Documente Academic
Documente Profesional
Documente Cultură
23-07 o
EXPERIENCE OF AC VOlTAGE TESTS WITH VARIABLE FREQUENCY
USING A L1GHTWEIGHT ON-SITE SERIES RESONANCE DEVICE
by
ISwitzerland)
Abstract a suita ble m eans of detecting meta l particies in particul ar after assembly. On
th e oth er han d, a dir ect volta ge imp oses diff erent st resses o n th e insulat ion
The construction of increasingly large GIS installations has resulted in unit capacities to the altern at ing volt age to w hic h it is exposed in servic e.
growing up and necessitated h. v. testing with transformers of ever increasing rates.
The transport and the energy supply of such heavy transfomners presents probrems.
today. Th erefore, t he t est dir ect vo ltage. as com m onl y used fo r cabl es,
although even th ere it is not without pro blerns, shoul d not be transferred to
Resonance test sets with fixed or variable inductances of the reacto rs are a good GIS insta llat ions because th e SF6 gas behaves qui t e differently to the oil/
alternative. In the first case, however, the test frequency is usually above the normal paper insulation system s of cab les.
power frequency. that means 50 to about 300 Hz,dependingon the testing capacity.
In ce ntrast. th e problem w ith te sts emp loyi ng alterna tin g voltage is t he
This paperprovides a summary of thedevelopment of resonance testsetsand shows
their salient features. Furth er. it examines the question as to wheather and to what large amount of reactive power th at is required, especially in view of th e rela-
extent the frequency influences the breakdown values of a pure gas gap and one ti vely high capacit ances of GIS install ation s. Apart from thi s high reactive
equipped with an insulator. pow er. t he pr ob lem of pow er: we ight rat io mu st also be me nt ioned. For an
on-site t est at onl y 40 0 kV. 1 A. em ploy ing conv entiona l test rneth ods, it is
Rel erring to some practical assignments on site, the functional capability 01 the set is necessary to tr ansport a tota l w eigh t of up to 200 kN to the site .
demonstrated, from which it would appear that particulary this method of testing is
ideal for detecting faults.
The displacement of suc h weights rapi dly leads to serio us tran spo rt
Key words : problems. A s alternative. proposals have been ma de f or an on-s ite te st with
osci llati ng sw itching im pulse vol tag e [31, a m ethod tha t has m eanwhile been
GISinstallation; on-site testing; serie resonancetest set; variable frequency; alternat- introduce d. Oscill aling sw it ch ing im pulse ge nerators are relat ively light in
ing voltage test. w eight and can thu s be tr ansp ort ed easily. Experience has shown, th ough,
th at t his m et hod of testing ls also able to det ect certain f aults in th e installa-
t ion reliablv, alt hough it has not been det ermined w ith certa int y. w hether
fo reig n partici es can also be reliabl y detect ed w it h this volt age f orm . Such a
I. Introd uction test is t herefo re part icul arly app ropriate, because every sw itch ing op erati on
Single component s of conv enti onal indoor or out door swi tchgear in- in t he network imposes sim ilar st resses on t he sw itc hgearto those pro duced
stallatio ns are normally subjected to th e stipulated typ e t ests and routine by th e osci llat in g sw itching im pulse. .
tests in the facto ry. When t he var ious com po nents have been assembled on
th e cust om er's site there is no real need fo r an on -site test, since the proo f of On th e other hand, fo r syst em s w it h lower rat ed voltages, t esti ng wi t h a
t he specified ins ulat ion level can be guara nte ed fo r th e com plete insta llatio n power-f requency voltage of abo ut 80% of the nom inal test valu e, perfo rm ed
on th e basis of t he factory tes ts [1]. on site, is st ill th e com mones t and best accepte d m et hod of test ing , and in
th is voltag e range is th e nearest equiv alent to the stresses encountered in
On the ot her hand, m etal- enciosed gas-in sulated sw itc hgear (GIS) has service.
to und ergo a voltage t est wh en it has bee n assem bled on sit e, additionally.
For t he very compact SF6 installat ions such a test serves a usef ul pur pose for In order to dispens e wi th th e heavy. convention al t esting tr ansform er,
a number of reasons. Firstl y, t he liv e part s cannot be insp ect ed visually when resonant cir cuit s can be em ploye d as a m eans of generat ing high alte rnat ing
th ey have been assernble d, on account of t hei r enciosure. Secondlv, owing test vo ltages. Likewise, th ey are th e sole on-s ite method of te sting th at is
to an ov ersight du rin g assembly on sit e it is poss ib le for fore ign part ici es to capable of dete cting fa ults reliably.
enter the equipme nt and possib ly resul t in the di elect ric stre ngt h being
red uced [2]. and t hirdly, damage suffered in t ransit cannot be altog ethe r
ruled out. Sy performing a vo ltage t est on sit e at the conci usion of assernb lv,
the custo me r is assur ed tha t t he installation w ill operate pro perly and t hat 11. Resonant Circuits for Generating High. Alternating
t he desi red or specifie d insul at ion level is adhered to. Test Voltages
It is w eil known that - apart from a f ew except ions - any elect rical in-
Present-day t esting pr acli ce for th e assurance of qualit y and pro ving sulat ion syst em can be regarded as a capaci tor w it h a high qualitv tac to r, i.e,
th e f uncti onal reliab ilit y is admittedly expensive, but it is selective. In addi- a low act ive pow er in relati on to the react ive power absorbed. Ther efore,
t ion to the norm al ma teria l tests beginnin g wi t h inspection of incom ing wh en t est ing equ ipm ent at high voltage with th e system f requency or
mat erials receiv ed by t he factorv, tests are p erform ed on t he vario us com po- higher, t he voltage generat or is almo st ent irel y load ed with react ive po w er.
nents wh ich are later assem bled to form a high-v oltage unit. A num ber of From th is aspect te st tran sformers are un econ om ical and tec hnically ext ra-
t ested components are put tog ether to form a shipping unit and subjected to vagant, bec ause the react ive powe r giv es rise to unnecessary losses in t he
the necessary electro-mechanical and dielectri c tests laid down in lEe 517.A various wi ndin gs. and the vol tag e adjusting device, w hich is abso lute ly
combined pow er- fr equency voltage t est togeth er w it h measurem ent of essent ial. has to be dim ensioned f or th e f ull react ive power. Integra l test in-
part ial discharg e in a single test has been fo und to be very st rict. W it h t his stall at ions, even wh en th e tes t transfor mer is of very co m pact desig n [101.
t est m ethod t he part ial discharg e ons et and offset voltages are checked, possess larg e dime nsions and are ve ry heavy. Th ey are co nsequent ly far
w hich should be appreciably hi gher t han 1.2 times t he voltage betw een from id eal for on -site testing of SF6 installat io ns. lt is t her efore und erst and- :
conduc tor and earth. On site th e shipp ing units are th en put tog ether to fo rm able t hat thi s m eth od of generat ing th e te st voltage for high-grade insular-
t he com plete installa t io n and fin ally subjecte d to an on-s it e tes t. as pro of of Ing system s is often called the " brute -fo rce" meth od .
co rrect assembl y.
The high react ive pow ers requi red fo r testi ng h.v. cables or capaci tors
If we now consider the various kinds of volta ge w hich could possibly be provid ed the initi al im petu s at an early stage for the develop ment of co m -
used for such t ests, it mus t be sta ted t hat a test w ith direct vo lta ge is indeed pensat ing circu its capa ble of relieving t he load, at least o n parts of the test
23 -07 - 2-
vol tage sourees. The sta ges of th is developm ent are illustrat ed in Fig. 1.
t hough only inf initely adjustabl e methods of cornpensat lon , possessing a
reso nanee eharaeteristi e are taken into aeeoun t.
I
ever, the power eons ump t ion of the entire installat ion is appreeiably
red ueed.
TT
Ct
RT VR == Fig. 2:
I Schematic diagram ot a HV-Series resonant
Vs ? l:fI] Vt
1
eircuit wit h variable t requency input voltage
>«
on ee m ore increase th e to tal we igh t of th e equipm ent.
V
s
~ IJ_ I A fu rt her reduet ion in th e size and weight of high-voltage te st equip -
me nt can be aehieved by op erat ing the series resonanee eireuit as per Fig. le
w ith a volta ge V s at var iable fr equ enc y. Then t here is no long er any need to
vary t he induetanee of t he reaetor HVR in orde r to obtain th e resonanee
cond it ion s needed for gen erat ion of high output voltages. Moreover, th e
f const b) reacto r ean be built as a very srna ll, eompact unit with low w eight. A eireuit
emp loying thi s pr ineip le is illu str at ed in Fig. 2 [11]. The volta ge sou ree w it h
eont inuous ly variab le f requen ey is a statie fre quency ch anger FC. wh ieh
feeds the resonant eireuit (L. C,) t hrou gh the exeit er t ransforme r Er. In [11]
IN bot h the dim ensions and the service eharaeterist ie were der ived and it was
show n tha t th e hi gh-v oltage react or L ean eonven ient ly by eom pose d of a
number of sm all elem ents if the f requ ency of t he high test vo ltage V, is to be
maintained w it hin elose limits. The size of t he reaetor does not depen d on
th e seleete d fr equen ey. This me ans tha t it is not the reaeti ve power w C, . V, 2.
inereasing wi t h the fr equeney. tha t governs its size, but rath er th e ma ximum
energ y C, . V , 2 stored in eapaeitor C" A direet result of this is th e major
advantage that induet ive voltage tr ansform ers, whieh would reaeh satura-
t ion at service frequeney and hi gh test vol tages. are able to remain in th e GIS
installat ion. Therefore th e series resonanee eireuit briefly deserib ed below
was designed for a nominal f requeney of about 100 Hz.
f const
111. A Mobile 800 kV Series Resonance Circuit of High
C) Output
Fig. l :
Diff erent methods tor capacitive a. Low voltag e parallel campensation The high-voltage reaetors of th is eireuit were dimensioned in such a
load campensations tor constant b. Low volt age series campensation w ay that at a resonan ee frequ ency of 100 Hz and a voltage of 800 kV. it is
testing tr equency c. High voltage series resonant circuit possible to test a load cap ac itanc e of about 12 nF under short-time eond i-
t ion s for about 10 minutes. From th e eond itions for resonan ee
Vs AC input voltage TI testin g transtormer
RT induction voltage reguiato r
VR variable reactor, low voltage
ET
C,
exciti ng tr anstormer
object under HV test f,=f,=~ (1)
HVR variable reactor, high voltage V, test voltage
It load eurrent th e induetanee L works out t o about 200 H. Thus th e rated eurr ent is 6 A. The
tot al appare nt power of t he reacto r, 4 800 kVar w as divided between four
ide ntiea l reaetor units, eaeh for 200 kV w it h an in duetanee of appr. 50 H. in
order to enh anee t he fl exib ility of th e equipme nt in appl ieation. thus result -
Oft en th e arrangem ent in Fig. la results in the hig h test voltage V, ing in t he units being sma ll and easily trans po rt able. The w eight of eaeh
co ntaining unw ant ed, st rong harmonies beeause the leakage reaetanees of reaetor uni t is onl y 3.75 kN. represe nti ng a ratio of w eight to po w er of o nly
th e test tra nsform er and regu lato r RT form series resonanee eireuits about 3 N per kVA. Th e we igh t of th is reaetor mak es it t he heaviest sin gle
together with th e t est obje et C,. Th ese eireuits ma y be exeit ed by harm onics item in the test installation (th e exeite r tra nsform er rat ed 50 kVA w eigh s 2
in t he syste m voltage V, or eurrent harmonies. Thi s drawback is avoid ed in kN. the f requeney chang er rated 50 kW we ighs 3 kN).
the arrangeme nt shown in Fig. t b, in w hich t he variable eom p ensat ing
reactor V R is in series with t he prima ry w inding of t he testtransfor m er TI [4] . The eompl ete installa ti on is illu strated in Fig. 3. in wh ieh all f our reaetors
But the other advantages and disadvantag es of th e preeeding arrange me nt are eonneeted in series to prov ide a test assemb ly for 800 kV. The toroidal
are retain ed. This method is qu ite often used, som etime s in mo dif ied form eleet rodes of eonduet ing plast ie material. w hich had to be very earefull y
[5] . But t he ov erall w eight of th e equ ipm ent is not signifi cantly redu eed. A dim ension ed [12]. are loeat ed between th e reaeto r units ; as separate
marked adva nee w as. howe ver, ma de w hen it became possibl e to eonstruet elements t hey ean also be easily t ransported and mounted on site. Th e fre-
reaetor s of variable indueta nee for high ov erall voltages [6 ]. for w hieh th e qu eney chang er is a speeially deve loped unit ; t he eleet ronieally eont rolled
prineiple of series reson anee wa s prima rily used (see Fig. lc). Thi s arrange- fr equeney variat ion in extrem ely f ine steps at a eonstant output voltage of
ment has rapidly found its w ay into praet iee in reeent years [7. 8] . th e larg e. about 500 V enables th e high vol tage to be fin ely adj uste d and kept stab le,
heavy t est tra nsformer having been replaeed by th e mueh Iighter variable simpl y by uti lizing th e resonanee eurve.
high -vol tage reaetor HVR. In th is series resonanee eireu it the tr ansfor me r is
redueed to a mere exeit ation tr ansfor m er Er of relatively low rat ing and Aeeording to eq. (1)th e aet ual f requency f , of th e alternating t est voltage
voltage because it now has only to cover the active losses in the high -voltage wi t h th e series or parall el connectio n of the reaetor units , as chose n in a
eomponents HVR and C,. If Q is t he qua lity f acto r of this series resonanee part ieular case, is det ermined by t he eapaeitanee of the load C,. Durin g the
circuit, the neeessary seeondary voltage of thi s exeit at ion tran sform er is only on-site te sting of SF6 inst allati ons this valu e is given by the eapaeit anee of
VJG. beeause it on ly has to provide the load eurrent. lt is quite evident th at t he act ual installation, but it is liable to fluetuate. At small values of C, it is in-
th e eapaeitive reaet iv e power of the te st obj ect is eompensated in an ideal deed possible to obtain a desir ed, tight fr equeney range by f ixed additional
rnanner, t hereby relieving optimally all th e elem ent s required for excit ation. eapaeitanees of graduated values. But the fact remains that it is diffieult to
Even if, for reasons of insulation, the various h.v. reaeto rs HVR can hardly be assure a given test frequeney exaet ly. This drawback is offset by the advan-
built for vo ltages higher than about 300-400 kV. it is nevertheless possible tage that by using a eombinati on of reaetors it is quit e easy to keep w ithin a
to obtain higher overa ll test volt ages V, by eonne cting several reaetors in narrow frequeney range and, furtherrnore, it is possibl e to te st quite high
series. eapaeitane es at redueed voltages. Sinee t he range of applicati on of a
-3- 23-0 7
_ .--- - - - -t-- f t
Hz
I--"..e.--~:I_---~~r_---+_- l 00 0
r
100 ----+----~tx:_~""'""...........---+---""'...
45
f
min
10 100
--- --1~
~
CdnF
Fig.3:
Generalview of the complete test set 1 isolating transformer
with 4 reactors in serie and the 2 control unit, frequency converter
necessary compon ent s 3 supply transformer
4 reactor f
5 voltage divider 2p
~ H,.Z-+-_ -I-_ _rt----::.,-L-_ _-+ -t_
reactor is defined, on the one hand, by its nominal voltage , i.e, considerations
of insulati on, and on the other by the rated curr ent (t emperature rise and 1000
l,~2p+4p------i------">.",;;-~~~~
=-
saturation of the iron circuit), but both quantities are independent of one
another, th e relat ionship betwee n the maximum possib le t est voltage V, max
of aseries resonance circuit and the given load capaci tance C, for C,;;:; C, is
also given by
,m
V " ~
v;:- = V -c,- , when V, m a x ::;;; V n- (2)
In th is formula V n and C, are both nominal values, for whi ch the rated
100
curren t In is obtai ned at rat ed voltage V n• From eq. (1) a similar relatio nship
can be derived fo r t he fr equenci es, i.e,
(3)
Fig.4 shows an evaluati on of eq. (1), (2) and (3) for the inst allation ill us-
trat ed in Fig. 3. Th e max imum test voltage V ,m"
and the f requencyf, thereby IV. Experiments
obtained are shown for the fo llowing circu itry combinations: A series of practical experiments was carried out to determi ne the
1 : 1 reactor unit extent to which a .t est freq uency diffe ring from t he industr ial freq uency
2s : 2 reactor units in series aff ect s the breakdown behaviour of GIS components when series resonance
3s : 3 reacto rs in series test circuits are used. The answer to thi s question is of importance, in that it
4s : 4 reacto rs in series has to be assured th at a testvo ltage at elevated f requency of the orderof 100
each in terms of C" Fig. 5 shows th e same informati on for th e comb inat ions : to 300 Hz enables equally c1earand defin ite statements to be made as a test
4p : 4 reacto rs in parallel at the industrial power frequency.
2p : 2 reactors in parallel
Zs/Zp : 2 pairs of reacto rs in series/pa rallel, 1. Dep endence on f requency
the conditions for a single reactor (1) being shown for comparison. For in- For the purposes of the test an experimental tank filled w ith SF6 and contain -
stanc e, if 300 Hz were specified as maximum perm issibl e t est frequ ency, it ing a plate/plate configuration w as used. In th e midd le of the plates there
would be possible to use th e 4s arran gem entlo test capacitances from 1.3 to was a slight identation in wh ich, for lat er tests, a cyl ind rical piece of insula-
12 nF at th e fu ll voltage of 800 kV, dur ing which the frequ ency would drop to tion could be inserted. With a distance of 28 mm belween plat es and an SF6
100 Hz. If the load capacitance w ere increased to 48 nF and the voltage pressure of 350 kPa, this arrangement was tested with the series resonance
reduced to 400 kV, th e fr equency wou ld drop f urt her to 50 Hz. But it is quite equipment. The various freq uency values were obtained by sw itchi ng high-
easy to restr iet th e frequency range to the select ed dimen sioning value of voltage capacitors in and out of circuit. In every case the voltage was raised
100 Hz by employing th e 3s, 2s or single -reactor arrangements at the until breakdown occur red. For the free gas gap (curve 1in Fig. 6) the electr ic
maximum possible t est voltages, as can easily be seen in the diagram . W ith strengt h wa s found to remain constant up to above 300 Hz, follo w ing which
parallel or series/p arallel arrangements (Fig. 5) the field of app lication is, in a sligh t rise in the breakdown voltage wa s observed. For each test point
principl e, extended to very high test capacitances with correspondingl y about ten breakdowns w ere produced, from which the arithmetic mean was
23-07 -4-
calcul ated . The diagram also shows the upp er and lowe r measured 3. Inhomogeneous arrangement
maximum valu es. Also t he relationship wi t h pressure and f requency of a very inhomoge neaus
arrangement (poi nt/ plate, distance 40 mm, angle of point 30 °, radius 2 mm )
was investigated . Fig. 8 shows how th e pr essure of the SFa gas varies as a
1 Gas gap function of the breakdown vol tage. The paramet er is the fr equency, wh ich
can be vari ed betwe en 73 and 536 Hz. Here, too , the (familiar) result was
u 2 Gas gap with tangent ial stressed obtained, the test voltage init ial rising w it h increasing gas pressure. This
maximum th en changes to a minimum and finally begins to rise again. Here
IV insulator the relationship between th e breakdown volt age and the frequency in the
400 lower pressure range also agr ees weil with the known result s [13]. This
r .m. dependence o n frequ ency (high f requencies also result in high breakdow n
voltages) is of no signific ance for GIS installation s in practi ce because the
pr essure normally emplo yed is out side this range. At 350 kPa, for insta nce,
(lJ 30 Fig. 8 shows that the breakdown volt age remains constant fo r all the
öl measured frequ encies.
'"
+->
'0
>
<::
I ~
-1
,: : 2B
2.
1r
:;:
0 \. 4""""""' )
-0
-'" hig her frequenc ies
'"s,
(lJ
100 '\1
..0
..,
rn
-----:~.,. Frequency ~z ......
0
Fig.6 :
>I
h
AC-Sl renglh versus frequency for plale :;:
elecl rodes wil h and wilh oul insulalo r 0
"':J f r equenci es /'
-'"
m
<lJ ,-/ t. 30°
Curve 2 also shows the relat ionship between the fr equency and the break-
2f4a
I-<
50
... / '
down voltag e, but in this case when a supporting insulator w as inserted . This ..0
epoxy insulator was loaded tangentia lly and reduc ed the breakdown volt age
in the we il known manner. For both of thes e basic arrangements the break- ,/ r=
down volt age w as found to remain const ant thraughout the entir e frequ ency
range of interest. The 50 Hz point s were oblained fram measurem ents w ith ,,- /"-'''-1 ~TI
a conv entional test set and w ere obt ained under oth erwi se identical cond i-
tians.
2. Dependence on pressure
1
Fig. 7 show s the f amil iar relationship between t he breakdow n voltage and 0 100 200 30 0 400
pressure for th e arrange ment cont aining the supportin g insulat or. Here, too,
the voltage was conti nuously raised w ith the series resonance test set and 3
to 5 breakdowns wer e produced per point. - - -.....~
. SF6 - ga s press ur e ffi
Fig.8 :
AC-breakdown vollage as a funcl ion Theorelical calculaled minimal
of SFa-gas pressure al frequencies vollage (13)
u 8elw een 73 Hz and 536 Hz
IV
r .m .S 400
U 500
kV
r
T.m.s.
300
-~
(lJ
(lJ öl 400
öl
'"
+->
'"
+->
~
0
0 >
>I I
<:: 300
s, :;:
(lJ 200 0
~
> -0
0 -'"
i1
~
s:
'"s,
;:::1- ~ :~
(lJ
Vl
200
~
'"
l.L.
..0
100
= 100 T
r SF6 gas press ure 350 kPa
5. Discussion of results
Fram th e m easur em ents take n an d iIIust rate d it is qu ite evid ent th at conven-
ti onal tes t tra nsforme rs and series reson ance equip m ent, w he n employed
und er th e same cond it ion s, yi eld the sam e resu lt s. Both p hysical and p rac -
tic al resu lts are equally atta inab le with both me th ods . Thi s particularly
ap plies to th e frequ ency range investigated, rangin g fram th e norma l power
frequency to several hundred Hz. This result is just as grati fying for the user
of series reson anc e tests sets as it is for the custorn ers, wh o can agree to
th eir use on sit e w it hout any hesi tat io n.
V. On site experience
M eanw hil e t he series resonant t est set wa s tested and used on- site by
d iff erent custome rs of GIS-i nstallat ion s. Th e fo llowing tab le shows th e Fig.lO :
list in g of tested sw it ch gears and giv es the actual and charact erist ic data s On site test in Norw ay.
result ing f rom test equipm ent, nu m ber of reactors, circ uit, capac ita nce and
so on:
' additional test oscillating switching surge 7. Th is is confi rm ed by on -site tes ts performed on GIS installations at
rate d voltages up to 4 20 kV, w hich have demonstrated th e pr actical, fl exible
Table: Examples of on site high voltage tests conducted on complete assembled and easy han dlin g of series resona nce t est sets w ith var iable fr equency as a
switchgear by means of the series resonant test set with variable frequency m eans of pe rf orm ing alt ernating vol tage test s.
to C I G R E P A PER 23 -07
equation (1) -
2 TI /LC
t