Breakdown of Transformer Oil

M. G. Danikas, Asea Brown Boveri Research Center

iquid dielectrics are used as mulatin media in

high voltage systems. They can fill any space to Factors affecting the dielectric strength of
L be insulated. They can also be easily circulated
transformer oil include the stabilization
and they can be used to dissipate the heat generated in

a given system. Their maindisad vantage is that they are phenomenon, oil and electrode pretreat
easily contaminated Transformer oil ls one of the in
ment, oil velocity, electrode area and gap
su!ahng liquids most frequently used for electrical in
spacing, and the effects of the several
sulation purposes.

Previous research in the field of breakdown of msulat capacitances present in the test circuit.
iii).; hqulds was concerned with <111 attempt to under

stand th physics of the breakdown mechanism [1,2] as

well as the dependence nf the dlelectnc strength of

transformer oil on 1
, variety of factors [3,4]. experimental wnrk reported here:

\rlany factors, such as clrclrodo area, gap spacing,

conditiomng, electrode and oil pretreatment, influence (1) small mild steel electrodes (cnch of Hl mm

the measured dielectric strength of transformer oil. Thi: dtamererl. The oil entered through a hole .it tbe

dielectric strcngtb o! transformer oil depends on center of the lower electrode mid spreed radially

various factors either due tu ti::.t condition, ur to the oil into the test gap. The gap spacing was adjusted by

itself. Therefore, thedielectnc strength of the oil is some a micrometer attached to the upper electrode,

thing which cannot be defined simply by Its numerical

value. Test conditions have to be described in detail if (2) Bruce profile brass electrodes, f1i,; 1 . wtth a flat
2
the numerical value IS to be meaningful area of 2.56 cm (the gap spacing being adjusted

Since most practical insult1ting liquids depart from with feeler gauges); aud

the ideal hquid state, it is convenient to call weak

links' an departures from the ideal state. The break (3) large rruld steel coastal .:ylrndrK,11,lr!!:lnxles

down mechanism is assumed to be initiated by a covering a wide r.mgL of g.ip spacings and areas.

'weak l i n k ' , i.e. any sort of lrrcgulariry that may exist

either on the electrode surfaces or in the liquid itself.

It has to be noted that with the small mild steel
lncreastng the number of "weak links" increases the
electrodes, gaps 1n the range of 0.1 0.5 nun were

probability for breakdown at a given electnc stress, investigated; with the Bruce electrodes, gaps in the
so a reduction in dielectric strength is expected. An range of I to 4 mm; and with the large mild steel
iru rease in the size of the insulating specimen under
electrodeagapsin the rangeof I to6mm. The large mild
test therefor o implies a r e d uc ti on 111 dielectric steel electrodes covered areas in the range l)f 30 to 1240
2
lrength (sire effer-t).
=L

With uus m mind, the wnrk described 1n this article

has ne, ,t mnm p1Hpose to investigate -ume of the

factors a ffectmg tho dielecmc :.tn,ngth of transformer Cleaning Procedure

oil. These f actors i n c l ude t he s t abiliza t iun The electrode prcparauon consisted of first washing in

phenomenon, ml and electrode pretrt<,thnent, the ef trichlorocthylcnc, then a standard surface polishing

tect ot oil velocity, the effect of t

i capacitance parallel with u 10Cl0 grade <ihcon carbide paper. The smoothed

to the test cdl, and the effects of electrode area and electrodes were then wached m <1 mature of hot water

gap pacmg and soap solution, uini.;- ,1 synthetu spoogi: to assist

with grca:.c and dut removal After :.uLLt'SSJVi: rinses in

hot and distilled 1<1.t1:r, the electrode, W('T(' dncd 111 a

lammar flow air cabinet, ,md fin.illy wiped with il lint

Experimental Apparatus and Procedure
tree lens tissue with acetone and then trichlorocfhyh-nc.

lhe rest of the electrode assembly was washed m hot

Electrodes water and soap solution .:md then nnM..'cl .1 above atter

Three kind s of elec t r ode s were u se d for the each test

Ol>Kl-75.S4/90/(l'l(ll).1)027$1Jl<)::CJJ990 IEH..

IEEE Electncol lrisulot1on Magazine September/October 1990-Val 6 No. 5 27

 E .------A--.,B
=
a
c c
.-
, e E
oo E
/

,''-'-'
0 10 20 30 40

F
Sequence of Breakdowns

Fig. 1. H;aJf contour of a Bruce profile electrode. Section liA

Hg. 2. Breakdown sequence with the brass electrodes and a
is the electrode profile. Section AC is described by a sine
gap of 2 mm, The last 20 measurements were taken with a
fund:ion based on axes QC and OA, where DB = 0C sin
570 pF capacitor conneded in parallel with the test cell.
(AD/A0'!/21.

Oil Handling and Processing

electrodes. The current in the transformer secondary
A quauty grade transform.er oil was used. The oil was was limited by a series electrolytic resistor (500 k.2)

stored in a converted oil drum staIKling 011 a frame four between the high voltage transformer and the test cell.
meters above ground level The oil was kept clean and

dry by a filter and silica gd drier in the line venting the

storage dru.m. Before entering the test cell, the oil went
Experimental Results
through a 6 m filter into an evacuated degassing

column, wh!J...it ensured sufficient air removal as well as

partial drying. The test cell was lower than the degass Stabilization

ing column. so that when sufficient oil had been After successive tests, the average dielectric strength

processed, it could be transferred to the test cell by of the oil reached a stable or plateau level. This

gravity (for more details on this particular oil purifica stabilization is influenced by a vanety of factors such as

tion process, see rs)). electrode area, gap spacing, oil quality, energy released

in the gap, electrode pretreatment and lime between

breakdowns. Some of the aforementioned factors are

Oil Quality Control inveshgated below. Stabilization 1 affected by the

Quality of the transformer oil was regularly controlled electrode area. For the large cylindrical mild steel

throughout the experimental work. The check was per electrodes, values were stable after about twenty break

formed with a standard commercial motor-operated downs; whereas for the brass electrodes and the small

Foster oil test cell having standard VDE electrodes and steel electrodes, values became stable after about ten

a gap spacing of 2 mm. Throughout the experimental and five breakdowns, respectively. It 1s suggested that

work, the oil used had a drelectrtc strength of ap stability may be attributed to the removal of surface

proximately 14 kV /mm. Control of the oil quality is irregularities by successive breakdowns [1] However,

always necess;ary, inct" both the reliability and it is also possible that smce the electrodes were _polished,

reproducibility of results, obtained with a variety of cleaned and dried in open air, they may have adsorbed

electrodes and gap spacings, depend on it. It would be gas Stabilization may also be attributed to the gradual

a pointless exercise to carry out experiments if the oil removal of gas from the electrode surfaces, due to sue

quality kept changing. cesstve breakdowns [61. Oil quality in the gap could be

changed either by continuous circulation ur by chang

ing the filter porosity. In most cases, continuous circula

Electrical Tests hon of the oil m the gap dunng the test (at a velocity of

All tests were conducted with sine wave ac voltage at 3 cm/ s) increased the percentage difference between

SO Hz. A 100 kV, 5 kVA transformer was used for the the first breakdown value and the plateau level. without

h.igh voltage supply. The applied voltage was raised m altenng the number of breakdowns before reaching the

all tests uniformly al a rate of 2 kV /sec. Acircmt breaker plateau level. This also suggests that stabilizmg can be

connected to the transformer primary removed the fault associated mainly with the removal of surface ir

current, whose prolonged flow might have caused car regularitles. A change of oil Jilter porosity (from Om

bomzahon of the test sample and p1ttmg of the to lSm) altered only slightly the percentage difference

28 IEEE Electrical JnsUOtlon Magazine September/October 1990-Vol. 6. No. 5

 9

E 7
Sequence ct Breakdowns
E

Fig. 3. llrekdow11 sequence with the brass tlectrodes and a

gap of 4 mm. The last lO mta!<.urements were \aken with a

-
>
.:,:.
6

570 pf cepeciter connected in parallel with the test cell.

-
s:

between Lhc first values and the plateau level.

-Ol
5

c
Tt-, carried nut on the Bruce electrodes, polished with

320 grade emery paper, 1UOO grade emery paper, and

very fine emery p<'lper. showed that st11biliLath)n

-"'
(f)
4

depends on the roughness of tilt' olortrod .. '>llrfh'<!'>.

o
1 he mugher the electrode surfaces, the longer the c::
tabili7.ing pemxi.

treated with a 320

1t was observed that the electrodes

grade emery paper reached the

-o
3

plateau level after 15 to 20 breakdowns; the electrodes

"'
treated with IOOO gradl emery p.ifl<-'T rear-lied ttic "'
zs 2

plateau lt!wl after about IObnMkdowns; aud lhose that

were highly polished with very fine emery pllpt.!T

1
presented practicnlly mi stabilization at 1111.

Tlwi,1 nw.;ulN lim to confirm conclusions previously

drawn m [I], where it was supported that successive

0
visible discharges seem to remove the most prorrunent
1 2 4 6 8 10
of the protrusions existing on the electrode surfaces.

The conclusion rcac hed elsewh ..r.. [7], that stabilization

Gap ( m m )
is 1)1,;111g to the removal by successive discharges of

small fibers suspended in the oil within the test gap, is

I --

in disagreement with cur results, since we replaced the Fig. 4. Gap "'pacin8 effect for the large steel electres. The

oil sample after each breakdown measurement. lt may plotted pointt1 are for etectrode areas of 61.92 t'm (po,nts

be soul that stablhzation ic:. due mainly to the removal plotted .1s tri.uigles and square,, for flowing and static on,
2
of protrusions from the electrode surfaces with succes respertivelyl; 1,nd 1:12.Sb cm Iopen circles for flowing oll,

solid circles for staric nil).

sive breakdowns. Some of the stabilization may also be

due to a gradual removal of the gas absorbed by the

electrode surfaces during their pretreatment [8].

Oil Quality
It should be noted here that the creation of an insulat
Impurities in oil can be classified into four hmxl gmups:
1ng layer on the electrodes because of chemical
1) particles of dust or fiber already present in the
transformations is mdcod po..,1hk [241. Therefore, a
liquid. These particles lower the dielectric strength
higher applied rnlla,e may to necessary to overcome
of the oil and they may e!'.Cape the filtration pro..'<'
SS

this barner. This has been observed [251, where it was

if they are extremely small,
shown that with one of the electrodes coated, the cur-

rent was reduced by two nrdor-, nf magnitude and the 2) particles produced. by previous discharges

current Zvoltage charactensucs showed different These can be carbon particles resulting from the

trends. depending on whether the cathode or the anode deccmpcsirion of the oil or minute m,t;1llk

was the coated electrode. Moreover, it is not expected particles removed from the electrode surfaces by

that Hewing oil would dihxlgc thi layer Lhe discharges;

IEEE aecncci Insulation Magazine September/October 1990-Vol. 6, No. 5 29

 - BO
BO

- 70 -
-E
70

E
E E

-
>
60 -
-
>
60

.x: -"'

- 50
- -
s:
50
s:
-Ol -Ol
"
c
"

c

Q) 40 - a, 40

-
Cl) -
Cl)

u 30
- o 30
"o'-
.::::
'o'b..
-
.::::

-u
Q)
o

a,

Q)
20
-
'-'- - -
Q) 20

0 -o __ 0
-o - -
10 - 10

o-----
o.i 0.2 0.3 0.4 0.5

Gap (mm) Gap ( m m )

fig. ,. Gap spacing dfect for the brus tltrodH, with Fig. 6. Gap spicing effKt for the small steel electrodes, wilh

Oowing oil (solld circles), and static oil (on cirdn), flowing oU (solid cU.les), .a.nd Matt.: oil Icpen drcluJ.

3) water; and Changing the oil filter porosity (fmm hm t

o 1.'im)

marked a decrease of the mean dielectric strength of

about 12% (tests were carried out with the Bruc:e

4) additives deliberately mtroduced into the oil in electrode, and for )PP spadngs of 1, 2, and 4 mm).

order to change its electrical properties. Increasing the number of purifo:ation cycles prior to

testing also increased the dielectric: strength of the oil.

Testscarrit!d l)UI with the brass electrodes at a 1 mm gap

Numerous studies were earned out on the influence ol spacing after 2. 5, and 8 purification cycles showed that

impurities on the dielectnc stttngth of various dielectric the mean dielectric strength increased by about 8% after

fluids (7, 9, IO, 11). Conducting impurities and/or im 5 purlficarton cycles. and by nbcct 10% after 8 purifrra

purities with a lugher dielectric coastanr lhan the liquid lion cycles - in comparison with the mean dielectric:

arc attracted into regaons of high electric stress and strength after 2 purification cycles (a 6m filter was

eventuallyfonn a bridge. Tlus particle bndge may pos used). The improvement in oil quality is .:i likely result

sibly lead to breakdown. of the fact that the distribution of particle sizes moves

30 IEEE Elect1ica1 Insulation Magazine September/October 1990-Vot 6. No. 5

tu vmaller particle sizes as the

number of purification cycles in

c rcascv
10
Conlin u ous tl ow of the o il

(velocity= 3 cmh) was found to

mt re,H,e the mean drclcct nc

brass

st r e n g th. The mean d ielec tric 5

strength uf the LHl increased by
>
.x:
aboul 15?0 for the brass electrodes

with gap spacmg<; of 1, 2, and 4 w

0,

mm, and by about 25% for the !)!

small mild steel electrodes with 0

gap spacings in the r;rnge ot [) 1 to

>
c;
0.5 mm. For the large steel

cylm d ncal electrodes and for 0
"O
gaps of l , 2, 3, and 6 mm, the
rn
mean dielectric strength was w

found to increase bv about!:!%. "

ro 0.5 small steel electrodes

The increase of the dielectric c;

rn
strength of the oil can be ex
co
plamed by assuming either that

the oil flow impedes the entry of "

tmpuntres mto the gap (as was

suggested in [51) or that the 011

motion dday-. the P<,t.ibh"hmt'nt o.1---1

of particle bridges between the
0.1 0.5 1 2 4
electrodes [3]. The oil flow did not
Gap (mm)
ceern to affect the standard devta-

tiun and the skewness of the

Fig. 7. Mean breakdown voltage vs. gap spacing, for buth brass and sm,:ill steel
diclcctrrc strength divtributum-,
electrodes, with flowing oil (solid circles) and sta tic oil (open circles).
in any particular way. Thts con

cl usron i;.a tv anancewi th I 31, where

il wasclamied that a norrnahs ahon

Effect of an External Capacitance
of the dielectric strength distribution takes place as a
The source impedance seen by the test gap includes
rr-vultof them 1 flow and a:,, the, elocity i-, increased.
two separate capacitances: the capacitance uf the volt
Calculations for the various electrode profiles,

agcpowcr source and tho <,p]f rnp..u rtance of the test cell.
usmg the 011 flow velocity stated above, indicate
These two components affect the discharge current
that the Re} nold I\umber was approximately equal
waveform at breakdown The first component cannot
to 6 for the small steel electrode system, about 10 for

normally bo modrtied muc rt 1-. an mtnnsic property of

the Bruce electrode system, and as high as 300 for

the largC' -.t(tI e-ler trnde- "}ltm Smee turbulent the power supply. The second componenl, however,

flow is associated with Reynolds Numbers of from can be modified since the test cell capacitance depends

2000 to 20,0CJO, it appears that the flow was indeed on the electrode area and the gap spacing It may also

larnrnar. be- modmed by adding m parallel with the tesl cell

As was already pomted out, the change in dielectric various values of capacitors. The capacitance ot the test

strength was ;,ignificant with an oil velocity of 3 on/s cell is important to the stabilization phenomenon (131,

Other researchers [4] claimed that the velocity needs lo since electrode damage aud oil dotonoratton an- af

be several tens of cm/s to cause a ;,ignihcant change m fected by the current waveform and ils energy dis

dielectric &trl'ngth. r-rn to be

I luv, however; docs not sc sipated m the gap. for the test cell with Bruce electrodes

nght m the light of more recent data [12J. An mdefnute and for gaps of 2 and 4 mm, it was found that a

increase of dielectric strength is not expected, since the capacitance of 'i7{] pF r-onra-rted m parallel with the test

111CT't:'J'*' nf the ml vp\oc1ty c auve-, the flow to becorne cell mcteased the dielecrnc strength of the oil, Figs. 2

turbulent. Turbulent flow results in a pressure dif and 3. The increase in dielectnc strength may be ex

tere.rce m the liquid and the consequent creation of gas plained by assuming that the energy released rn the gap,

bubbles, which reduce the dielectric strength when ii breakdown occurv, 1-. optimum for Jt>t,tn,ying

IEEE Electrical lnsulatlOfl Magallne September/October 1990-Vol 6, No. 5 3l

the most prominent protruvronv cm

the electrode surfaces. The energy

released in the gap makes the

electrode surfaces smoother, thus

increasing the plateau dielectric

strength of the 011. A capacitance

15
connected in parallel with the test

cell affects the dielectric strength in

E
a complex way The optimum value E

of the capacitance depends on the

actual test cell as well as on the dis

tance of the capacitance from it [14].

Capacitance values less than the op-

'5i 10
c

tun um will probably not smooth s

out all the electrode protrusions suf ii5

ficiently, whereas values greater

than optimum will result in exces

u
srve electrode surface damage ,md 1l 5

0
oil deterioration

Effect of Gap Spacing

The dependence of dielectrrc

1 10 100 1000
strength on gap spacing has been

reported by vanous workers m the Area (cm2)

field of liquid dielectrics [4, 6]. In

this arbcle, the effect of gap spacing

on the ac dielectric strength will be Fig. 8. Area effect for the large steel electrodes and the brass electrodes al 2 mm gap

du..c__
-uvsed. Tests with the brav, and
spacing, with flowing oil (solid circles) ;i.nd stalk oil Copen circles).

the small steel electrodes for gaps of

0 1 4 mm W('H' earned out, a wdl

as tests with the large steel electrodes for gaps of 1 -- 6 It is genera Hy arccpted that the gap pacmg dft'Ct on

mm. All tests were performed with both static (i.e , dielectric strength is an essentially complex phenome

change of the 011 sample after each breakdown) and non. An increase of gap spacing may increase the

flowing oil (i.e , contmuous circulation with constant population of impurities entering the gap. These im

velocity of 3 cm/). Some of the results obtained with purities may act as nucleation sites which eventually

the large steel electrodes are presented in Fig. 4. It can reduce the dielectnc strength of the oil. The gap spacing

be seen that the gap effect is approximately 80 effect will be discussed again below, in relation to the

area effect.
kV/cm/decade change (and in fact it remains so for

areas up to 1240 cm2). The variation of the mean break

down voltage with the logarithm of gap spacing for the

Effect of Electrode Area
brass and small steel electrodes is shown m Figs. 5 and
t
I is well known that the dielectric strength of oil
6. The mean breakdown voltage varies linearly with the
increases with decreasing electrode area [I , 4]. The
loganthm ot gap pacing according to the following
results of experiments carried out with the Bruce elec
expression.
trodes and U1e large steel electrodes using static and

flowing oil are shown in Fig. 8 (static and flowing are

V = Kd''
used in the previously defmed sense). The reader will

realize that the dielectnc strength decreases with in-

where K is a constant, d the gap spacing and n an creasing electrode area, but the reduction r decade

exponential factor between O and l, as suggested by did not have the same value for the whole range of the

other researchers [15, 16] In the present situation, n is tested electrode areas. For small areas, up to 20 cm2, the

about 0.7, wluch shows that the breakdown voltage is rate of change was approximately 70 kV /cm/ decade;

somewhat less than proportional to the gap spacing, bul rt fell at about 35 kV /cm/decade for areas m the

Fig 7, a result agreeing with data published elsewhere range of 30 to IOOJ cm2; and to almost zero for areas

[17]. greater than 1000 cm2. A non-linear .relallonslup exists

32 IEEE Electrical Insulation Magazir)0 September/October 1990-Vol. 6. No. 5

bclwecn the dielectnc strength and the logarithm resulting in ii total voltage effect, i e , a dich-crnc

of the area, and a levelling off of the dielectric strength strongly dependent on gap spacing (19].

strength with 1ncreasmg clectrod area is ob

served The fact thc1t the area effect is weaker than the gap

Bt-armg in mmd the present area effect data and the spacing effect means that not only weak links but also

preccdmg results on the gap spacing effect. one can physical size tactors mfluence the breakdown mecha

justifiably c1rgue that a simple volume effect docs not nism d('CJ<,Jvdy lt J<, worth noting that what 1 being

exist, bccau-, otlu-rwice area and gap '>pilcmg effects reported here was also noted by others. who reported

should be equal. One may mfer that the dielectric that both the e.lectrode area and the stressed oil volume

strength is a function ot two variables, the area and may affect the dielectric strength of large 011 volumes:

the gap spacing; and that it is not uniquely dependent and that depending un thr- expenmental conditions.

on the stressed oil volume. Srnr e the gap <pacing either parameter may be Lhe most prominent (20]. It has

effect 1'i stronger than the area effect, rt may be said also to be said that the stronger dependence of dielectric

that the ac breakdown m transformer oil depends strength on gap spacing rather than on electrode area

very much on electrode geometry The strong gap has been found to hold true tor ,lid m<,ufatmg

pacing etfect may be attributed to th,, rrnp'n-vse-d matcrtal-, [21J. Future wmk which may elucidate lhe

voltage which increases as the gap mcreascs, thus possible common denorrunator underlying this be

creating more dcomposition products and electrode havior in both solid and hquid insulants should be

damage [4]. It has already been suggested [18] that undertaken.

although both weak links frxisting ctther in the ml

vulurm- or lJJ\ the electrode surfaces) and physical size

factors affect the dielectric strength, the latter may Conclusions and Suggestions for
sometime;, prevail in causing breakdown, being more Future Research
sensitive to gap spacing than to electrode area. Pos

&ihk phvvn-al 1,.;P tac tors are Experiments ca rrrcd out with umform field brass

and mild steel electrodes have shown that the

I) the collect 1011 process of particles frnm OlJ tside dielectric strength depends on various factors such

the stress-ed volume via the gap edge. The profile as stabilization, electrode area, gap pacmg, 01!

and size of the electrode edge mfluence the rate of veluuty and self capaotance of the lr.,\ cell. The

particle accumulation in the gap which m turn i breakdown phenomenon, although it appears to be

dependent, ,1mong other factors, on gC'ometry, weak link rnifiated, dews not result m a simple size

electrode size, electric held and the number of efrect. Possibly, physical size tac tors play an impor

particles. their dnsity and their shape; tant role in the breakdown process togclhC'r wilh

Llw statrstu-al srze factors.

2) liquid flow in the gap trom both deliberate The stabrhzanon process depends mainly on lhe

external pumpmg and intnm,K oil motion due to electrode area and its roughness. Oil flow was

electric stress. Liquid flow affects particle motion shown to be important, even when the velmity 1.,

and consequently the breakdown It is alreadv only d few cm/s The ronnerlion of 1
, capaolor

known that it 1& more difficult to uniformly flush parallel to the tt'i,,t cell may have ii beneficial effect

a gap ot only ,
l tew mm <1& the electrode area on the dielectric strength of the oil. It has also been

n-croases: shown that the volume effect cannot be expressed

as a umgue function of the area and thr- gilp spvc

3) stored electrostatic energy m the neighborhood rng, but rather di,, a fundion of the 11... 0 components.

of the gap. If insufficient energy is available, the It is suggested that the greater ,c,ensitivity of the

d.15eharge proce% takes longer to develop and to dielectric strength to gap spacing is due to the faster

lead to breakdown; accumulation of large particls m the gap cntenng

the stressed volume via the non-umform field

4) m.i.bilily to reproduce the small system guabty in region at the electrode edges.

a large system closer to industnal practice. It is very The work descnbed here did not attempt to reveal any

difficult to reproduce the same quabty of surface new aspects of the physics concerning the breakdown

fmish and oil purification for large systems since mechanism It was, howe ... er, an attempt to better un

tluv involves a lot of ume and cost, and dcrstand ;,omt' of the acpect-, of msulatmg liquids,

.\Je1tht>1 the statements made m th1, article nor the

5) the charge on ii particle may be dependent on the concepts put forward are thought to be the final one;

gap voltage and not iut the local electric field, thus Further research needs to be donr- The stronicr de-

IEEE Electncol Ir-surotron Mag071ne September/October IQ90.Vci 6. No 5 33

pendeoce of <lielectnc strength on gap spacing than on [hi Maks,P,...wokt, J T, and Tropper, H , Some Factors Affl"ctmg

the Measurement of the Electric Strength of Orvnic I iqmds .

electrode area needs further experimental evidence,
l'roc 11:1:, Vol. 101, pp 183-190 (1954)
and the factors resulting in such behavior have to be
[7] Watwn, PK and Higham, JB "Electric Breakdown of Trans
dari6ed. Fast time-resolved current and voltage meas former Oil", Proc IEE, Vol. 100, pp 168-173 (1953)

urement of the breakdown event [22, 23], as well as high [SI Cro,s, ID dnd Ma.:uNk, B., "Fast Cathode Processe m

speed optical '.tudie-., might be fruitful in this respect. Vacuum Du;charge Development, T'roc Xllth Int Symp on DIS

moe. Ins. Vac., pp. 47-48 (l 98!i)

The oil flow and its influence on dielectric strength also
[9] Zem El-Orne, M E and Trapper, H, "The Electric Strength of
needs further clarification, since the transition velocity
Transform CT 01!", Proc IEE, Vol. IU3, pp 35 45, (1956)

o be deter
fat which the onset of cavitation occurs) has t [10] Crine, JP, "Newl,- Dl"Hlopl"rl Analv!Ical Techmques for

mined as wdl as the factors which possibly affect it. Characterization of lnsulatmg Oils", Electncal lnsulatmg Oils,

SJ'P 998, hds. Ame11can iely 101 Testing and M.,teri.,ls,

Ph1lad,.Jphia, rr so-an (1988)

[11] Zdky, A A et .,1 . "Eff...._t of Ele<trod<' Polanty and Add,tiv,

Acknowledgments
on t
he lJuect Breakdown Voltage of S1l1cone 011 under Highly

Non-Uniform Fields', IEEE Trim, on Elec lmul, Vol El23, No

This work has been performed at the University of 6, pp 961-970 0988).

Newcastle-upon- Tyne, Newcastle-u pon-Tyne, [12] Cross, J . D , -Breakclown AuoM, ., D,elednc Spacer m In

5ulatmg Oil and the Role of I:lectrohvdrodynam1cs 1n L,quid

England. TI,e author is grateful to Dr. W. R. Bell for
Breakdown", JEEE Trans on Elec Insul, Vol. EI 17, No 6, pp.
stimulating discussions during the experimental work.
493-498 (1982).

He deeply thanks Dr. IJ. Fruth,of the Asea Brown Boveri [13] Hancol<. R and Tropper, I I , '"Th .. Braldown ofTr,mformer

Research Center, Baden-Dattwil, Switzerland, for a Oil under Impulse Voltages", Proc. [EE, Vol 111311., pp 250-262

(1958)
critical reading of the manuscript.
[14] Jaksts, A and Lross, I . D , "High peed Slre.,l l'hologr.,phy

of the Breakdown of TrnnformPT Oil and th<> DPp<"nd<>ncP ol it

Nature on Local Stored Energy , [EEE Trans on Elec Insul , Vol

Michael G. Danikas received the B.SC. degree in El-IS, No 6, pp 5\1'1-604 (1983).

EJectrical and Electronic Engineering and the M.Sc. de [15] Kawaguchi, Y. ct al , "Breakdown of Tranformer Oil". TFFF

gree from the University of Newcastle-upon-Tyne, Trans on Pow. App Sys!., Vu!. PAS-91, No 1, pp 9-19 (1972)

[16] Bell, \l'i K, 'lnlluence of pec,men Si1e on the D1dednc

England, in 1!ffl(I and 1982, respectively, and the Ph.D.
Strength of Electrical lnsulatmn-, Ph n Th,.s1s, Umv of Ncw
degree from Queen Mary College, University of Lon
c.,tle-upon Tyne, England (1974)

don, England, in 1985. He spent two years doing re [17] Ganger, BE., '"!he Hreakdown Vollage of OJI Gaps '""1th

search on vacuum insulation and on partial discharges High DC" Voltag,.""", IEEE TrM on Puw App Srt., Vol l'AS-87,

in polyethylene at Eind hoven University of Technology, No. 10, pp. 1840-1843 (1968)

[IS] Hell, W.K. aud Damkas, M , Fddors Affectmg the Break

the !\"etherlands, before joining Asea Brown Boveri Cor
down Stttngth of Transformer 011 ', Proc !CI:r lnt Symp on
porate Research at Baden-Datlwil, Switzerland, where
Elec. Insul, pp. 264-267 n982)

he studied long-term performance of rotating 119] Fel.enlh.,L P dnd Vonnegut, B., "Enhanced Charge Tuns fer

machinery insulation, partial discharges in epoxy-resin m Dielectric Flmds Conta1n1ng Conductmi, Puhcl ..s", Bnt. I

samples, as well as the discharge behavior of Kapton Appl. Phys. Vol. re. pp 1801-IBO!i (1%7)

[201 G1<to Tnnh, N. ct ,1l., tMistical Dielectric Degradatrnn of

cables for space applications.
Large Volume 0!1 lnsulatton, ICEC Trans on Pov. App Sys!.,

Vol PAS-IOI, K o. 10, pp 1-9 1982)

(

[21] Cygan, S a nd Wghar1. J R .. 'D ..pemlence of the Ekdm

References Strl"ngth on Thickness. Area an d V olume oi l'ol\prop)lene,

IEEE Tran un Fl...- l n su l , Vol FT-22, No 6, pp 83'i-83 7 (1'187)

[I] 5harlxlugh, A.H. et ,1, "Progro,M m the held of hleclr,c [22] Fuhr, J el .ii, "Spdrk Bredkdown of L1qmd Hydrocarbons -

Bre,ikdown m Ou,l=tnc Liqmds", !Elli: Trans on Elec Insul, I. t Current

l:'a, a nd Voltage Measurements of the SpMk B re,1k

Vol. El-13, Nu. '1, pp 24\1-276 0978\. down in Ltqutd n-Hexan, J Appl P hys, Vol 59 (Il l pp

3694-3 701 0986 ).

[21 Chad band, W.C., "'lhe l'os1t,veS<reamer .16 the Link belwt:en

Breakdown Pmces,;es ,n Gases and L,qmd,-, Proc. lnl. Con!. on [23] Fuhr, J and Sc hmidt. WF., "Sp.,rk BN!akdown of 1
L qm d

Oicl. M.i.t., as. and Appl. pp. 5-8 (l9SS) Hvdrocarbons - Il Temporal Development of the Electric Spark

(3] Nel.sun, J.K. el al, "Electrn; Stn'n!,,i;h of Tansfurmer Oil for R esistance m n-Pen tne, n-Hexane, 2,2-Dlmcth}lbutJne, and n

L.trge Electrode AR""as", Prnc IEE, Vol. llS, No. 2, pp 388--393

oecaae-, J. Appl Phys., Vol. 59 (11), p p. 3702-3708 (1986)

(1970. [24] Fuhr, J nd Schm,dl. W r. Th,. lnflucncl" of Succcs51vc

(4] Bell, W.K, "lnrtuence of SP",inu,n s;.,,., un the D,eli;ctnc le!ectr1ca! Breakdown on the Breakdown Voltage of n Hexane",

Strength. of Tr.insfonncr Od". U:t::h "li"an on Elec. lnsul., Vol IEEE Trdn". on a. lnsul, Vol. er-zs. No 1, pp '185-187 (1988)

El 12. Nu. 4, pp. 281-292 (19771 [25] Ze,n Cl-Dine, M E . el 1 . 'Influence ol Insulatrng f1lm5 on

Conduction, Proc IEE, Vol. 112. Nu 1, pp 5H!l-58'i (1%'i\

151 Ck,th.ier, N.D. et ,
i i .
. "Sol,d Particle lmpuT1ties m Power

Transformers", Elec. R-v, Vo!. 166, pp MS-65() (1970)

34 IEEE Electrical lr'\SllkJtlon MagaZ1ne September/October 1990-Vol. 6, No. 5