Influence of humidity on the breakdown voltage

of d-c. and a.c. voltages in air

K. Feser

Translated from Bulletin ASE 63 (1972) 6, p. 278.. .281

1 BE 756 1
Influence of humidity on the breakdown voltage
of d.c. and a.c. voltages in air

Kurt Feser

Work is reported which has had as its objective the in- b) Negative pre-breakdowns: Humidity has no effect on
vestigation of the influence of humidity on the breakdown any known type of pre-breakdown.
voltage of spark gaps in air. In the spark gaps investigated, Arising from this comparison comes the fact that, up to
breakdown resulted from the impulseless glow corona. Up now, not much notice has been given to the effect of
to now no complete investigations exist of the effect of humidity on breakdown voltage, when a impulseless
humidity on the glow corona. The steps necessary in glow corona precedes the breakdown.
practice and the regulations for the correction of humidity For this reason investigations are described in the fol-
are indicated. lowing, the purpose of which is to measure the effect of
humidity on the breakdown voltage of air gaps in which
the impulseless glow corona precedes the breakdown at
the positive electrode. The impulseless glow corona is a
1. Introduction form of discharge which can be observed in a.c. and d.c.
voltages particularly at sharp edges and on thin wires.
In recent years different research work has become A glow corona in formation acts in a stabilising manner,
available [l] [2] [3] [4], in which investigations have been that is to say, the breakdown voltage of a gap can be in-
carried out on the breakdown voltage of air gaps. From creased through this form of discharge. In practice. fo1
these investigations therefore it is seen, that the effect of example, grid electrodes are used as shield electrodes
humidity on the breakdown voltage in air can always be
observed, if pre-breakdown precedes the breakdown on
a positive electrode. Corresponding to this main con-
dition, humidity does not in practice effect the breakdown 2. Test arrangement and evaluation of results
voltage in an homogeneous or near homogeneous field.
As long as the breakdown occurs without pre-breakdown, In the following the effect of humidity on the breakdown
the effect of humidity on the breakdown is slight (e.g. voltage from the glow corona in d.c. and a.c. voltages is
0.25%/g/m as shown [l]). Similarly the effect of humidity investigated. With lightning voltages impulseless glow
on the breakdown voltage can be disregarded, if pre- corona does not occur [9]. The measurements were taken
breakdown occurs in the spark gap only on a negative in the air-conditioned laboratory of the High Voltage
electrode [ 11. With impulse-like pre-breakdown (streamer) Institute of the Technical University, Munich (6.5 x 6.5 x
on a positive electrode, the effect of humidity on the 6.5 m) [l] [6]. A 400 kV transformer was used as the
breakdown voltage is dependent to a critical extent on the source for the a.c. voltage which was supplied from the
dimensional and durational development of the pre- mains via a single-phase transformer. Measurement of
breakdown. the a.c. voltage was carried out with a capacitive voltage
A study of the literature on the different forms of pre- divider (C = 65 pF). It was possible to attach a one-way
breakdown, relative to the effect of humidity on the rectifier (rectifier, smoothing capacitor) to the transformer
breakdown voltage, shows that, with the exception of the for the generation of the d.c. voltage, so that the d.c.
impulseless glow corona, there are already measurement voltage could be measured with an ohmic voltage divider.
results available which show the main gradients for every With the aid of a bushing the test voltage could be
form of pre-breakdown. For practical use, there are al- switched through to the test arrangement.
ready correction graphs in existence, although these do The chosen arrangements ought preferably to build-up
not explain the physical reasons for the effect of humidity the impulseless glow corona. Employed as test objects
on the formation of the pre-breakdown and thus of the for this reason were either a vertical built-in rod-plane
breakdown voltage. spark gap (rod diameter: 50 mm rod-end with an edge
The following are the principal considerations in the radius r = 1 mm at a spark distance of a = 10 cm or
effect of humidity on the breakdown voltage which have r = 2 mm at a = 20 cm and a = 30 cm), or a cylindrical
been measured up to now : spark gap (outer diameter: 40 cm or 20 cm. Diameter of
the inner conductor 10 mm or 6 mm).
a) Positive pre-breakdowns : These dimensions were necessary for the rod-plane gap,
aa) Dark discharges (no pre-breakdown): Effect of hu- in order to prove that the breakdown resulted from the
midity can be disregarded. glow corona and not from the shaft of the rod during the
ab) Impulse-like pre-breakdown : Greatly influenced by action of the glow corona. The cylindrical spark gaps
humidity [5]. were so dimensioned that the breakdown occurred in the
ac) Impulseless pre-breakdown: Effect of humiditv un- cylinder field.
known. The measured values of the breakdown voltages were
evaluated in accordance with statistical methods. 20 300
breakdown tests were carried out for each measured
point. The average value of the breakdown-voltages, the
standard deviation and in the event of a normal distri-
bution also the confidence limits of 95% were determined
graphically and by calculation, with the help of the proba-
bility-paper. The breakdown voltages were also calcu- 200
lated under the normal conditions (20C 760 torr) cor-
responding to the recommendations of the International kV
Committee of Electronic Techniques (CEI) [7].
t
-ii G l o w voltage
+ 100
d.c.,: * :
ue ,*7--r- - - -
300 a.=.- -------x

kV

O-
10 g/m3 15
y:---

Fig. 2: Effect of absolute humidity cpa on the breakdown voltage Ud of

a 20 cm rod-plane spark gap
Type of electrode: blunt with 2 mm radius
Glow voltage See Fig. 1 for further notations
d.C..

U, d . c . - - * - - --x
a.=,-5--,-- --x-- -
that if the breakdown results from the impulse-like
streamer discharge (from cpa > 6.5 g/m3), the breakdown
voltage increases with rising humidity.
In the results in a.c. voltage (Fig. 3) with a 10 cm gap
distance, the following is worthy of note. In this arrange-
3 5 10 g/m=
1

15
ment, breakdowns occur in a certain voltage region, for
Ta- example: at cpa = 9.5 g/m3 from 68. . . .81 kV, after the
onset voltage. The breakdowns however are still not able
Fig. 1: Effect of absolute humidity cpa on the breakdown voltage Ud of to heat up a low-ohmic channel. If the voltage is increased
a 30 cm rod-plane spark gap
Type of electrode: blunt with 2 mm radius
further (for example: at cpa = 9.5 g/m3 at 81 kV), the
Parameter: Voltage breakdowns cease. The region in which these breakdowns
Positive d.c. voltage (d.c.) can be observed is indicated by means of vertical dotted
a.~. voltage (a.c.) lines in Fig. 3. The breakdown, which also ends in an
U, onset voltage
intensive short circuit arc, occurs at this humidity at
Ud breakdown voltage
rpa absolute humidity about 130 kV. If the voltage is increased, not to the final
breakdown at about 130 kV, but only to about 120 kV,
then subsequently reduced further then the breakdowns

3. Results
200-

The results of the effect of humidity on the breakdown

kV
voltage for the rod-plane gaps are illustrated in Figs. 1 to
3. It is recognised that the onset voltage U, in d.c. and
a.c. voltages is about equal and increases minimally with
rising absolute humidity. After the impulse-like streamer
lOO--
discharge, the impulseless glow corona appeared, in the t
course of which, the onset voltage of the impulseless glow ii
3%
corona was measured only at the d.c. voltages as in Fig. 1
ac.
and 2. The d.c. breakdown voltage shows a decreasing
characteristic for all three gaps, that is with increasing
absolute humidity, the d.c. breakdown voltage decreases 0 1 I
if the breakdown results from the impulseless glow corona. 0 5 10 g/m3 15
The a.c. breakdown voltage shows the same tendency in -b --)

the glow corona. Because of the form of the a.c. voltage,

Fig. 3; Effect of absolute humidity cpa on the breakdown voltage Ud of
the glow corona developing in each half-wave is not so a rod-plane spark gap
stable, so that the breakdown occurs earlier at lower Type of electrode: blunt with 1 mm radius
absolute voltage values. In Fig. 1 it can be ascertained, See Fig. 1 for further notations

4
also occur again in the characterised region (68 kV 6 U < correction of humidity must unfailingly concern them-
81 kV), on reduction of the voltage. Explanation of this selves with the nature of the pre-breakdown, when a
behaviour lies in the pre-breakdown. An impulse-like correction graph is employed. The correction graphs
streamer discharge which is able to reach as far as the given in previous recommendations and regulations [7]
opposing electrode occurs as the first form of discharge [lo] are only valid, if streamer-corona occur at the positive
following the onset voltage, as the onset voltage for this electrode.
arrangement is higher (64 kV) than the voltage require-
ment of pre-breakdown (5 kV/cm for the streamer dis-
charge means 50 kV at 10 cm) up to the opposing elec-
trode. If a impulseless glow corona occurs at all, this has 5. Summary
a stabilising effect. That is to say, the breakdown voltage
only occurs at a higher voltage. From the measurements it is clear, that the breakdown
voltage of spark gaps in a non-homogeneous field is in-
fluenced by the absolute humidity. If the breakdown
results from the impulseless glow corona on a positive
electrode, the breakdown voltage falls with increasing
absolute humidity. At high absolute humidities no im-
pulseless glow corona occurs.

o;
0 5
9=-
10 g/m3 15

Fig. 4: Effect of absolute humidity cpa on the breakdown voltage Ud of

a cylindrical spark gap
kV

Outer cylinder D = 40 cm, Inner conductor d = 10 mm, Length of the

outer cylinder: 90 cm
See Fig. 1 for further notations - I

Even with the cylindrical spark gap, at which a further

considerably stabilising glow corona can be formed, Fig. 5: Effect of absolute humidity cpa on the d.c. voltage breakdown Ud
results comparable in practice were achieved. As is to be of a cylindrical spark gap
seen in Fig. 4 and 5, here too, the onset voltage is in Outer cylinder 20 cm. Inner conductor d = 6 mm. Length of outer
practice not affected by the absolute humidity, whilst the cylinder 60 cm
See Fig. 1 for further notations
breakdown voltage decreases with increasing absolute
humidity, as long as the breakdown results from the The measurements presented were carried out at the Institute for High
impulseless glow corona. Voltage at the Technical University, Munich.

4. Discussion on the results References

If the breakdown results from the impulseless glow [I] E. Peschke: Effect of humidity on the breakdown and
corona, the breakdown voltage thus decreases with the flash-over at high d.c. voltages in air. ETZ-A 90 (1969) 1.
increasing absolute humidity. This drop in the breakdown P. 7...13
voltage means that the voltage region in which the glow [2] H. Winkelnkemper and W. Baucke: Effect of humidity
corona can occur becomes less and less. In high absolute on the breakdown in air of high a.c. voltage. ETZ-A 90
humidity no glow corona can be observed. This limit (in (1969) 25. P. 667.. .670
the arrangements investigated) lies under 15 g/ms.
A possible explanation of the drop in the breakdown [3] S. Guindehi: Effect of air humidity on the breakdown
voltage could lie in the adsorption of the photons. Whilst voltage in air in various types of voltage and forms of
this adsorption in streamer discharge works to increase electrodes. Bull. SEV 62 (1970). P. 97.. .104
strength, with the impulseless glow corona a weakening
[4] K. Feser: Effect of humidity on the breakdown be-
of the glow corona occurs, since too few negative ions
haviour in a.c. voltages. ETZ-A 91 (1970). P. 584.. .586
(which are important to keep the glow corona alive) [8]
exist in areas of weak fields. [5] K. Feser: Effect of the type of voltage on the breakdown
The falling graph line in rising absolute humidity indi- behaviour in spark gaps in air with non-homogeneous
cates, that national and international regulations on the field. Energie und Technik 22 (1970) 9. P. 3 19.. .324

5
[6] H. Print: Fire, lightning and spark. Munich. Bruck-
mann, 1965.

[7] High tension tests. Publication No. 60 of the CEI.

Second edition. Geneva. Central Office of the Inter-
national Electrotechnical Commission. 1962

[8] W. Hemstein: The streamer discharge and its tran-

sition to glow corona. Arch. Elektrotechn. 45 (1960) 3.
P. 209...224

[9] E. Nasser: Ionising potential waves at the spark break-

down. Zeitschr. Physik 172 (1963) 4. P. 405.. .428

[lo] Generation and measurement of high voltages. Part I :

Regulations for the generation and use of a.c. and d.c.
voltages for test purposes. VDE Regulation 0433.
Part I/ 1.66

Address of the author:

Dr. (Eng) K. Feser, Emil Haefely & Co. AG, 4028 Basel,
Switzerland.