum component in strokes
tical distribution of the crest currents and S'. Direct stroke records,16 all components
to make some reasonable extrapolations
to currents higher than any yet recorded, After considering the data and its pos- basis of one discharge per year, any
the data of Figure 4 are replotted to arith- sible limitations at high currents, it seems lightning arrester may be expected to dis-
metical probability coordinates in Figures reasonable to assume curve C' of Figure charge currents at least as high as stated
5(A) and 5(B).'7",8 These figures are 5(B) to be representative of the distribu- in the first column of Table I as often as
discussed at greater length in the appendix. tion of crest currents for both lightning stated in the second column.
To be brief, the distribution of the loga- arresters in stations and at distribution The rare occurrence of very high dis-
rithms of the current crests are straight transformers, for purposes of estimating charge currents is apparent, and is corrob-
lines, indicating that the logarithms have a duty. It may be a little light on the orated by the fact that none has been
normal, chance distribution. The diver- distribution arresters and heavy on those recorded. It is the author's opinion that
gence between the lines A' and B' for in stations. Nevertheless, if it errs, it the 65,000- and 100,000-ampere tests now
distribution lightning arresters and those errs on the severe side for station arresters in use as criteria of the ability of a light-
in stations indicate a significant difference of which more is expected than of distri- ning arrester to stand up in service have
between the two sets of data. Extrapo- bution or line types. Curve B', applying not been well chosen although the magni-
lation of A', B', and C' indicates a more specifically to arresters in stations tudes may be impressive. At least, they
small probability of discharge currents of may be preferable for use in estimating the do not seem significant for valve lightning
65,000 amperes and higher. The break protective levels provided by arresters in arresters. A high surge current has merit
in the A' line to A" suggests that the stations without line or station shielding. for expulsion lightning arresters, because
distribution of discharge currents higher The frequency with which lightning
than 20,000 amperes is different from that arresters on distribution systems and in
below this value, and that the probability stations connected to unshielded lines Table I
of very high currents is less than one may be expected to discharge currents __________________
might expect. Unfortunately data at with crests at least as great as any prob- Amperes Frequency
the high currents are scarce. For the able magnitude is shown by the solid
present it must be admitted that we do curves in Figure 6. They are derived 100,000 . Once inl14,300 year
not know the facts. If the data for from curve C' of Figure 5(B) assuming 4,00°0
6000."
direct strokes, all components, from average frequencies of discharge of 1/2, 1, 10,0 o o ..: 45...
Figure 13 of reference 16 are plotted in
the same manner, they do not show this
and 1'!, times per year per lightning
arrester. The station data alone are
5
5000
830 "..
. " "
131/
2
aberration. plotted dashed, for information. On the _100 ..

1136 Beck- The Duty on Lightning Arresters for A-C: Systems ATEE TRANSACTIONS

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
_6 A

32t L LMt IllXt tUX W W : E1 32 --

28 t t X g 50 28 < + XS 4 _ 0

24-----1+AS A F 945 2g.-24 - 4'..+


4.5
-- - 20)
4--0
U) - - - - - - - - - - -~~~~
-x w3 16

Y. F. 0 0.

4.0 ct w
B Arithmetic
------distribution of crest 3.0rents /h0Ii 12
w~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~5 90 8 06

9590807060 40 20 0 5 2 0.5 0.2 0.050.1PRCN0TLATA RA SODNT


20

AS0 ASPERCEN PER CENT


AT LEAST GREAT ORDINATE
REA ASORINAE c

vlop5(B).o Probable of09.A25dB


-----dbLSty orsonIncludes
oA both
Figur 5(A.Prbabledistibuton
Figure5(A). Probabletiondistrit
of crest ofcres curentsFigure
currents in
distribution of all crest currents.
distribution and lightning arresters in stations, 5,164 records
25 & -

A. Arithmetic distribution of crest currents in distribution lightning C. Arithmetic distribution of all currents. Use ordinates on left
arresters. Use ordinates on left C'. Distribution of logarithms of all currents. Use ordinates on right.
B. Arithmetic distribution of crest currents in lightning arresters in The dashed lines embracing C' are the boundaries of the confidence
stations. Use ordinates on left envelope for a probability of 0.95. A' and B' correspond to A' dnd
.A'. Distribution of logarithms of crest currents in distribution lightning B' of Figure 5(A), for reference
arresters. Use ordinates on right
A". Possible distribution of logarithms of crest currents above 20,000
amperes in distribution lightning arresters. Use ordinates on right arsesi ttosi niae nFgr
B'. Distribution of logarithms of crest currents in lightning arresters in arresters in stations iS indicated in Figure
stations. Use ordinates on right 7, reproduced from reference 9. Similar
The ddshed lines embracing A' and B' are the boundaries of their indications have been obtained in another
confidence envelopes for a probability of 0.95 study'1 which has yielded a limited
amount of data on systems up to and
including 230 kv. The data from this
investigation, if plotted on Figure 6)
it subjects the arcing chambers of these arrester.' In well shielded systems the would lie above the curve for 0.5 dis-
lightning arresters to great mechanical probability of this is remote. It would charge per year.
forces as a result of the high and sudden result from a backflash from the shield The available data on the crest magni-
gas pressures produced by high surge cur- wires to the system conductors or from a tudes of lightning arrester discharge cur-
rents. Experience indicates it to be a stroke that side flashes to the conductors rents may be summarized as follows. The
good criterion for expulsion lightning despite the shielding. This is a possi- curves of Figure 6 give the probable
arresters. bility, but in well shielded systems it is a expectancy of crest magnitudes for rural
It has already been said that the data of rare occurrence, as is evident from the distribution systems and medium voltage,
Figure 6 are based on records obtained, excellent flashover records of well shielded highly insulated systems in isokeraunic
for the most part, on highly insulated lines. The probable distribution of dis- levels of 30 to 40 in altitudes below 5,000
and unshielded lines of medium voltage. charge currents in high voltage, shielded feet. The curves may be corrected in
Although the volume of data on high- stations is affected by numerous factors proportion for other isokeraunic levels,
voltage lines, both unshielded and such as the arrangement and extent of the by increasing or decreasing the average
shielded, is not large enough to warrant shielding and the ground resistances. number of discharges per lightning ar-
conclusions based on records, it would Field measurements in significant number rester per year. It is possible but not
seem that the curves of Figure 6 should are lacking. It can be said with some definitely proved generally, that at high
be valid for unshielded high-voltage lines confidence that in a well shielded high- altitudes the curves may have different
also. They may be pessimistic to some voltage station, the lightning arrester shapes, having greater slope and ordi-
degree. On the other hand, high-voltage discharge currents seldom will exceed nates even for a higher frequency of dis-
conductors are located farther above the 5,000 amperes, except in the case of a charge.'9 It is believed that the curves
ground and should be subjected to more side flash to a conductor which is un- are valid also for high-voltage unshielded
disturbances of significant maguitude. likely. A theoretical study of the prob- systems. For high-voltage systems well
Lightning arresters in stations that have able occurrence of certain lightning shielded against direct strokes, it is prob-
adequate direct stroke shielding, con- arrester discharge currents based on defi- able that the frequency of discharge is
nected to lines that are also well shielded nite assumptions will be found in refer- less and that the discharge currents sel-
for at least a half mile or a mile from the ence 12. dom exceed 5,000 amperes. The curves
station, will experience high surge currents The effect of line insulation and shield- lie too low for urban distribution systems,
less frequently. The high currents re- ing on the frequency and the crests of on which both the frequlency and magni-
sult from strokes to the circuits near the discharge currents through lightning tude of the discharge currents are lower.6

1951, VOLUME 70 Beck-The Duty on Lightning Arresters for A-C Systems 1137

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
Figure 6 (left). Probable Frequency of occur-
rence of lightning arrester discharge current
10,000 < +V.||||4 4 4 = = { | | | crests
8000 __1 _Abscissae-amperes
6000 t <i1iX X - A } 1 T X DOrdinates-number of years in which a
lightning arrester; or total number of lightning
4000 l l l l X X 4 X X| F sX j/
a drresters of which, in a year, one will dis-
3000 ~~~~~~~~~~~~~~~~~~~~~~charge
a current at least as great as abscissa
/ // Solid lines-All data compiled from Figure
5(B) curve C'
/ // Dashed lines-lightning arresters in stations
only, from Figure 5(A), curve B'
Numbers on curves-assumed total number of
lightning arrester discharges per year per
lightning arrester
4 I2
I____ lsokeraunic level-30
00L l llI I IX t /X III
400 _ _ _ _ __ ___ 2. // _

4
0 ~~~~~~~~~~~~~~~~~~~~~~~~Elevation
not more than 5,000 f.eet
4
200
w
0 oo L L Xable data on a basis of probability as
oi 100 ~~~~~~~~~~~~~~~~~~~~determined
by field test results.
the
4 ao0 pfI ||]||
t ti| t0Q2 rl 153 Excluding the shor front, low current
>_
60 ____ _ __ / / / = __ _discharges, the shortest fronts appear to
w l l XXr-|X|sWa
gbe+ 11/2
+ lmicroseconds.
|ll l l The highest light-
F40 ____ / 7) z 4ffi / ___ _ ning arrester current measured with this
cc 30 } + 4 >5tfl, K tl front is 7,900 amperes,8 indicating an
20 _ _/___
2 / ai j ___ _ average rate of rise of about 5,000
io, oel ~~~~~~amperes per microsecond. That is the
highest rate of rise recorded so far. It is
o0L I I I A 2AXXH = g
|+reported that rates of rise of 43,000
8 === amperes per microsecond have occurred I
fwFH vJ zZ , _ _ _ _ ___ _
6 _ _L in strokes. The most significant curve
of the statistical distribution of wave
4 ~~~~~~~~~~~~~~~~~~~~fronts
is curve C of Figure 9, for above
__1- _ z///of _ lw ___ _ ___ _ 5,000 amperes. From Figure 8 it appears
____ thatE| IIIIII| ;4probability
ithe A W X of occurrence of
> :~o ',
000 ~~~~~~~~~~~~~5,000
amperes per microsecond is two out
of 100 lightning arresters per year in 0

I__ 1 Htisokeraunic
ItIltt tAM-t 'E levels
RES of 30 to 40, on the kinds
AMPERES
-- of systems studied; on ~ ~ subtra'nsmission
° N-I0-d°° 0<Dl 0 o
co
o-° 0 > 006 wood poles with no shielding. The re-
CD¢ 0 ° cording equipment was placed in locations
where past experience had indicated that
lightning disturbances were likely to be
Wave Fronts or Rates of Rise of amperes with fronts of not more than 1 more severe than elsewhere. With a
Lightning Arrester Discharge microsecond. Because of the low crest probability of two discharges per 100
Currents currents these records are not of practical lightning arresters per year on wood pole,
interest in a consideration of the effect on unshielded lines the probability on
lightning arresters or apparatus protected shielded systems will be much less.
Wave fronts of currents discharged by by lightning arresters. The rapid rise of Calculations of the probability of 5,000
lightning arresters in service have been current in the case of a large number of the amperes per microsecond on high voltage,
measured by magnetic surge front re- low crest magnitude discharges results shielded systems12 indicate that it may be
corders. 2 A few have been obtained from the abrupt change in the surge cir- from zero to five per 100 lightning arrest-
with cathode-ray oscillographs.2' Con- cuit produced by the sudden sparking of ers per year, depending on the shielding,
clusions drawn from the available data the lightning arrester series gap. the ground resistance, the insulation
are plotted in Figure 8. They are de- The rate of rise of current in the system levels of the lines, and the lightning
rived from the background data of before lightning arrester sparkover can be arrester ratings. The probability of zero
Figure 95 and additional data accumu- considerably slower. It is thus not rea- applies to highly insulated (1,500 kv,
lated later. Both figures indicate that the sonable to lump all of the recorded wave li/2X40) shielded lines with tower foot-
shortest fronts accompany low currents, fronts together in a statistical distribu- ing resistances of 10 to 20 ohms or less.
such as curve E of Figure 9 for currents of tion culrve and to apply this wave front The higher probabilities apply to lower
less than 1,000-ampere crest, and the curve to crest currents of all magnitudes. insulated (500 kv, 1.5 X40) shielded lines
point on Figure 8 marked 28. This point Some discretion must be used in estimat- with tower footing resistances of 50
siguifies that out of 100 lightning arrest- ing the probable wave fronts or rates of ohms. It is apparent that tower footing
ers, 28 might discharge per year, cur- rise that are significant. Figure 8, com- resistance plays an important part in this
rents whose crests are not more than 500 piled by A. M. Opsahl, presents the avail- matter, and that it should be low, at

1138 Beck-The Duty on Lightning Arresters for A-C Systems AIEE TRANSACTIONS

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
2 FlT T FT T rr 1 C LIGHTNING CURRENTS THROUGH ARRESTERS
I 100 ll
l.IO°° 25 AND 34.5 KV WOOD POLE LINES- NO SHIELD
4
v WIRES-NUMBER INDICATES DISCHARGES PER
T100 ARRESTERS PER YEAR-48 ARRESTER
0.5 ( - POLE YEARSEOD
cr ~~~~~~~~~~~~2 w

cr
Curves show
0.2~~~~~~~~~~~~~~~~~~~~~~~~~ 0

w
0.1 ..6000 2-
0 A 8n m
C. Steel tower lines without wrounow o

a.
40
a.w C 44000 2- 22i
nubradcetmsiud
(3 0.02 8c flstis Fgr8.0 Prbblt ftmst rs o ihnn retrdshrs urnso aiu a
w
~~~~~~~~~~~~~4 o
0.01~~~~~~~~~~~~~~~~~~~~~~~~~~~
A. Wodpl ie ihotgon ie 6 6
2 4 6 ~8 10 12
cl 28 6
4

I___LI___6
6 ____2
10 ___ 12
CREST KILOAMPERES 0L
0 2 ~4 8 14 16
Figure
Stee7 (above).
D.0 Ioe Iie Effect of shielding
wihgon ie on TIME TO CREST - MICROSECONDS
number and crest magnitude of lightning Figure 8. Probability of times to crest of lightning arrester discharge currents of various magni-
arrester currentso tudes
Curves show number of discharges per three
pole lightning arrester year with crest current Times to Half Crest Value of distribution lightning arrester discharges,
at least as great as abscissa Lightning Arrester Discharge the times to half value were greater, the
Woodof pole
A.curren ineetlines without ground wires
hyivlete Currents
aesaedaot4 irscnsaat median
af auvalueewebeing about 40 microseconds
itiuinadsa
C. Sotel tower
Cmintede lines withoutground
towerlinesorathout ground wires
eathode-ray ingdeviceofgreatercurrentsensitivit. and 80 per centThebeing nottomore than 75
microseconds. times half value of
D.steeltogra. lins wthe gretound btwires ITwo instruments have been used to currents in direct strokes are in general
record the general wave shape of lightning longer than of currents in lightning arrest-
least for the first several towers out from arrester discharge current; the fulchrono- ers, although the curves for direct strokes
the station. graph,20 later followed by the photo- and distribution lightning arresters are
There are several considerations that recorder. similar. There seems to be no conspicu
make the rate of rise of arrester discharge chronograph are that the individual links ous reason for the difference in times to
current of interest. They involve the are spaced about 40 microseconds apart, half value between distribution and sta-
discharge characteristics of valve light- so that fast fronts are not readily re- tion lightning arresters shown in Figure
ning arresters and inductive drops in corded. Also, the lower limit of sensi- 10, unless it be the fact that almost all of
leads. The voltage across a valve ele- tivity of the device is about 50 amperes. the records on distribution lightning
ment generally reaches its crest in a Its particular usefulness resides in its arresters were obtained on delta systems,
shorter time than the current. The ability to record wave tails and multiple whereas most of the records on lightning
actual voltage and its rate of rise for any discharges of significant magnitude. arresters in stations were obtained on
practical discharge current can be deter- The photo-recorder is a current record- systems where the neutrals of the source
mined in the laboratory from cathode-ray ing device of greater current sensitivity, transformers were grounded solidly or
oscillograms. With the relation between It can record currents from 0.2 ampere to tnthrough resistance.
lightning arrester discharge current and very high magnitudes. It cannot, how-
voltage known from test, the rates of rise ever, distinguish very short time inter- Total Durations of Lightning
of the currents, if they are of appreciable vals. In this it is less sensitive than the Arrester Discharge Currents
crest magnitude, give an indication of the fulchronograph. The usefulness of the
front of the surge voltage produced at photo-recorder is in its ability to record The total duration of the discharge cur-
the lightning arrester by lightning, currents of long duration and low magni- rents as recorded by fulchronographs and
This may be important from the stand- tude. It does not lend itself readily to photo-recorders are shown in Figure 11.
point of the oscillations that may be set the recording of wave fronts or times to In the case of the total durations re-
up at the terminals of protected appara- half crest value, corded by fulchronographs, the duration
tus. Furthermore, the crest value of the Data from field records on the time is that time to the point where the cur-
voltage across the valve element is a from the beginning of discharge current rent decays to 50 amperes or less, since
funtio ofte ae frierahr hnh to the point where the current has passed the fulchonograph does noreorI cr

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
too Figure 9. Statistical dis- u Figure 10. Times to half
- L.A* tribution of the duration of :) o crest value of all indi-
ax | 1l74 fronts of currents discharged m vidual components as re-
A. Direct
3r <
B.Disnby lightning arresters5 Acorded by fulchronographd
I .c 80

D(n I w d5
.60
B e 60 -
_j

w~~~ U> --~ ~ ~ ~~~~~z


~~~~~~~~~~~~~~~~~~~~~~~~~D
c

cr Z >< cr~~~40 -
W~~~ ~~~~~~~~W
Fn 0 ~~~~~~~~~~~0
U_ w ~~~~~~~~~~~~~cn
20

:3~ ~~~~~~~~~~~~i
Discargs0wth recrd exeee
rss-ad5prcn 4,0
a exee irscns
4,0 ln
ico2 uaio hnhg urnsohr

ld B end 1,000ance g 5d 0 . e 2 orded wih tthcres ove 1,000 ae reg s onl


ae.thatcheearges brestaisseow5
irect likeyto
strokesB ampresi, 2 and 5 per- centmaexceed7,000 microse- distribution
tharoug duain2wt lightning 2h3smarresters.
hrg.Ti
A.t Direcate nurladteohr Aestrokss b.tee Ditributolihnngro arredisters,b8ureordsys
B.aton
All aretr recoer dischre
lightning Tecuob'icagdb ihnn C. Lihtin c rrestersinattatlnastdlomrecords,
c.rrAbovem5,000ampertes,
iAmpeesofhnn 17
amperes . The recordstes arsteons.
phoitou- oning arresterThseviemaximum durterionse- has D. lightning
been arresters brecatoRauecords
indica ed inaginlstatos. troes
reord withdalightain
isolate neutral
wrecobtnerdsan on the other
twoesys- rerestercee ,00mcoeodsbetweng
cre hanOn drthen
th, doterhand uhngrounde
itwslae distribution sys-t
fboundth
B.ceptbothrib
tems,
lihatningrod arrser shded;wone Chom s Dt p waischargednt
pole, mrecurrents th of arraction
C.tisong
Lighnin e on the Wcnsnabye
in sttos mthsloest
colmbadlbut N records,ohrta occurred
bandi
The urvsfrecordens consinnin wheraalbe no longeritantecord
t ted cuie nrsett olmsaddm
grounderd tre lou reosis tac elonger mond o temorent one superosed dur and a69-ky ciircuit tmheshame pls.
durations
strokeof d of lightning arrester discharge The coulombs discharged by lightning It was concluded that at least some of the
currents may bereiordoexpected
evrlts
on such systems
elsewhere
arresters
butlo
in
50 pner entofd theotaneWiscon-srior unwerabe
service can be determined by lightning arrester
tsoaelwitcurretsgo
damage was caused by
thousnds f miroseonds tha thee sin ricord e xdceeded colmicrsecondsght
4,00
than on those effectively grounded.'5 integration of the discharge current this. Furthermore, it iscurresents.lon mdurain tihanhingh ohr
possible that the
The fulchronograph data were obtained waves. In some cases they have been coulombs registered may have been influ-
on various systems, solidly grounded, iso- recorded by means of the magnetic inte- enced by such crosses. The Wisconsin
lated and resistance
amemnaresiof grounded.Thesisotovty grator."i
some duaigson. onds.flhrngah.O h data therefore
hascon-ncoulombs edasybeenoreporryd
cannot be regardedtxests.o
as con-
Frecorderbdakaroere
The curves for currents
obtaie on lightning
in two
Wscon- Onitiuinssemegtrcrswr
The available data indicate that the disharesclusive, inoftferrespect onents
hand, to coulombs and dam-
lateond thort
tesin Pboth
arresters and directCopole,unsieled
Servic strokes are similar, one charge
Cuo mbsneinidisaigcharged
most cases varied Lgtning on to distribution
between a drteWio,rscotngsin
grete age lightningaocurredo
sstmallsse arresters.
except
with that there
apeasothate netral
maximu
are thesothcur-
more currents in cuob
small fraction of a coulomb and about No
Aretheeowhcbalyreahe records
betwee, the other than those obtainedsys-
ugreaopepndedisribution in
grounteoded
strokes throug
wa rstancees,
of durations of several tens Longer of one coulomb.
olmb.Tooteels One record was obtainedthretoand cargek
Wisconsin ocicuit
were durng the
associated with
same poflthe
damage
thousands of microseconds, than there are which indicated 3.7 coulombs in a light- to modern lightning arresters.
in lightning arresters. The longest dura- ning arrester in a station on an isolated The charges involved in direct strokes
tions in lightning arresters were found on neutral system. It is, however, not cer- are considerably higher.'61 About 50 per
a distribution system of the Wisconsin tan that this record was one discharge or cent of the strokes have at least 9 cou-
Public Service Corporation, a delta sys- consisted of more than one superimposed lombs, and a total charge of more than 160
tem in a region of high soil resistivity, on the fulchronograph. On the Wiscon- coulombs has been reported. Except for
From the background data of the Wiscon- sin distribution system eight records were discharges of few components and short
sin Public Service Corporation curve,2" obtained indicating charge greater than 1 duration, resulting in small amounts of
it appears that the maximum crest cur- coulomb, three of which barely reached charge, the great preponderance of the
rent recorded was 6,000 amperes, and 1.5 coulombs. Two of these last three total charge occurs during the flow of the
that it was the discharges of relative low were associated with crest currents of less continuing stroke current of low magni-
crest magnitude, 2,000 amperes or less than 1,500 amperes and the third with a tude. The high current peak portion of
t'hat hadr the longdurain.cetcretoslgtyoe6,0am astoecretmyb1000aprs

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
Figure 11. Statistical distribution Coulombs. Less in lightning arresters.
of the durations of all components Number significant components. Less
in lightningofarresters.
A w \ <
of records of lightning arrester
B
~~~~discharge currents5
so 5 \4c \ \ |\, ' C Switching Surges
D \ \ \ . D When a line is switched, a transient
4 \ X \\ \\ voltage is likely to occur. These volt-
60 ages may be from one to several times the
E
\ \\ \A_line to neutral voltage of the sys-
normal
tem. Usually they are not high enough
1:
4__ _____\
\ \ s \ to discharge lightning arresters, or en-
danger apparatus insulation. If they ex-
w \ \ \\ \ ¢,F ceed the voltage required to spark the
zs \ \ \\\ 5i s,_G lightning arrester gap, the lightning ar-
rester must dissipate all or part of the en-
o<20 \ss \ \ \\ \ \ ergy in the switching surge. This en-
ergy, very generally speaking is 0.5CE'
\ where E is the crest of the switching
surge voltage. If a lightning arrester
discharges on a switching surge, the dis-
10 100 1000 lOpOO 100,000 charge current is of low crest magnitude
DURATION IN MICROSECONDS
with a long tail.* It may be said to be
I. As measured by fulchronograph, sensi- similar to a low, long duration, lightning
tivity 50 amperes: discharge.
A. Direct strokes,"6 123 records were serviced on the average of once a
B. Lightning arresters in stations. All week and usually after every storm. Power Follow Current
records, 268 Comparison between the sets of data
C. Wisconsin Public Service Corporation, obtained with the fulchronograph and the A distinction must be made between the
distribution lightning arresters, 56 records photo-recorder show more components duty imposed by power follow current on
D. Other distribution records, 41 for the photo-recorder. This may indi- valve and expulsion lightning arresters.
E. All distribution records, 97 cate components of less than 50-amperes In the case of the valve lightning arrest-
11. As measured with photo recorder, crest, not detected by the fulchrono- ers, the magnitude of the power follow
sensitivity 0.2 ampere:'
F. Direct strokes,'6 30 records graph, or some of the apparent compo- current is independent of the short circuit
G. Lightning arresters in stdtions, 137 nents may be the result of system disturb- capacity of practical systems. It is de-
records ances such as those mentioned in the sec- termined by the 60-cycle voltage across
tion under Frequency of Lightning Ar- the lightning arrester terminals and on the
rester Discharges. As far as duty on the relations of the instantaneous values of
Multiple Discharges lightning arresters is concerned, the ful- the surge and 60-cycle voltage across the
chronograph curve is considered the sig- lighting arrester during discharge. The
Multiple or repetitive discharges are nificant one. On this basis, about half latter factor will influence also the dura-
those that have more than one impulse or the discharges may have more than one tion of the power follow current. Usually
well defined crest occurring in the same component and 5 per cent may have more the duration of power follow in a valve
disturbance. The several crests are called than seven. lightning arrester, if the 60-cycle voltage
components. Between these components applied to it does not exceed the lightning
the current ray actually cease, or there General Comparison of the arrester voltage rating, is less than a half
may be a continuing current of low magni- Characteristics of Currents in cycle. It may vary between zero and al-
tude. The available data on the statisti- Strokes and in Lightning Arresters most a half cycle depending on the already
cal distribution of the numbers of compo- mentioned relation of surge and system
nents in lightning arrester discharges are . . . voltage waves. Lightning arresters usu-
given in Figure 12. It is possible that rest magnstude. Considerably less In ally do not have 60-cycle voltage equal to
some of the components considered to
occur in the same discharge actually oc-
lgtning arreters. In gene
ratio is about one to ten
their rating impressed on them, except in
laboratory tests or during abnormal sys-
curred in separate discharges. However, Duration of front. About the same in tem voltage conditions. Records ob-
after considering the probability of this it lightning arresters as in strokes. tained with the photo-recorder indicate
is believed that the curves of Figuire 12 Rate of rise of current. Less in light- that occasionally the powerfollow current
are reasonably representative. The data ning arresters. Ratio is about one to ten. may persist for more than a half cycle
plotted in Figure 12 were obtained on ... without damage to the lightning arrester.
lightning arresters in stations. The data Times to half crest value. Less in light- The background data obtained with the
secured on distribution lightning arresters ning arresters in stations than in strokes. photo-recorder, used to derive the curves
are ot nclded Siee t ws nt pac- Roughly the same in distribution light- of Figure 11, show that only 25 per cent
tical to service the distribution lightning ning arresters as in strokes. of the component discharges had power
arrester installations as frequently as Total duration. Fulchronograph rec- follow durations approaching a half cycle
those in stations, the distribution records ords less in lightning arresters in stations or more of 60-cycles; 75 per cent were
were not considered sufficiently signifi- than in strokes. Slightly less in distribu- less. On one system, with isolated neu-
cant to use. The installations in stations tion lightning arrester than in strokes. tral, three records indicated power follow

1951, VOLUME 70 Beck-The Duty on Lightning Arresters for A-C Systems 1141

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
Figure 1 2. Statistical distribution of 100 quate without 60-cycle voltage. Because
numbers of components in records of < the operating characteristics of valve and
lightning arrester discharge currents' u5 expulsion lightning arresters differ, they
LLUcn
A. Measured with photo recorder, sen- ° I 80 should be considered separately. The
sitivity 0.2 ampere, 31 records z most significant tests are not the samefor
B. Measured with fulchronograph, sen- 2 \ both.
sitivity 50 amperes, 228 records Z -' 60 Since the nature of the surge current
C. Direct strokes'4 I. may be a factor in the valve lightning ar-
3z t t! - A rester's ability to handle the power follow
current of three half cycles and six of two Ot, 4Q B current, its operating cycle should be ini-
half cycles. On a similar system grounded o '\
through resistance and located in a dif- -i -f5
ferent part of the country, two records of ° 0 ;>
amost three half cycles and two of two a: w 40
half cycles were obtained.' This may i - -
result from a long continuing discharge, or , - 4 8 2 6 20 24 28323
a multiple discharge, or restriking in the 4 8 1 6 2 4 2 23
lightning arrester. Restriking in the NUMBER OF COMPONENTS
lightning arrester may be caused by 60-
cycle voltage across the lightning arrester Basis for Testing Lightning Arresters tiated by a surge that conforms to a rea-
in excess of its rating, not unlikely on iso- sonably severe service condition, and
lated or resistance grounded systems. A laboratory or routine test whose pur-l which can be reproduced readily with
pose is to verify the ability of a lightning available surge testing equipment. The
Generalized Wave Shape of arrester to withstand severe service 60-cycle voltage applied to its terminals
Lightning Arrester Discharge should involve discharging the lightning should be equal to the lightning arrester
Currents arrester by a surge of some severity while voltage rating. The short circuit capa-
the lightning arrester is connected to a
ity of the 60-cycle source should be suf-
It is evident from the recorded data source of 60-cycle voltage so that power
ficient to maintain at least 90 per cent of
that the characteristics of the surge cur- follow current will flow. This is the so-
the open circuit voltage during the flow of
rents discharged through lightning ar- power follow current. This conforms called operating duty cycle test referred
resters in service vary widely. There is to in AIBE and NEMA standards for
with existing standards. In view of the
nofixeddutyforthesedevices. Figure13 lightning arresters. Tests with surge
severe duty imposed on valve lightning
is an attempt to present a generalized current alone, with the lightning arrester
arresters by longsurges, the discharge cur-
wave shape with values that are thought not connected to a 60-cycle source do not
rent used should have a long duration.
torbesignificant. It is quite possible that have sufficient significane. Tests made
The most substantiallightning arresters
when more data accumulate changes in to verify the impulse protective charac-
are those identified as station type light-
the values may be in order. For the time ning arresters The dutyeycle tests onteristics of lightning arresters, that is, the
being Figure 13 seems a representative theseshould bemore rigorous than online
voltages that appear aross their termi-
aggregate of the expected discharges. or distribution types. AIBE and NEMA nals when discharged by surges, are ade-
standards presently prescribe a duty cycle
test for station lightning arresters of 30
Idischarges with 60-cycle voltage con-
that the charctersc oFigure 13. Generalized wave nected, using a surge current of 10,000
l\ ~~shape of lightning arrester amperes crest and 10 by 20 microsecond
2p follodischarge currents wave shape. This test is not representa-
rents discharged through lightning ar- called operatingdutycycletestrefered witive of the conditions with which lightning
resters in service vary widely. There is to in AIEEandNEMAstandardsfor arresters of this type may have to contend
nofixeddutyfor these devices. Figure13 lightningarresters.Testswithsurg because the duration of the surge current
F-
+ H~~~~~~~~~~~~~~no
TIME
not long enough. The present test is
longer a criterion that distinguishes be-
~~~~~~~~~~~~~~~iS
is an attempt to present a generalized currentalone,withthelightnintween lightning arresters of the station
wand other types, nor is it indicative of
I. Crest currents from less than 100 amperes up what modern station type lightning ar-
resters will do. The standards also re-
Distribution Station quire a surge discharge of 100,000 am-

1,500 a or more
Median alue 83 a
32%d 27%
710- peres crest of 5 by 10 wave shape with no
60-cycle voltage connected. This is im-
10,000 a 2.5 1.15 pressive but not as significant as a lower
0.2 20,000 a 0.6 current of relatively long duration with
60-cycle voltage. It is suggested that this
F. Fronts, 1.5-25 microseconds (dS). Durations of fronts increase with currents requirement for valve lightning arresters
maximum recorded rate of rise, about 5,000 amperes per ,us
H. Times to half value from less than 10 to 75 ,us. Median value for arresters in be discarded.
stations, 26 Ms, for distribution lightning arresters, 40 ,us This paper proposes that a different dis-
T. Total signiicont durations 40-10,000 s. Medin for lightning arresters in charge current be used for the purpose of
stations, 80 <s5 for distribution lightning arresters, 200 fs verifying the ability of station type light-

1142 Beck The Duty on Lightning Arresters for -C Systems AIEE TRANSACTIONS

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
ning arresters to stand up under the de- the 30 operations be retained. It is rec- check the protective characteristics of
mands of service, and to classify the ommended that the firing angle be be- valve lightning arresters. For this pur-
several types of lightning arresters more tween 15 and 0 degrees of the crest of the pose it is believed that the 10 by 20 micro-
accurately than is now the case, so that 60-cycle wave of the same polarity as the second waves now prescribed in standards
purcbasers may have a clear picture of surge. are adequate.
what they are specifying. The proposal made has the benefit of In the case of expulsion lightning ar-
Consideration must be given to labora- considerable experience."5 The discharge resters, the power follow current is of ma-
tory facilities and problems. It is sug- current described above has been used for jor importance. It is influenced by nu-
gested that the crest magnitude of 10,000 15 years in making routine quality con- merous factors such as the short circuit
amperes be retained, but that the shape trol duty cycle tests on valve elements of capacity and recovery voltage of the sys-
of the discharge current be changed to a Autovalve station type lightning arrest- tem, the location of the lightning arrester,
wave which reaches crest in a time be- ers. its characteristics, the power factor of the
tween 5 and 15 microseconds and decays The field experience with lightning follow current circuit including the light-
to half crest value in 45-60 microseconds. arresters built with such surge tested ning arrester, and the electrical angle of
The time to half value corresponds to the valve elements has been excellent. It is the 60-cycle voltage at which it is fired.
median found for distribution lightning of interest that on earlier constructions of These conditions vary in service. The
arresters, which is about 50 per cent these station lightning arresters, not subject is discussed in reference 1. The
greater than that for lightning arresters in tested in the manner described but able present paper does not attempt to make
stations. easily to meet the requirement of the any recommendations except that the
In power follow testing the total dura- 100,000 ampere 5 by 10 wave, sufficient duty cycle tests for expulsion lightning
tion of the wave through the lightning ar- cases of damage occurred to cause some arresters be different from those for valve
rester-will:be determined both by the concern. Field research then indicated lightning arresters.
surge circuit constants and by the 60-cycle the importance of long tailed surges of
power source in parallel with the lightning lower magnitude and led to the adoption
arrester. An energized transformner of a rgorous routine duty cycle testing Appendix 1. The Probability
shunting the arrester is a circuit with in- procedure. Curves Figures 5(A) and 5(B)
ductance and resistance and will dis- To summarize the proposed test crite-
charge some of the long tail of a surge."5 rion for station .. lightning
type .. s
arresters,
. '
it
~~The method of plotting the curves of
In defining the total duration of the surge is suggested that it be a duty cycle test, Figures 5(A) and 5(B) with their confidence
discharge through the lightning arrester with the test specimen connected to a envelopes, the meaning of the confidence
this must be taken into account. There- source of 60-cycle voltage equivalent to envelopes and the conclusions that can be
fore it is recommended that the source of its rating. The number of operations drawn from such curves are discussed in
power follow be a two winding trans- suggested is 30, with the firing angle 15-0 references and 18. ifNormal
tribution is17indicated fall ondis-a
the datachance
former. The technique of determining aheadahead
degreesdegrees off therestothe6 ycle
the crest of the 60-cycle straight line when plotted on probability
the total duration is to connect the sec- voltage of the same polarity as the surge. graph paper. The distributions of the
ondary of the transformer to the test spec- The proposed discharge current is one amperes of crest current, curves A, B, and C
imen's terminals, to disconnect the pri- with a crest of 10,000 amperes, a front of of Figures 4 and 5(A) and 5(B) are not
to short mirstraight
lines. However, the distributions
mary from the source of power, to short
circuit the primary to simulate a source of
5-15 microseconds,
mary from the a time
source of to
talf
power, value

45-60 microseconds, and a total duration


of of the logarithms of the currents, A', B',
and C' are straight lines and therefore indi-
low impedance, and then to discharge the determined as discussed earlier of not cate a normal distribution. In Figure 5(A),
surge generator into this circuit. The less than 600 microseconds. the high currents appear to deviate from the
duration of the discharge through the The test is suggested as a design verifi- line A'.
lightning arrester can be measured with a cation test. It is not recommended that bility level of 0.95, shown by the dashed
cathode-ray oscillograph either by record- test specimens subjected to it be put in lines above and below the curves of the
ing current or voltage. It is recom- service, because of the severity of the test. logarithms, are derived from the data'8 and,
mended that measured in this manner, the The life of a lightning arrester subjected to as the term indicates, denote the degree of
confidence with which the data may be re-
surge current actu-
total duration of thetoalduatonofth atu srg uren scduy yle will
such duty cycles not bbe uliitd
wllno unlimited. garded. They signify that in 100 investiga-
ally discharged through the lightning ar- For line and distribution lightning ar- tions of the kind made, the data from at
rester be not less than 600 microseconds. resters it is suggested that a similar, modi- least 95 would fall within the confidence
Testing with such surges has indicated fied, surge be adopted. It might be a zone, or that for one additional set of data,
that with the transformer disconnectedr
* >- lower current, a shorter time . to half value ~~~~~~~~~~~~~~~~the
chances would be 19 to 1 that it would
fall within the zone.
from the lightning arrester, the total dura- and duration, or fewer discharges. Al- The confidence envelopes for A' and B'
tion of the discharge is well over 2,000 though the field data indicate that the do not overlap at currents higher thai' about
microseconds. probability of severe duty may be greater 100 amperes. Therefore it may be assumed
that there is a significant difference between
During the actual duty cycle test, the in the case of distribution lightning ar- these twosetsof data. A', B', and C' being
short circuit is removed from the trans- resters, it is not practical to expect dis- straight lines, extrapolation to higher cur-
former primary and it is connected to a tribution or line lightning arresters to rents than any recorded by extending the
60-cycle source. The present standards withstand tests as severe as those which lines appears justified. However, a ques-
specify 30 operations. From Figure 6 station lightning arresters will handle. t.°iois posed by the observt.aion th5at thne
this number may appear excessive. The station lightning arresters are ex- fidence zone. It appears that the loga-
However, probabilities are involved to a pected to be the most durable and most rithms for currents higher than 20,000
high degree, multiple discharges and effective arresters that can be made eco- amperes may have a different distribution,
switching surges occur, and modern ar- nomically. The others are smaller and on another of
lying probability straight line A N, and that
'
resters are quite able to handle the situa-
.
less expensive, ~~~~~~~~~~~~~~~~their
hnoewudepc.rmtecnieal occurrence may be less
tion. Therefore it is recommended that The foregoing tests are not intended to vrolume of data at lower c-urrents. There is,

19S1, VOLUME 70 BeckThe Duty onLightning Arrestersfor A-C Systems 1143

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
on the other hand, the possibility that the actions, volume 70, part I, 1951, pages 1-7. (AIEE Transactions), volume 54, August 1935,
high currents recorded reached the limit of 2. REPORT ON STANDARD FOR VALVE AND EXPUL- pages 837-43.
sensitivity of the recording devices, and that SION TYPE LIGHTNING ARRESTERS, AIEE Standard 15. LIGHTNING AND LIGHTNING PROTECTION ON,
the actual currents may have been higher number 28A, February 1950. DISTRIBUTION SYSTEMS, R. C. Bergvall, E. Beck.
than indicated. This would tend to put the 3. SURGE CURRENTS IN PROTECTIVE DEVICES, T4a42
AIEE sactio-9, volume 59, 1940, pages
measurements into the confidence envelope. A. M. Opsahl.volume
Electrical Engineering (AIEE
54, February 1935, pages 16. THE MEASUREMENT OP LIGHTNING CURRENTS
Transactions), IN DIRECT STROKES, G. D. McCann. AIEE
All of the data are gathered together into 200-04.
one curve, C', in Figure 5(B). The same 4 FIELD INVESTIGATION OP TEE CHARACTERISTICS Transactions, volume 63, 1944, pages 1157-64.
tendency is discernible at the high currents, OF LIGHTNING CURRENTS DISCHARGED BY ARREST- 17. GRAPHICAL STATISTICS-AN ENGINEERING
although now the points lie within the con- ERS, I. W. Gross, G. D. McCann, E. Beck. AIEE APPROACH, L. R. Hill, P. L. Schmidt. Westing-
fidence zone for C'. Apparently, in the Transactions, volume 61, 1942, pages 266-71. house Engineer (East Pittsburgh, Pa.), March 1950,
region above 20,000 amperes some specula- 5. FIELD RESEARCH ON LIGHTNING ARRESTER page 121.
tion will be involved until there is available a DISCHARGES, G. D. McCann, E. Beck. AIEE 18. GRAPHICAL STATISTICS-AN ENGINEERING
larger volume of data at those currents. Transactions, volume 66, 1947, pages 625-29. APPROACH, L. R. Hill, P. L. Schmidt. Westing-
However, some conjectuires seem warranted. 6. DISCHARGE CURRENTS IN DiSTRIBUTION AR- Engineer (East Pittsburgh, Pa.), May 1950,
house 157.
At cnd Figures 5(A
and ,5(B)
Figures 5(A) and ing RESTERS, K. B. McEachron, W. A. McMorris. Elec- page
After considering
the possibility of a lower probability of cur-
'), trical Engineering (AI EE Transactions), volume
57, June 1938, pages 307-14.
19. LIGHTNING INVESTIGATION AT HIGH ALTI-
TUDES IN COLORADO, L. M. Robertson, W. W.
rents above 20,000 amperes than indicated 7. LIGHTNING INVESTIGATION ON A RURAL Dis- Lewis, C. M. Foust. AIEE Transactions, volume
by A' or C', and the limitations of our pres- TRIBUTION SYSTEM, D. A. MacCarthy, D. A. Stann, 61, 1942, pages 201-08.
ent knowledge, it is concluded that C' is D. R. Edge, W. C. McKinley. AIEE Transac- 20. NEw INSTRUMENTS FOR RECORDING LIGHT-
reasonably representative of the distribution tions, volume 68, 1949, pages 428-37. NING CURRENTS, C. F. Wagner, G. D. McCann.
of the lightning airester discharge current 8. LIGHTNING INVESTIGATION ON THE 25 KV AIEE Transactions, volume 50, 1940, pages 1061-8.
crests for all lightning arresters. Curve C' SYSTEM OP THE WEST PENN POWER COMPANY, 21. FIELD INVESTIGATION OF LIGHTNING SURGES
AT SUBSTATIONS, I. W. Gross, G. D. McCann.
.tions, volumeBowen,
William C. Edward Beck. AIEE Transac-
passes through
passs troug th
the loaritm correspond-
logarithm 55, corespnd- 66 1947 pag 831 6.AIEE Transactions, volume 66, 1947, pages
ing to 100,000 amperes, at the probability of 9 LIGHTNING CURRENTS IN ARRESTERS IN STA 569-76.
about 0.007. This is seven in 100,000 or
one in 14,300 discharges. Since the con-
.TIONS, I. W. Gross, W. A. McMorris. AIEE
Transactions, volume 59, 1940, pages 417-22.
22. Discussion by A. M. Opsahl of BASIC IMPULSE
INSULATION LEVELS FOR HIGH VOLTAGE SYSTEMS,
fidence envelope boundaries pass through 10. LIGHTNING INVESTIGATION ON 33 KV WOOD J. E. Clem, J. R. Meador, W. J. Rudge, A. H.
the probabilities of 0.012 and 0.005 at this POLE LINES, F. E. Andrews, G. D. McCann. Powell. AIEE Transactions, volume 69, part II,
same logarithm, the probability might be, AIEE Transactions, volume 64, 1945, pages 768-77. 1950 pages 953-60-
in 95 cases out of 100, 12 or five in 100,000. 11. AN EIGHT YEAR INVESTIGATION OF LIGHTNING 23. FIELD INVESTIGATION OP LIGHTNING, C. F.
Considering the trcnd indicated by the line CURRENTS AND PREV}ENTIVE LIGHTNING PROTEC- Wagner, G. D. McCann, E. Beck. AIEE Transac-
A' of Figure
A Figure
of 5(A), even the probability of TION ON A TRANSMISSION SYSTEM, E. Hansson,
S. K. Waldorf. AIEE Transactions, volume 63,
tions, volume 60, 1941, pages 1222-9.
24. Discussion by E. Beck of TILE MEASUREMENT
five discharges of 100,000 amperes or more 1944, pages 251-8. OF LIGHTNING CURRENTS IN DIRECT STROKES,
in 100,000 may be too high. Nevertheless, Di b C F W G. D. McCann. AIEE Transactions, volume 63,
it is probably safer to err on the high than INSULATION LEVELS FOR HWgerVof BASIC IMPULSE 1944, pages 1157-64.
on the low side, consequently C' is assumed J. E. Clem, J. R. Meador, W. J. Rudge, A. H. 25. MEASURED LIGHTNING CURRENTS THROUGH
to be the distribution. Powell. AIEE Transactions, volume 69, part II, DISTRIBUTION ARRESTERS ON 4800 VOLT RURAL
1950, pages 953-60. CIRCUITS, H. R. Wilbur, W. A. McMorris. General
13. THE SURGE CREST AMMETER, C. M. Foust Electric Review (Schenectady, N. Y.), voltume 44,
References H. P. Kuehni. General Electric Review (Schenec- March 1941, page 159
tady, N. Y.), volume 35, 1932, page 644. 26. LIGHTNING CURRENTS MEASURED, H. W.
1. THE POWER INTERRUPTION TESTING OF LIGHT- 14. LIGHTNING CURRENTS IN FIELD AND LABORA- Collins. Electrical World (New York, N. V.),
NING ARRESTERS, Otto Ackermann. AIEE Trans- TORY, P. L. Bellaschi. Electrical Engineering May 12, 1934.

Discussion fection is required. We, therefore, con-


sider it essential to maintain at least the
arresters in Mr. Beck's paper, one comes to
realize that expulsion lightning arresters,
present 65,000 surge ampere requirements built to handle power flow are not sensitive
W. A. Sumner (Westinghouse Electric Cor- for these distribution-type expulsion light- to surge discharges. On the other hand,
poration, Sharon, Pa.): We are glad to note ning arresters. current flow in the expulsion type is ac-
that Mr. Beck recognizes the merit of the companied by the consumption of gas pro-
high surge current test for expulsion light- ducing material. Therefore, the weight of
ning arresters and finds that it is a good 0. Ackermann (Westinghouse Electric the question of service duty is shifted from
criterion for this type of lightning arrester. Corporation, East Pittsburgh, Pa.): Only a the characteristics of lightning surges to the
From a close field follow of expulsion light- rather short section of Mr. Beck's paper is frequency and character of power follow
ning arresters mounted in distribution trans- allotted to power follow, mainly because operations.
formers, representing some 2,500,000 light- valve-type lightning arresters limit this cur- Regarding lightning arresters for dis-
ning arrester years and representing a rent to a very small amount compared to tribution service, the sparkover of expujl-
nation-wide distribution, much of which ex- their surge discharge capacity. That the sion and of valve types is nearly enough
perience was associated with rural lines matter is quite different with expulsion-type the same to warrant the conclusion that ex-
having severe lightning exposure, we agree, lightning arresters is indicated but briefly by pulsion lightning arresters operate as fre-
in general, with the trend as shown for the the author. The subject is, however, all im- quently as valve arresters which, according
"Probable Frequency of Occurrence for portant as far as the service life of expulsion to Mr. Beck is, on the average, about once
Lightning Arrester Discharge Current lightning arresters is concerned; therefore I per year. Another source of information, re-
Crests," as set forth in Figure 6. However, wish to enlarge on this point at least in a porting expulsion lightning arrester opera-
our data would indicate that the frequency discussion to the paper on lightning arrester tions only, is the lightning investigation
of occurrence of the heavier currents in the duty. paper by McCarthy and others.1 Based on
range of 65,000 to 100,000 amperes might be Expulsion lightning arresters do not limit experience covering 4,054 expulsion lightning
appreciably greater than shown in the curve, power follow current nearly as much as the arrester years, that paper states that, on the
It is true that one stroke per year at 65,000 valve type; in fact, many of them limit it average, an expulsion lightning arrester
amperes for each 4,000 lightning arresters, hardly at all. The latter in particular must carries power follow 0.65 times a year, and
or 0.025 per cent per lightning arrester year, therefore be designed to handle full system that this power follow is restricted to one
is a relatively small figure. However, when fault current for about one cycle at a time. half cycle or less in all but 1 per cent of the
one considers the volume with which these For a 1,000-ampere rms fault, the flow of cases. It is gratifying to note that the re-
expulsion lightning arresters are mann- charge per cycle amounts to 15 coulombs. sults of these two separate investigations are,
factured for distribution transformer appli- If one compares this with the figures re- as I am to show, surprisingly much alike.
cation, it is realized that a high state of per- ported to be critical to valve-type lightning It is well known that expulsion lightning

1144 Beck-The Duty on Lightning Arresters for A-C Systems AJEE TRANSACTIONS

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
arresters discharge surge currents without with which they may be expected on posed in Figure 13 of the paper. It is in the
power follow if the surge occurs at times of shielded and unshielded lines, is of con- application of this test wave that some ques-
low instantaneous power voltage. With siderable interest. Figure 5(B) curve C, for tion may arise.
many types of expulsion lightning arresters example, shows the distribution of crest The proposed timing of this impulse with
this region roughly covers the time within 30 currents in valve-type lightning arresters in respect to the 60-cycle voltage does not differ
electrical degrees of voltage zero, which is stations and at distribution transformers. essentially from the present standard duty
one-third of the total voltage cycle. As This figure shows that 90 per cent of the cycle test, that is, with the impulse and 60-
random surge discharges are distributed uni- surge currents discharged through lightning cycle voltage of the same polarity. This
formly over the range of the voltage cycle, arresters are 5,000 amperes or less. Another may have been chosen originally because it
and as most of them last only a short part of measure of this 5,000-ampere co-ordination was thought to be the most severe condi-
such a cycle, one must conclude that only criterion is its frequency of occurrence. This tion, but it is known now that this is not
about two-thirds of them cause power fol- is given in the paper as once in 13 years. necessarily true.1 Furthermore, natural
low. Hence, the 0.65 figure for yearly The analysis regarding the probabilities lightning is apt to strike at any point on
power follow operations per lightning of having a rate of rise of 5,000 amperes either polarity of 60-cycle voltage. Would it
arrester indicates that the total surge dis- occur is of considerable interest. However, not be more representative to space the
charges per unit are about one per year; the probabilities of having from 0 to 5 such timing angle of each successive operation
this checks Mr. Beck's estimate. rates of rise on shielded lines and 2 such some value such as 15 degrees.
The American standards consider an rates of rise on unshielded lines per 100 As in the existing standard duty cycle
expulsion lightning arrester to give satisfac- lightning arresters per year, do not appear test, Mr. Beck is concerned with the ability
tory service if it is in acceptable operating to be consistent. Perhaps more discussion of the valve element to withstand service
condition after five test operations of the of this point would clarify it. conditions. The requirement of a high
most severe kind. This most severe test Having a current discharge through a fault current 60-cycle supply circuit im-
condition is approached only if the lightning lightning arrester in a station of 65,000 am- poses the most severe duty on the valve
arrester discharge starts at an early moment peres once in 4,000 years has been calcu- element, but not necessarily the gap ele-
in the voltage cycle. For any other moment, lated and is shown in Figure 6 of the paper, ment. In such a circuit the follow current
the follow current is lower in amplitude and An analysis of the field data made for flow is limited almost entirely by the light-
shorter in duration. Therefore, lightning another purpose corroborates this. That is, ning arrester valve resistor, and the circuit
arrester life in terms of power follow opera- the records over many years show that only recovery voltage appearing across the gap
tions is considerably longer with random 0.5 of 1 per cent of the currents discharged element at follow current zero is very low.
firing than with firing timed for maximum are 30,000 amperes or more. On this basis, In a lower fault current circuit the higher
severity discharge, even if the lightning the use of duty cycle tests on valve-type circuit reactance causes a greater disparity
arrester is installed at a location of maxi- lightning arresters of 65,000 or 100,000 am- in time between follow current zero and 60-
mum permissible fault current. If one peres is much higher than necessary. cycle voltage zero. The circuit recovery
appraises this effect by the factor of 2.5, the The statement is made that "both Figures voltage, though still very low, is larger than
multiplication of all factors involved indi- 8 and 9 indicate that the shortest fronts ac- in the high current test circuit. The multiple
cates that an expulsion arrester, installed company low currents." This is apparently gaps' ability to withstand these very small
at the most critical location permitted by its based on 25 records shown in curve E of recovery voltages determines the lightning
rating, has an average life of 19 years (1/0.65 Figure 9 and on the 28 points with less than arrester's ability to interrupt 60-cycle follow
X 5 X 2.5). Now, remembering that ex- 2-microsecond fronts and magnitudes below current.
pulsion lighting arrester life increases about 2,000 amperes. This does not appear to be A considerable proportion of lightning
as the ratio of the maximum current rating sufficient data to use for such a conclusion. arrester failures in stations have been as-
to the available fault current, it is evident The crest value of the voltage across a scribed to excessive system voltage. This
that, on the average, an expulsion lightning valve element is a function of the rate of rise implies gap failure, admittedly under con-
arrester has a service life well over 20 years of current rather than of the current crest ditions of abnormally high follow current
just like any other piece of properly designed value. Because of this characteristic be- through the valve resistor element. Since
and applied distribution equipment. havior, what is the magnitude of the dis- severe system voltage conditions as well as
It would, of course, be assuring if this charge voltage across a 37-kv station-type abnormal lightning conditions impose severe
contention could be proved by some sort of lightning arrester, for example, when it is duty upon the lightning arrester, should we
life test. I feel that this can be done and, subjected to a 5,000-ampere per micro- not consider a test which explores lightning
therefore, I wish to conclude this discussion second, 5,000-ampere crest current, com- arrester performance under more severe re-
by calling attention to what I consider a pared to what it would be on the standard covery voltage conditions. Such a test
significant life test for high current rating 5,000-ampere, 10-microsecond front-test could be an additional combined impulse and
expulsion lightning arresters, applying to wave? power voltage test using not a high fault
ratings of 6,000 amperes or more. It is test- The proposed current, wave shape, and current circuit, but one whose short circuit
ing at 90 per cent of rated lightning arrester duration duty cycle test is a 10,000-ampere impedance was at least 3 ohms per kilovolt
voltage, and advancing the moment of firing wave which reaches its crest in 5 to 15 micro- of arrester rating, predominantly reactive.
by 80 electrical degrees for successive dis- seconds and decays to half tail value in 45 to Crest magnitude of the impulse current
charg.es. Tests of this nature are reported 60 microseconds, which is applied for 600 would be unimportant, but total duration
and discussed in the paper on the spiral microseconds, with a 60-cycle power follow should be in the order of 100 microseconds to
groove lightning arrester2 which also is being source connected to the lightning arrester. start follow current easily. Polarity would
presented in this session. It is suggested that a 5-microsecond front, a be chosen as that at which follow current
REFERENCES 45-microsecond half tail and a duration of could be started at the earliest point on the
1. See reference 5 of the paper. 600 microseconds would be more satisfac- voltage wave, or if preferred the timing
2. SPIRAL ARC CHOKES POWER FLOW IN NEW tory to use from a users point of view. could be spaced 15 degrees between opera-
LIGHTNING ARRESTERS, Otto Ackermann, E. J. Another related problem, which is not tions as suggested above.
DeVal. AIEE Transactions, volume 70, 1951, discussed, is the need for a low current, long Mr. Beck has suggested an apparently
pages 995-98. duration test for valve-type lightning arres- logical change in the impulse wave shape
ters. The author's comments on the use of a used on the standard lightning arrester duty
H. A. Cornelius (Public Service Company of 300-ampere crest with a 4- or 5-micro- cycle test. The long tail portion is moderate
Northern Illinois, Chicago, Ill.): The second front and half tail value at 1,000 enough to represent possible severe operat-
author is congratulated for presenting a microseconds would be appreciated. ing conditions even though a transformer
clear, comprehensive paper which discusses winding will by-pass considerable long
several topics of value to the industry. duration current from the lightning arrester.
The use of probability theory in regard to J. W. Kalb (Ohio Brass Company, Barber- The preceding comments are merely for
lightning arrester discharge currents, as has ton, Ohio): Mr. Beck's proposal should go consideration in further increasing the use-
been done in this paper, should be quite far toward better defining lightning arrester fulness of the duty cycle test.
helpful in providing forecasting data for performance with respect to actual service
users of lightning arresters. conditions. The field data cited and statis- REFERENCE
The mention of 5,000-ampere discharge tical analysis thereof offer convincing evi- 1. ESTIMAT1ON OF LIGHTNING ARRE3STER CHRAC-
currents and rates of rise, the frequency dence for the generalized wave shape pro- TE:RISTICS AND ASSOCIATED MEASUREMENT PROS-
1951, VOLUME 70 Beck-The Duty on Lightning Arresters for A-C Systems 1145

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
LEMS, Bror Svensson. Report 337, International in the field, the high-current short-duration The power-follow current that a lightning
Conference on Large High-Voltage Electric Sys- component and the low-current long-dura- arrester can be expected to withstand has
tems (Paris, France), 1946. always been an interesting question. Mr.
tion discharge.
If specifications can be drawp up for the Beck has included data that should be help-
Arthur R. Koerber (General Electric Com- combined long-duration power test which ful in determining the degree of risk that a
pany, Pittsfield, Mass.): Mr. Beck has care- will permit all laboratories to obtain uniform user may anticipate for dynamic over-
fully reviewed field data from many inde- results without these specifications becoming voltages.
pendent sources and has suggested a com- overly complex, then such a combined test It has been the experience of the Bonne-
posite impulse test wave shape to be used in should be reconsidered by the industry. ville Power Administration that lightning
combination with power. A test, similar to arrester failures rarely occur from lightning,
this, that combines a long-duration impulse REFERENCE and that the greatest hazard is the dynamic
wave such as encountered in service with a 1. See reference 1 of preceding discussion. transients that exceed the lightning arrester
duty cycle test would be highly desirable. 2. LONG DURATION SURGE TESTING OF LIGETNING rating. We find that lightning arresters are
Any test to be used as a basis for standard- ARRESTERS, S. B. Howard, T. J. Carpenter, H.
ization should be easily duplicated in dif- Schwartz. AIEE Transactions, volume 70, 1951, namic overvoltages of short duration but of
ferent laboratories and should not require pages 1487-92. high enough magnitude to flashover the
complex specifications to insure identical lightning arrester series gaps, or (2) dy-
test conditions. namic overvoltages of relatively long dura-
Mr. Beck specifies that the duration of the R. H. Mertz (The Detroit Edison Company, tion which do not flash over the series gaps
impulse discharge be determined with the Detroit, Mich.): My discussion resolves it- ning arrester blocks. Voltages in the first
primary of the power transformer short self in the form of a question which the category may occur from unusual switching
circuited to simulate a source of low im-
pedance. In making the actual test, how-
operator is confronted with after the light-
ning arresters have seen some service.
catego r from unusua-connechin
surges or o auls on deltaoected
ever, the reactance of the power source is, of Let us assume that the lightning arresters systems Voltages of the latter type may
course, inserted in the circuit. This addi- have been built to meet the conditions sug- octiony
tional reactance increases the duration of the r.have had,w let us
gested by Mr. Beck. They rejection.
~~~~~Goodeconomy dictates that some degree
impulse discharge. Since the reactance of say, ten years of service, they are then re- of risk must be assumed in balancing cost
the power source will not be the same for moved for some reason such as a rebuilding versus the degree of protection. We would
different laboratories, the duration of the operation. What tests are necessary to de- appreciate Mr. Beck's comments relative to
impulse used in the actual test will vary from termine if they are satisfactory for reuse? these hazards and the probability of light-
laboratory to laboratory, making the sever-
ity of the test nonuniform.
For present types of 6-kv distribution
lightning arresters that have been removed
th ester ailure rom themt
ning arrester failure from them.
The principal difference between the pro- from service, we are thinking in terms of an
posed test and the present standard duty impulse of around 100 kv, 2,000 amperes and E. Beck: The discussions by Sumner and
cycle test is the long-duration tail of the im- a power follow current of 400 amperes. No Ackermann are useful contributions to the
pulse wave. The amplitude of the tail of wave-shape control. Is this sufficient cur- information needed to develop satisfactory
the impulse current wave should be more rent to determine if the lightning arrester is tests for expulsion-type distribution light-
exactly defined than to simply state that satisfactory for reuse? ning arresters. We have no comments to
the total duration of the surge current make except to suggest that more detailed
actually discharged through the lightning discussion of a proposal for practical test
arrester be not less than 600 microseconds H. L. Rorden (Bonneville Power Adminis- criteria for these lightning arresters would be
with the primary of the test transformer tration, Portland, Oreg.): In recent trends of value to the standardizing committees.
short circuited. for reduction of insulation levels for the The discussion by Cornelius raises several
Past attempts to combine long-duration transmission of power at high voltages, questions. One pertains to the probabilities
impulse and power have led to considerable lightning arresters have been recognized as of rates of rise of lightning arrester dis-
difficulty due to the interaction between the the protective device upon which the insula- charge current of 5,000 amperes or more per
impulse and power circuits. Svennson re- tion levels are based. Major economies can microsecond on unshielded and shielded
ported in 1946 that this interaction directly be achieved by establishing as close a margin systems. It is well that this question is
affects the severity of the duty on the arres- as is feasible between basic insulation levels raised because the wording of the paper is
ter.I Our tests have confirmed his results. and the levels to which lightning surges may not exactly clear on this point. The paper
Furthermore, this interaction is dependent be limited by protective devices. This be- states that on shielded lines the probabilities
upon the test circuit constants which can be comes of increasing importance with the of such steep currents are less than on un-
varied without departing from Mr. Beck's higher voltages. shielded systems. This was intended to
specifications. If lightning arresters are to become a major apply to very well shielded systems where
Although Mr. Beck may have used this part of our scheme of protection, the system the probability of back flashes from the
test for some time, it would appear that more designer must know the limitations as well shield wires to the conductors is negligible.
study is necessary to arrive at a set of speci- as the protective characteristics of lightning In another paragraph of the paper the state-
fications which will insure that different arresters that are to be used on his system. ment is made that on high-voltage shielded
laboratories, using iinpulse generators and Mr. Beck's paper contains a vast amount of systems the probability of lightning arrester
power circuits with different constants, can information useful to the system designer discharge currents of 5,000 or more amperes
obtain equivalent results. which has not been previously available to per microsecond may be from zero to 5 per
To demonstrate the adequacy of lightning the industry, and which can only come from 100 arresters per year. These figures might
arresters to handle long-duration discharges, one who has devoted a major effort to indicate off hand that on shielded systems
we have proposed,2 in addition to the present understanding the performance characteris- there can be a greater probability of steep
standard test, a long-duration surge-with- tics of lightning arresters. We are particu- currents than on unshielded systems, which,
stand test for station-type lightning arrest- larly impressed with his suggestions and as Cornelius points out, is not consistent.
ers. This test consists of 20 successive recommendations for duty cycle tests and The shielded systems with a higher proba-
applications of a surge having an essentially his discussion of power-follow currents. bility of steep lightning arrester currents are
constant magnitude of 150 amperes (less Data of this nature are necessary in the not well shielded. The analysis made by
than 10 per cent decrement), to be main- application of lightning arresters if the sys- C. F. Wagner, reference 12 of the paper,
tamned for at least 2,000 microseconds. This tern designer is to provide adequate protec- from which the data were taken, brings this
test wave is believed to be a fair representa- tion without excessive and too conservative out. Wagner calculated the probabilities of
tion of the long-time low-current discharge margins, which in turn dictate high and ex- steep arrester currents on the basis of the
resulting from lightniiig or switching surges. pensive insulation levels, probability of back flashes. If a back flash
The present standard discharge current We believwe that Mr. Beck's suggestions, occurs, the voltage and current on the line
withstand test together with the proposed particularly those recommending a long- conductors rise abruptly, more or less re-
long-duration test shoulld demonstrate the wave duty-cycle test and elimination of the gardless of the rate of rise of current that the
adequacy of the discharge capacity of the high-current short-time tests, should be con- lightning has produced in the shield wires
lightning arrester at both extremes expected sidered for inclusion in national standards. before the back flash occurs. Thus there

1146) Beck-The Duty onz Lightning Arresters for A-C Systems AJEE TRANSACTIONS

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
600 insulation level of 115-kv class insulation is
shown for reference. Comparing curves B
and C with A and considering the areas
50C under them it appears likely that there is
actually less margin between C and A than
between B and A. From this standpoint
400 7
400_alone
should
the 10 by 20 wave of discharge current
r e ~be~ preferable because it produces dis-
charge voltages which in magnitude and
wave shape seem to approximate the sever-
>
300 ity of the 1.5 by 40 voltage wave that deter-
mines the insulation level of equipment.
Cornelius suggests that instead of the dis-
200__ ______ ______ charge current wave shape proposed in the
paper for the duty cycle test, a wave shape
of 5 by 45 microseconds with a total dura-
tion of 600 microseconds would be more
loo satisfactory from the users' point of view.
We have no particular comment on this al-
though we would question why this wave
I_____ _____ ______ ______ should be more satisfactory. We would like
4 8 20 40 to point out that in arriving at these tests,
,i SEC'S there are some limitations as to what can be
done in laboratories and routine test floors.
Figure 1. 121-kv station type autovalve lightning arrester. Discharge voltages for discharge The wave shape proposed in the paper has
currents of two different wave shapes, compared to basic impulse insulation level of 11 5-kv class been used for a number of years and it seems
to satisfy the criterion that is really the most
Curve A. 1.5X40 microsecond voltage, basic impulse insulation level of 115-kv class important; that is, that lightning arresters
B. Lightning arrester discharge voltage, discharge current of 4,800 amperes, 3X23 which meet these tests give good service in
microseconds the field.
C. Lightning arrester dischdrge voltage, discharge current of 4,800 amperes, 10X28 We are not entirely convinced that tests
microseconds in addition to those proposed in the paper
are necessary to control the performance of
lightning arresters and to differentiate be-
tween the different types. We would how-
may be a greater probability of steep arrester with a rate of rise of 5,000 amperes per ever, go so far as to say that a low current
discharge currents in a poorly shielded sys- microsecond, with a crest of 5,000 amperes. long duration test of perhaps the general
tem than in an unshielded one. As indi- Before replying to this question we wish to order proposed by Cornelius at the end of his
cated by Wagner, the higher probabilities of make the following remarks. The meaning discussion has more significance than the
steep currents are associated with systems of rate of rise of discharge current should be 65,000- and 100,000-ampere test now speci-
that have overhead ground wires but lower defined accurately. A discharge current as fied for valve lightning arresters in stand-
insulation and higher ground resistances, on described is difficult to obtain in the labora- ards. We are of the opinion that the 65,000-
which back flashes are more probable. That tory. From certain available data we do ampere tests should be retained for distribu-
low tower footing resistances in the vicinity not think that the current crest has much, tion-type expulsion lightning arresters but
of a station are highly desirable for this if any, effect on the crest of the discharge we think these high current tests might well
reason, is pointed out by Wagner and in the voltage provided the initial rate of rise of be discarded in the case of the valve light-
paper. current is held constant. We urge caution in ning arresters.
Cornelius questions the validity of the using crest magnitude of discharge voltage Kalb raises several points of interest in
conclusions that the shortest fronts of without consideration of the wave shape in reference to the tests. The firing angle sug-
lightning arrester discharge currents ac- relating lightning arrester performance to gested in the paper was proposed because its
company low magnitudes of current. The insulation withstand strength because in- use in such tests over a number of years2 has
data available indicate this to be so. It sulation withstand strength, basic impulse resulted in good service experience indicating
is recognized that the volume of data is not insulation level, is itself defined on the basis the test procedure to be a good criterion.
great. More data would be useful and we of both crest and wave shape. We estimate Also, it is a convenient firing point. In
hope that more may be accumulated. The that an initial rate of rise of 5,000 amperes selecting a test procedure it is not practical
operators of transmission and distribution per microsecond will produce, in a 37-kv to cover all possible conditions covered by
systems could contribute to such an Autovalve station-type lightning arrester, a their statistical distribution. One selects a
accumulation of data. The conclusion is discharge voltage crest of the order of 147 criterion that is representative, is severe
based on all the data that are available to kv. This is about the same as the voltage enough to assure satisfactory service and
the industry at present. There are certain crest produced by a 10 by 20 discharge cur- can be obtained by practical means in the
considerations that also point toward the rent of 20,000 amperes crest, although the laboratory or shop. We do not believe vary-
same conclusion, such as the one mentioned discharge voltage wave shapes will be dif- ing the firing angle is niecessary in case of
in the paper that when an arrester gap ferent. valve lightning arresters.
sparks, a sudden rise in current in the It is not recommended that the crest Kalb's suggestion of a reactive circuit has
lightning arrester is to be expected, up to a voltage just estimated be used as a basis for merit for lightning arresters whose follow
magnitude depending on the characteristics comparison with insulation. For this pur- current depends significantly on the im-
of the lightning arrester and the voltage pose the discharge voltages obtained with pedances of the power system. This is the
across the lightning arrester at the instant the 10 by 20 current wave now standard are case with expulsion lightning airesters. In
sparkover occurs. This is something like more suitable. The change in both crest the case of practical systems and valve
the case of a back flash from a struck shield and wave shape of the voltage produced by lightning arresters the power follow circuit
wire to a line conductor. In a paper1 by I. change in the time to crest of the discharge is highly resistive because of the characteris-
W. Gross and G. D. Lippert results are current is shownl by A. M. Opsahl in ref- tic of the valve element. The effect that
given of measurements On1 rates of rise of erence 22 of the paper. In Figure 1 of this phase displacement between voltage and
surge voltage. Figure 4 of that paper is of a discussion, Opsahl's data are used to plot current may have on the ability of gaps to
special interest and indicates again the the discharge voltages of a 121-ky Autovalve recover and interrupt is well known but we
considerably greater prevalence of moderate station-type lightning arrester with two dis;- do not believe it is a problem in arresters of
rates of rise over steep onles. charge currents, both of 4,800 amperes crest the valve types as they are made today.
Cornelius asks for the magnlitude of the but with different wave shapes as indicatedl In reply to Koerber's discussion, we agree
discharge voltage of a 37-ky station-type in the figure's caption. The 1.5 by 40) wave that there are interactions between impulse
lightning arrester while discharging a current of vo)ltage that designates the basic impulse and power circuits thlat affect the duty on

1951, VOLUME 70 Beck-Thze Duty on L,ightning Arresters for A-C Systems 1147

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.
the lightning arrester. This was considered they are as good as new would require more its rating and during this time the arrester is
when the duty cycle test proposed in the exhaustive ones made under more severe sparked over by a surge, then follow current
paper was established as a routine quality conditions. After having subjected a 10- flows and the arrester attempts to clear the
control test for station-type Autovalve year lightning arrester to the latter tests follow current against a dynamic voltage
lightning arresters some years ago.3 This is there would probably arise serious question that is in excess of the lightning arrester
a factor to a degree also in testing with 10 by whether to put it back in service because rating. The limit of the dynamic voltage
20 waves of discharge current. The tail of these severe tests subject them to more against which the lightning arrester can
that discharge current is less well defined severe duty than they would experience accomplish this is designated by its voltage
than the wave proposed in the paper. We during a number of years of average service. rating. If the dynamic voltage is above the
recognize that probably debate and dis- We are much interested in Rorden's con- lightning arrester rating, the arrester must
cussion is to be expected before a new clusion that lightning arrester failures try to interrupt a higher than normal follow
standard is evolved. The industry has had rarely occur from lightning and that the current against a higher than normal volt-
similar problems that have been reconciled. greatest hazard is dynamic transients that age. It will probably be unable to ac-
A long duration test of the general charac- exceed the lightning arrester rating. This is complish this, power follow current will con-
ter mentioned by Koerber has merit, in our also our conclusion. In the case of station- tinue to flow and the lightning arrester will
opinion, as a surge current withstand test type lightning arresters we are quite con- be damaged.
and we are in general agreement with a pro- vinced that lightning alone without con- We believe the situation just described is
posal of this nature. As in the case of the tributing factors is a negligible cause of the chief hazard to lightning arresters of the
other proposal, some further discussion of it damage. We do not feel that this is equally station type. If applications are carefully
is desirable. We do not believe it takes the true in the case of line or distribution types. made, the risk is negligible.
place of a duty cycle test such as that pro- We concur with Rorden that dynamic over- Rorden's request for comments relative to
posed in the paper. voltages high enough to spark the lightning the hazards and probabilities of lightning
The question raised by Mertz opens a arrester series gap produce damage. This arrester damage is a difficult one to answer
subject on which much discussion might would require considerable voltage in the quantitatively. The risk of damage in-
take place. If there is no limitation as to case of station-type lightning arresters since creases as the dynamic voltage applied
test facilities and the expense warranted, a their 60-cycle sparkover, dry or wet, is at across the lightning arrester increases be-
series of tests might be desirable. One test least 1 1/2 times and usually more than 11/2 yond the lightning arrester rating. With the
we would consider to be significant is a 60- times the lightning arrester rating. In the voltage no higher than the arrester rating,
cycle sparkover or withstand test in ac- case of high-voltage lightning arresters, other the probability of damage is negligible, and
cordance with AIEE standards. To check than the station type, contamination and this seems to have been borne out very well
the protective characteristics of valve moisture on the outside surfaces can cause by service experience. If the dynamic
lightning arresters, we believe that a volt trouble of this nature. Such a condition will voltage applied across the arrester terminals
time oscillogram using a 10 by 20 discharge reduce the 60-cycle sparkover below normal. increases above the arrester rating the risk
current would be desirable. For such a test a This is the major reason why the use of line- mounts. As the dynamic voltage applied to
current crest of 1,500 amperes would be type lightning arresters with ratings higher the lightning arrester terminal increases, the
adequate although other current magnitudes than 73 kv is not recommended. risk increases and finally if the dynamic
might be used. In the case of the expulsion Mr. Rorden mentions, also as a cause of voltage becomes equal to gap sparkover,
lightning arresters, inspection of the series lightning arrester damage, dynamic over- damage is practically certain.
gap and the interrupter should be sufficient. voltages of long duration which do not In closing, I wish to express appreciation
To check the follow current interrupting sparkover the series gaps but result in ex- to the discussors of the paper because the
ability of a valve lightning arrester that has cessive heating of the lightning arrester discussions have raised points of interest and
been in service for 10 years, a test such as blocks. This question must be examined in provided an opportunity for expanding on
Mertz describes, that is 2,000 amperes surge some detail. It seems to us that a dynamic the subject beyond the original scope of the
current and a 60-cycle circuit of 400 amperes voltage in excess of the lightning arrester paper.
should, in our opinion, be satisfactory. We rating but below the sparkover voltage of
assume the 400 amperes mentioned by the lightning arrester, not accompanied by REFERENCES
Mertz applies to the short circuit current
capacity of the 60-cycle power circuit. On
any other transient system disturbance,
would not be harmful because if the gaps do
1. LIGHTNING INVESTIGATION ON 132-Kv SYSTEM
OF TH1 AMERICAN GAS AND ELECTRIC COMPANY,
the expulsion lightning arresters this would not spark, the current through the arrester is I. W. Gross, G. D. Lippert. AIEE Transactions,
be a rather mild circuit and on these we rec- limited to that which is passed by the shunt- volume 64, 1945, pages 74-84.
ommend inspection rather than the duty ing resistance elements that parallel the 2. CONTROL OF LIGHTNING ARRESTER QUALITY,
cycle test. Such tests will not tell whether series gap elements. This current is small T. L. Dyer, E. Beck. Westinghouse Engineer (East
these 10-year lightning arresters are as good and to the best of our knowledge would not Pittsburgh, Pa.), May 1951, pages 100-01.
as new in all respects but they will indicate damage the lightning arrester valve ele- 3. SURGING LIGHTNING ARRESTERS CONNECTED TO
that more service can be expected from ments. If, however, there is a dynamic POWER VOLTAGEt
Ab DrFohrbeP, 935, pages 162
them. To make tests to determine whether voltage on the lightning arrester in excess of 65.

1148 Beck-The Duty on Lightning Arresters for A-C Systems AJEE TRANSACTIONS

Authorized licensd use limted to: IE Xplore. Downlade on May 13,20 at 1:4239 UTC from IE Xplore. Restricon aply.

Anda mungkin juga menyukai