IEEE Transctiors. or Electrical Insulation Vol. EI-22 No.6, December 87 81

COMMUNICATION

A NEW APPROACH To MEASUREMENT OF THE FIGURE-OF-MERITFOR STRONGLY ELECTRONEGATIVE GASES AND GAS MIXTURES

Y. Qiu and Y.F. Liu

Xi'an Jiaotong UniversityXi'an, China

ABSTRACT

This communication presents a new approach to measurementof the figure-of-merit for strongly electronegative gasesand gas mixtures. Compared with Pedersen's method, whichneeds very high precision measurements, the present methodcan be used in most laboratories. The difference betweenour results and those obtained by Pedersen et al. is notmore than 3%, which is acceptable for engineering applicationresearch work.

INTRODUCT I ON

It is well known that local field enhancement causedby electrode defects greatly affects the breakdownstrength of strongly electronegative gases or gas mix-tures. To assess the sensitivity of gaseous dielectricsto the effect of electrode defects, Pedersen [1] intro-duced a figure-of-merit M which is defined as

M = _ K6 (EIP) LimMI. K (1)

This communication presents an alternative method,which does not need very high--precision of measure-ments, yet can give satisfactory results for engineer-ing application research work.

DIFFICULTIES ASSOCIATED WITH PEDERSEN' S METHOD

Applying Eqs. (2) and (3) to a uniform field gap witha gap length d, one can easily find a linear relation-ship between the breakdown voltage Vb and the Pd value.

where K is a constant, but variable for different gasesand gas mixtures, in the streamer breakdown criterion;and S (E/P)liim can be found in the net ionization ex-pression.

fXc (u-r ) dx = K (2)a - n= [E -(E/P) Zim P] ( 3)

This figure-of-merit M together with the theoreticalbreakdown threshold (E/IP)iim can be used to evaluatethe discharge behavior of gases or gas mixtures in anelectrode system with surface defects [2].

To determine the figure-of-merit, Pedersen and hiscolleagues measured the breakdown voltage of the gas orgas mixture under investigation in a uniform field gapwithin the Pd range where the linear part of thePaschen curve can be observed [2,3].

Though very simple in principle, Pedersen's methodrequired a very high degree of precision, which is quitedifficult to achieve in many laboratories. To date,however, that method has been the only way of measuringthe figure-of-merit for electron-attaching gases andtheir mixures [2-5].

K=-=V - (E/P) Lim Pd (4)

For pure SF6, K/,>0.35 kV, (E/P)jim-88.6 V/Pa.m. Ithas been reported that the linear relationship betweenVb and Pd can still be observed when Pd goes down to0.067 kPa.m [6], but Pedersen and his colleagues choseto measure the figure of merit at a Pd value of 0.7kPa.m corresponding to a breakdown voltage of 61.8 kVin SF6 [2,3]. It is therefore quite clear that to en-sure a 5% precision for the K/, value, the measurementprecision for Vb in Pedersen's case should be 2.8xl0-4.

Moreover, Pedersen's method needs such a high degreeof field uniformity for the gap that the following ex-pression [3] should hold

Eo - Emin < ME 0 Pd (5)

where Eo and Emin are the average and minimum fieldstrengths, respectively, along the gap axis.

Take pure SF6 as an example. When Pd=0.7 kPa.m aswas the case in Pedersen's experiments, the above con-dition becomes

g.ZI 5- , I -f- _,-- ' 7-7c7 Ir-

3z I

TEEE Trsnss:-tions on ElectriCalI TnSulatio- V.;a. EI-22 No.6, December 1987

E- EminE0 <5"7xI 03

Therefore, great care must be taken in designing theelectrode profile to achieve a very high degree offield uniformity.

THE PRESENT METHOD

1.0

2 0.8

0.6For a nonuniform field gap, the following expression

can be written.

0.4

f,x [E - (E/P) lim FP] dx = KlS (6)

If the electric field distribution function E=f(V,x)and the (E/P)Lim value are known, then it is possibleto determine the K/, value according to the dischargeinception voltage V0 (or the breakdown voltage Vb for aslightly nonuniform field gap).

It is not difficult to compute the electric fielddistribution for commonly used electrode configurations.In the present work a sphere-plane gap was used. Thediameter of the sphere was 2 cm with the gap lengthbeing 1 cm. A charge simulation method was used tocompute the electric field distribution with potentialerrors at check points on the electrode surface shownin Fig. 1 being <1X1O-4.

2 cm

d=1 cm

Fig. 1: The disposition of a point charge and ninering charges for fieLd computation (x: potentiaZcheck point on the electrode surface).

Fig. 2 compares the electric field distribution, com-puted using the charge simulation method (solid curve),with that calculated using the following approximateequation (dotted curve)

E(x) = (r+x)Emax (7)

in which x is the distance from the surface of thesphere electrode, and Emax can be expressed as

Emax = fV/d (8)

where f is the computed field nonuniformity factor,which was found to be 1.864 for the 1 cm sphere-planegap used in this work.

Fig. 3 gives two examples for pure SF6 and an SF6 gasmixture containing 90% N2 by volume. It can be seenthat the critical avalanche length in SF6 and the SF6

0.2 _

0 0.2 0.4 0.6 0.8 1.0i (cm)

Fig. 2: Electric field distribution in the 1 cmsphere-plane gap. soZid Zine: computed usingthe charge simulation method; dotted Zine:calculated using Eq. (7).

100

E Z^ 88,6 (SF6)80

60 \ 570 ( 10'oSF6+90EN2)

40- ~ ~ _

20

0 0.2 0.4 0.6 0.8 1.0

x (cm)

Fig. 3: Computation exampZes for pure SF6 and theSF6/N2 mixture containing 10% SF6.

gas mixture is less than 1 mm and 1.5 mm respectively,which are much smaller than the fraction of the gaplength where the computed field distribution coincideswith Eq. (7). Therefore, in calculating the K/6 valueEq. (7) can be used.

The (E/P)lim value can be derived from measurementsof the ionization and attachment coefficients. It canalso be obtained with sufficient accuracy from break-down experiments in a uniform field gap.

(E/P)Lim V1bm K (9)

Qiu and Liu: The -figure-of-merit for strongly electronegative gases and gas mixtures

When Pd.l km.Pa, the second term on the right-handside of Eq. (9) is less than 0.5% of the first term.In the present method an accuracy of 0.5% for (E/P)Jimis sufficient. Hence the following approximation canbe taken in evaluating the (E/P) Zim value.

(E/F) Znm = Pd1)

Fig. 4 shows that the (E/P) rim values for SF6/N2 gasmixtures obtained in our breakdown experiments are ingood agreement with those given by Pedersen [2], andalso by Aschwanden [7] from measurement of swarmparameters.

100

-

le,

_E

was kept at 0.1 MPa in all experiments. Negativepolarity dc voltages were applied to the sphere elec-trode and were measured by an electrostatic voltmeterwhich had been calibrated using a 0.1% precision re-sistive voltage divider. The applied voltage wasraised quickly to about 1% below the expected Vb, andthereafter at a rate of approximately 1% of Vb perminute until breakdown occurred. The scatter of themeasured breakdown voltages was not more than 1%(Table 1) even without any artificial irradiationwhich is very crucial for impulse breakdown voltagemeasurements in small gaps [8,9].

Table

Breakdown voltages of SF6/N2 gas mixturesin the 1 cm sphere-plane gap.

60_

40J

20_l_I

0 20 40 60 80 100SF6 (%)

Fig. 4: Percentage values of (E/P) Lim for SF6/N2gas mixtures. o: the present work; V: Pedersenet al. [2,3]; x: Aschwanden [7].

The critical avalanche length can be calculated usingEq. (7).

xc = r -Ea/F -r

(E/P) rim

K/6 can thus be determinedequation.

K-,I L ~- E ma-

(11)

by solving the following

(E/F)rim P] dx

(12)

Table 2 lists the K/6 and M values obtained for SF6and SF6/N2 gas mixtures. The results obtained byPedersen et al. were also included in Table 2 for com-parison purposes. The difference between the resultsof the present work and those given by Pedersen et al.is not more than 3%.

Table 2

K/3 and M values for SF6/N2 gas mixtures.

SF6 (P) 1 K (kV) k(,a m.MPa)l (kV/ni,MPa) this work Refr43] this work Ref4.3]

100 88.6 0.344 0.35 3.88 4.o

73.1 85.1 0.355 0.37 4.17 4.3

50 79.0 o.416 O.42 5.27 5.3

10 57.0 0.829 o.84 14.5 14.7

f r xcVb

d(r+x0)

RESULTS AND DISCUSSION

Both the 2 cm diameter sphere and the plane which hada flat part of 5 cm diameter with a simplified Rogowskiprofile were made of brass and were polished before theexperiments. The pressure of SF6 and SF6/N2 mixtures

It is interesting to note that Eq. (12) is similarin form to Eq. (4), yet the terms on the right-handside of Eq. (12) are about lOx smaller than those cor-responding terms in Eq. (4). That is why the preci-sion requirement for breakdown voltage measurements inthe present work can be one order less than that inPedersen's method. The limitation of the presentmethod is that in addition to the sphere-plane gap, a

SF6 Vb (kV)

(%) No. 1 NO. 2 No..3 No.4 No.5 aver.

100 53.3 53.3 54.1 53.5 54.o 53.64

73.1 51.8 51.8 51.9 51.9 51.3 51.74

50 48.4 48.6 49.1 49.1 48.6 48.76

10 38.4 38.2 38.6 38.3 38.5 38.40_~~~~~~.I I

E3-T-T

IEEE Transactions on Electrical Insulation Vol. EI-22 No.6, December 1987

uniform or quasi-uniform field gap is still needed fordetermining the (E/P)lim value so that the criticalavalanche length in the sphere-plane gap can be known.Otherwise, as stated by McAllister [10], the sphere-plane gap is not suitable for measuring the figure-of-merit.

CONCLUS IONS

1. The present method, whose precision requirement forbreakdown voltage measurements can be one order lessthan that of Pedersen's method, can readily be used inmost laboratories to determine the figure of merit forstrongly electronegative gases or gas mixtures.

2. The difference between the M values for SF6 andSF6/N2 gas mixtures obtained in the present work andthose published by Pedersen et al. is not more than 3%which is quite acceptable for engineering applicationresearch.

REFERENCES

[1] A. Pedersen, "Evaluation of the effect ofsurface defects on breakdown in strongly electro-negative gases or gas mixtures", Gaseous Dielec-trics II (Edited by L.G. Christophorou), Perga-mon Press, pp. 201-209, 1980.

[2] A. Pedersen, "On the assessment of new gaseousdielectrics for GIS", IEEE Trans. Power Appara-tus and Systems, Vol. PAS-104, pp. 2233-2237,1985.

[3] J. Berril, J.M. Christensen, A. Pedersen,"Measurement of the figure of merit relatedto the effect of electrode surface defectson breakdown for strongly electronegativegases or gas mixtures", Proc. 7th Inter.Conf. on Gas Discharges and Their Appli-cations, pp. 266-268, 1982.

[4] J. Berril, J.M. Christensen, I.W. McAllister,"Measurement of the figure of merit M for1-C3F6 1,2-C2C12F4 and C2ClF5", Proc. 8thInter. Conf. on Gas Discharges and TheirApplications, pp. 239-242, 1985.

[5] J. Berril, J.M. Christensen, I.W. McAllister,"Measurement of the figure of merit M forseveral strongly electronegative gases",Conference Record of 1986 IEEE Inter. Symp. onElectrical Insulation, pp. 251-257, 1986.

[6] H.A. Boyd, G.C. Crichton, "Uniform-fieldbreakdown-voltage measurements in sulphurhexafluoride", Proc. IEE, Vol. 119, pp. 275-276, 1972.

[7] Th. Aschwanden, "Swarn parameters in SF6 andSF6/N2-mixtures determined from a time re-solved discharge study", Gaseous DielectricsIV (Edited by L.G. Christophorou and M.O. Pace),Pergamon Press, pp. 24-34, 1984.

[8] Y. Qiu, "Impulse breakdown characteristics ofenclosed sphere gaps", Fourth Inter. Symp.on High Voltage Engineering, 33.07, 1983.

[9] Y. Qiu, M. Zhang, R. Liu, I.D. Chalmers,"Effect of wavefront on impulse breakdownvoltage of slightly nonuniform field gapsin nitrogen, air and SF6", IEEE Trans. Elec.Ins., Vol. EI-21, pp. 575-578, 1986.

[10] I.W. McAllister, "On the choice of a test gapfor strongly electronegative gases and gasmixtures", Gaseous Dielectrics IV (Edited byL.G. Christophorou and M.O. Pace), PergamonPress, pp. 195-203, 1984.

Manuscript was received on 25 September 1986, inrevised form 3 February 1987.

84