Relationship between power factor and resistivity of oil and the power factor of impregnated paper

E. C. R. Scarfe, B.Sc, A.C.G.I., A.M.I.E.E., and R. G. Pulsford, B.Sc, A.R.C.S. Synopsis

An investigation has been carried out into the effect of ionic and other additives on the power factor of oil-impregnated paper, with the object of relating the power factor of the impregnated paper to the power factor of the impregnant. For power-factor measurements, an improved test cell was used which was designed to operate at high stresses over a wide temperature range. It was observed that there was a variable relationship between the power factor and the volume resistivity of oils depending upon the type of additive. There was no correlation between the power factor of impregnated paper and the power factor of the impregnant, but a definite relationship was found between the power factor of impregnated paper and the volume resistivity of the impregnant, this relationship being independent of the type of additive.

Introduction

In order to produce oil-impregnated-paper dielectrics with the low power factors required for e.h.v. cables, it is necessary to avoid the use of materials that might increase polar losses or ionic conduction. Water-washed wood-pulp paper with low power factor is selected, and the cleanliness of the oil impregnant is also normally judged by its power factor. A considerable amount of data exists on the effects of various contaminants on the power factor of hydrocarbon oils'" 6 and on the power factor of impregnated paper.7"10 Attempts have been made to relate the power factor of impregnated paper to the power factor of the impregnant, but these have usually been unsuccessful. Jt has also been observed that there is no relation between the power factor of oil extracted from oil-filled cables and the power factor of the cables either directly after manufacture or after they have been in service.6 This is often attributed to difficulties in sampling from cables, but the work reported here suggests that such a relationship probably exists only for certain limited types of oil contaminants. Our own investigation into the effect of oxidation products on the electrical properties of oils lead to the observation that, whereas contaminants affected both the power factor and volume resistivity p, there was no constant relationship between power factor and volume resistivity. It has been shown by Balsbaugh and Assaf," and confirmed by Feick, Eich and Olds,12 that the dielectric losses in oil are mainly ionic, and it is suggested that d.c. measurements may be substituted for power-factor measurements. We have found that this applies only to a limited class of materials as shown in Fig. I, where power-factor measurements made on contaminated oils are plotted against the equivalent volume resistivities. The theoretical relationship between power factor and resistivity, assuming all losses are ionic, is given by p tan S = 1-63 x I010 for an oil with a dielectric constant of 2-2 and a measuring frequency of 50c/s, where p = resistivity (ohmcm) and tan 8 = power factor for the low values under consideration. This is represented in Fig. I by the line towards the lower left-hand corner. Only a few measurements fall near this line, and as might have been expected, they were of oils to which fatty acids had been added. Most other substances affected power factor to a greater extent
26th February 1964. It was presented at the Conference on Dielectric and Insulating Materials, 9th April 1964 Mr. Scarfe and Mr. Pulsford are with Associated Electrical Industries
Ltd.

Paper 4606 S, first received 2nd December 1963 and in revised form

PROC. IEE, Vol. 112, No. 3, MARCH

1965

than resistivity, and so, as shown in Fig. 1, the theoretical relationship did not apply. The reason for this variable, but controlled, relationship was not at first obvious, and variables such as moisture content and testing stress were examined. The d.c. measurements were normally carried out at a stress of 3-3kV/cm and the a.c. measurements at 6-6kV/cm, but varying these had very little effect on the power-factor/volume-resistivity relationship. Moisture contents were varied between 5 and 50 parts in 106,also without effect. This variable relationship is undoubtedly due to the behaviour of different types of contaminant when under an electric stress. It is known that ionic materials in oil cause high a.c. losses, as well as d.c. conductance, and it is also found that materials which are not ionic in character, but are of a colloidal nature, produce high a.c. losses owing to the motion in an alternating field of large charged particles. Materials of a colloidal nature can be formed in oil by oxidation, and are also produced under corona discharge.4 Lt is convenient to represent them experimentally by a material such as bitumen which readily forms a colloidal system simply by addition to oil. Those materials that are largely colloidal, such as bitumen, cause high a.c. losses, but the colloidal particles are rapidly swept away in a d.c. field by electrophoresis and hence do not have such an appreciable effect on the volume resistivity (which is normally measured after 1 min electrification). Such materials would thus cause greater a.c. losses than would be expected from d.c. measurements, and if plotted in Fig. 1 the results would fall above the theoretical line. A carboxylic acid is capable of dissociating into ions, particularly in an electric field, and the 1 min volume-resistivity value and the 50c/s power factor will both be a measure of the ionic losses and these measurements will fall on or near the theoretical line. This power-factor/volume-resistivity relationship is of interest in itself as a guide to the type of material causing high dielectric losses in an oil, but of greater importance is the effect on the power factor of oil-impregnated paper. It was predicted that those oils with power-factor/volumeresistivity relationships towards the upper right-hand corner of Fig. 1 would not affect the power factor of impregnated paper because of the colloidal nature of the additives. The colloidal particles would be large compared with the pore size of the paper and so would be trapped in the paper, their effect being eliminated or considerably reduced. Ionic materials would be expected to increase the power factor of 613

impregnated paper because they would continuously dissociate in the oil phase, and, owing to their small magnitude, would be less liable to become trapped in the paper fibres. It was thus expected that high power factors would be produced when paper was impregnated with oils whose powerfactor/volume-resistivity relationship occurred towards the

the oil sample degassed and finally passed into the cell to impregnate a paper sample under vacuum. A strip of lowdensity, low-impermeability, water-washed paper, 0-005in thick, was wrapped tightly round the inner electrode assembly,

0-1

001 \o

>. o

1)

0-001

7s

guards

\ %.

D10 ptan 8-1-63 x 1 <

0-0001

Fig. 1

10 100 volume resistivity, 0 cm

1000x1012

stand

Relation between volume resistivity and power factor of oils containing additives, at 85C
Additives: V bitumen A additive A O elastomers + fatty acids oil from cable ducts

Fig. 2
High-voltage test cell

lower left-hand corner of Fig. 1. This would divide contaminants into categories similar to Piper's A andB, 9 but it would also provide a convenient means of predicting these categories solely by the measurement of power factor and volume resistivity without knowing any details of the nature of the contaminant. Jn order to examine the effect of various additives on the power factor of impregnated paper, the following method was employed. 2 Experimental method

extending well beyond the upper and lower guards, the number of turns being such as to fill completely the radial test gap of 1 mm. The cell was heated by insertion into a thermostatically controlled oil bath to a temperature of 100C 0-5C and evacuated to 0 1 mm Hg. This heating and vacuum treatment was continued until constant power-factor readings were obtained, applying a low test voltage to prevent ionisation. The paper sample was then assumed dry under these particular conditions. The oil sample, after degassing, was slowly passed into the cell under a maintained pressure of 0 1 mmHg. When the cell was filled, the vacuum was broken with dry air, and power-factor measurements were taken at 85 C and 20 C for various stresses. Additives The oil used in these investigations was the lowviscosity cable oil generally used in Britain for the impregnation of oil-filled cables. It is a naphthenic-based oil blended with sufficient aromatics to give a positive hydrogen absorption under conditions of corona discharge. It has a viscosity, at 60C, of about 4-5cs and is refined to give a low power factor. When tested in a standard Henley cell at 85 C and a stress of 6-6kV/cm, the power factor of this oil is normally less than 0 0005. The additives were chosen to represent ionic and colloidal materials such as fatty acids and bitumen, respectively. Other substances were used in order to examine their possible effect on the performance of oil-filled cables. Typical of these were additive A (the purpose of which is not relevant to the subject of this paper) and synthetic-rubber compounds which have been considered for use in the manufacture of cable-jointing accessories.
PROC. IEE, Vol. 112, No. 3, MARCH 1965

For these experiments, use was made of a high-voltage oil test cell originally developed for the determination of the dielectric loss of low-viscosity cable oils at high electric stresses and at elevated temperatures.v The cell incorporates the basic design principles of the well known Henley oil cell13 but modified to enable it to be filled under controlled vacuum conditions and to permit testing stresses up to 120kV/cm over a temperature range up to 120C. This represents conditions approximating to those in oil-filled cables. Fig. 2 shows a sectional view of the cell. All metal parts that are in contact with the oil under test are of stainless steel and the supporting and separating insulators are of ground quartz. The O-ring technique was used for sealing under reduced pressures. A particular feature of this cell is that volume-resistivity measurements can be made directly and without calibration, since the test specimen is of defined geometric form. In the present experiments, the cell was incorporated in a closed circuit in which the cell could be evacuated and heated,
614

An initial series of measurements on oil-impregnated paper was made using oils containing a variety of contaminants in sufficient concentration to give power factors of 0 01 when measured under standard conditions in a Henley cell. The concentrations were then varied to give a range of power factors both below and above 0 01. Results The initial series of oils tested, that is those having a power factor of 0 01 at 85 C, verified the prediction that contaminants could be divided into categories depending upon their power-factor/volume-resistivity relationship. For instance, oils to which bitumen had been added to give a power factor of 0 01 had little, if any, effect on the power factor of impregnated paper when measured at 85 C or at room temperature, whereas oil contaminated with oleic acid to give a power factor of 0 01 had a considerable effect on the power factor of impregnated paper, particularly at elevated temperatures. It thus became clear that no correlation exists between the power factor of impregnated paper and the power factor of the impregnant. This is demonstrated in Fig. 3, which shows the power factor of impregnated paper plotted against the power factor of the oil. After further consideration, it became clear that the most important property of the impregnant was its volume resistivity, rather than power factor, and a definite relationship was found between the volume resistivity of the oil and the power factor of impregnated paper. This relationship is shown in Fig. 4. The power-factor values shown were those obtained on the impregnated paper at a stress of 60kV/cm. There was little variation in power factor with stress for paper impregnated with oil of high resistivity, and there was a slight reduction in power factor as the stress was increased on paper impregnated with oil of low resistivity. The lower the resistivity of the oil, the greater was the fall in power factor of the impregnated paper with increase in stress. Discussion Many attempts have been made in the past to compare the power factor of impregnated paper with the power factor of the impregnant, and it has usually been concluded that the
01

two properties are quite independent of each other. The work reported here confirms this view, but it shows that a definite relationship exists between the power factor of the impregnated paper and the volume resistivity of the impregnant. This relationship applies to a considerable range of contaminants over a considerable range of volume resistivities.

O-l

O-OI

O-OOI 01 volume

I resistivity of oils,

IO rv.cm

IOOXIO

Fig. 4
Relation between power factor of impregnated paper and volume resistivity {at 85C) of the impregnant in the presence of additives.
Additives: A O +
V

additive A elastomers fatty acids oil from cable ducts

bitumen

The reason for this effect lies in the difference in behaviour of ionic and colloidal particles, as suggested in Section 1. The trapping of colloidal particles in the paper fibres must be very complete, however, since it was found that an oil with a power factor as high as 0 06 had no more effect on the power factor of impregnated paper than had an oil with a power factor of 0 0001 (provided that the high power factor was caused by a nonionic colloidal material). The presence of ionic materials, or those materials capable of dissociating into ions under an electric stress, produces conduction in the oil phase and probably increases the conduction of the paper itself. Volume resistivity is solely a measure of ionic conduction and hence is of greater significance than power factor when judging the suitability of an oil as an impregnant for oil-paper dielectrics.

O-OO6
.

-J~

/ O-OO4 / / < c b -

o-oi

4O

2 o

f^

= =
.

+
9 * O-OO2 /

o-ooi

O-OOI

O-Ol

O-l

I-O

power factor of oil

Fig. 3
Relation between power factor of impregnated paper and power factor of the impregnant in the presence of additives, at 85C
Additives: V bitumen A additive A O elastomers 4- fatty acids oil from cable ducts

O-l

O-2

O-3

O4

OS

O-6

O7

percentoge

moisture

Fig. 5
Power factor of impregnated paper, at 20 C
a b c d Unimpregnated paper Impregnant: oil, tan 8 = 00001, p > 560 x lO'^Qcm Impregnant: oil, tan S = 0039, p = 56 x lO'^ficm Impregnant: oil, tan 8 = 0019, p = 0 9 5 x lO'^flcm

PROC. IEE, Vol. 112, No. 3, MARCH

1965

615

Conclusions It has been shown that there is no correlation between the power factor of the impregnant and the power factor of impregnated paper, but there is a relation between the volume resistivity of the impregnant and the power factor of impregnated paper. These conclusions have resulted from measurements in a test cell where the paper has been applied as a roll. Although such conditions exist in practical applications such as bushings, in cables the paper is applied as tapes with butt gaps between turns to permit bending. The thickness of the oil in the butt gaps is considerably greater than the thickness of the oil film between paper layers, and it is possible that colloidal particles in these butt gaps may contribute to the power factor of the impregnated paper. Also, the dryness of the paper in the test cell is undoubtedly better than can be achieved during cable manufacture, and it is possible that some materials might contribute to a greater extent towards increasing power factor in the presence of traces of moisture. These two aspects have formed the subject of a separate investigation using an apparatus designed so that the moisture content of the paper can be controlled and slots can be introduced into the paper to represent butt gaps. It is not the intention to give details here, but typical results are shown in Fig. 5. These results confirm the above conclusions, even though the paper contains moisture and excess oil in the butt gaps.

Acknowledgment The authors wish to thank Associated Electrical Industries Ltd. for permission to publish this paper. 8
1 2 3

References
PIPER, J. D., THOMAS, D. E. F., and SMITH, c. c . : 'Liquid dielectrics',

Industr. Engng. Client., 1936, 28, p. 309

PIPER, J. D . , FLEIGER, A. G., SMITH, C. C , a n d KERSTEIN, N. A . :

'Liquid dielectrics', ibid., 1939, 31, p. 307

PIPER, J. D . , SMITH, C. C , KERSTEIN, N. A., a n d FLEIGER, A. G.:

'Liquid dielectrics', ibid., 1940, 32, p. 1510 4 STICHER, j . , and PIPER, J. D. : 'Corona discharge on liquid dielectrics', ibid., 1941, 33, p. 1567
5 6 PIPER, J. D . , FLEIGER, A. G., SMITH, C. C , a n d KERSTEIN, N. A. I

'Liquid dielectrics', ibid., 1942, 34, p. 1505

GAZZANA-PRIAROGGIA, P., PALANDRI, G. L., a n d PELAGATTI, U. A . :

'Influence of ageing on the characteristics of oil-filled cable dielectric', Proc. IEE, 1960, 108 A, p. 467
7 8 PIPER, J. D., THOMAS, D. E. F., and SMITH, c. c.: 'Oil-impregnated

paper', Industr. Engng. Chem., 1936, 28, p. 843

PIPER, J. D., KERSTEIN, N. A., and FLEIGER, A. G.: 'Oil-impregnated

paper', ibid., 1937, 29, p. 1040 9 PIPER, J. D., and KERSTEIN, N. A.: 'Oil paper dielectrics', ibid., 1944, 36, p. 1104
10 PIPER, J. D., TREEND, P., and BEVIS, K. s.: 'Oil paper dielectrics',

ibid., 1948, 40, p. 323 11 BALSBAUGH, j . c , and ASSAF, A. G. : 'Electrical stability of electrical insulating oils under limited oxidation', Trans. Amer. Inst. Elect. Engrs., 1943,62, p. 311
12 FEICK, G., OLDS, w. F., and EICH, E. D. : 'The chemistry of cable oil

deterioration', / . Electrochem. Soc, 1961, 108, p. 56 13 BRAMLEY, L. N. : 'An improved cell for the a.c. and d.c. testing of insulating oils', / . IEE, 1940, 86, p. 326

616

PROC. IEE, Vol. 112, No. 3, MARCH 1965