Behavior of Insulation and Electrical Equipment at Low Temperatures

A. A. Vorob'ev

Introduction

In connection with the rapid growth of industry in the north and east of the Soviet Union, the study of the behavior of various electric-current installations under low-temperature conditions has acquired great importance[^1].

The expanded conference on electrical insulating materials, held in May 1940 in Moscow, pointed to the need “to intensify work on investigating the behavior of the insulation of electrical machines and apparatus under operating conditions”[^2].

The need for accelerated work on studying the behavior of the insulation of machines and apparatus under low-temperature conditions was also indicated in the resolutions of the conference on combating energy losses at thermal power stations and accident rates in the Kuzbass networks, convened by the Tomsk Industrial Institute in June 1940[^3].

For a number of years, the author of the present article, together with staff members of the high-voltage laboratory of the Siberian Physico-Technical Institute, and subsequently in the high-voltage laboratory of the Tomsk Industrial Institute, has been carrying out a systematic study of the operation of the insulation of electrical apparatus and of the apparatus themselves under low-temperature conditions.

The cold winters of 1938 and 1939 once again emphasized the importance of this scientific problem for the electric-power engineering of the Kuzbass regions, where the temperature fell to minus 57°C and for a considerable time remained below minus 40°C.

The apparatus of outdoor electrical substations, transformers, oil circuit breakers, and their insulation, installed in the open air, are subjected to the action of low temperatures.

The existing practice of operating electrical apparatus and electrotechnical materials, based chiefly on operating experience in regions with a high mean annual temperature, does not answer a whole series of questions concerning the behavior of apparatus under low-temperature conditions, whereas these questions are of acute interest to operating engineers and designers.

The present state of our knowledge of the structure of dielectrics and their behavior under operating conditions makes it possible to foresee the character of changes in certain properties of dielectrics when the temperature is lowered. However, from a practical standpoint, knowledge of the quantitative aspect of these changes for technical dielectrics is of interest, and physical theories do not yet provide it.

We investigated changes in the electrophysical characteristics of technical insulating materials—transformer oil, paper, varnished fabrics, porcelain, compounds, and other materials—when the tempera-

ture. On the basis of the experimental materials obtained on the behavior of dielectrics at low temperatures, studies were carried out on oil-filled transformers and circuit breakers.

In the present article an attempt is made to set forth systematically a part of the materials obtained.

MEASUREMENT OF THE DENSITY OF TRANSFORMER OILS AS A FUNCTION OF TEMPERATURE CHANGE

The change in the density of oil as the temperature is lowered is of direct interest. An oil whose density at low temperature will be greater than the density of ice at the same temperatures is undesirable for use in service, since in this case ice may float up and short-circuit the current-carrying parts,\(^{4}\) although it should be noted that such rise will be impeded by the high viscosity of oils at low temperatures.

In our experiments, the density of the oils was measured by the pycnometer method. Some of the results obtained by us are presented in Fig. 1. As can be seen from Fig. 1, the temperature course of the density of various oils is linear, and the angular coefficients for the various straight lines are different.

Fig. 1.

The dependence of the density of a liquid on temperature is given by the formula

\[ d_t = \frac{d_0}{1+\beta t}, \tag{1} \]

where \(d_0\) is the density of the liquid at \(0^\circ\), \(d_t\) is the density of the liquid at temperature \(t^\circ\), and \(\beta\) is the mean coefficient of volume expansion between \(0^\circ\) and \(t^\circ\). According to expression (1), the dependence between the quantities \(d_t\) and \(t\) is hyperbolic. The linear dependence obtained by us experimentally evidently indicates the inapplicability of this formula to oils containing water, or hydrocarbons freezing out at comparatively high temperatures. The temperature dependence of the density of such a two-phase system (liquid + solid), as can be seen from Fig. 1, follows a linear law and is expressed by the equation

\[ d_t = a + bt. \tag{2} \]

For normal operation it is essential that the density of oils remain less than the density of ice at the corresponding temperatures. The density of ice in the temperature interval from \(0^\circ\) to minus \(30^\circ\text{C}\) changes from 0.917 to 0.919 g/cm\(^3\), while the densities of the oils tested by us in these intervals varied from 0.885 to 0.927 g/cm\(^3\).

DEPENDENCE OF THE VISCOSITY OF TRANSFORMER OILS ON TEMPERATURE

Viscosity is one of the important characteristics of oils used for filling oil circuit breakers. An oil with high viscosity is unsuitable for these apparatuses, since in it the interruption of current and the extinction of the arc will occur too slowly. Low viscosity of oils is desirable also because then particles suspended in the oil settle more readily to the bottom and, conse-

Consequently, the oil is less prone to contamination. In less viscous oils, under the action of an arc a smaller amount of carbon is formed, and they are more readily filtered.

A great many works have been devoted to the study of the nature of the viscosity of liquids. A survey of the literature on this question may be found in the article by A. A. Leont’eva^5).

In a theoretical consideration of the question of the temperature dependence of the viscosity of liquids, this dependence is usually obtained as exponential. However, the corresponding formulas describing the change in the viscosity of a liquid with temperature contain a considerable number of constants that cannot be determined from experiment. Theoretical formulas, while correctly describing the qualitative aspect of the phenomenon of the change in viscosity of simple liquids with change in temperature, do not make it possible to carry out quantitative calculations.

For the multiphase system of interest to us—which oil becomes at low temperatures after the crystallization of water and of hydrocarbons boiling at comparatively high temperatures—the formulas of the theory of the viscosity of liquids are as yet inapplicable.

We measured viscosity by the falling-ball method, with subsequent calculation of the viscosity according to Stokes’ formula with the Ladenburg correction. For these measurements several different viscometers were constructed in our laboratory.

The measurement of viscosity by the Stokes method was chosen because this method better imitated the role of oil viscosity in the operation of an oil circuit breaker than methods of measurement based on the flow of liquid through a tube (the Poiseuille method and its modifications).

For the twelve oils we measured, two types of dependence of viscosity on temperature were obtained; these are shown in Fig. 2 and Fig. 3. In Fig. 2

Fig. 2.

the dependence of the logarithm of the viscosity, expressed in poises, on temperature is shown. Such a dependence was obtained for a very light-colored oil.

In the temperature interval investigated (\(+13^\circ\) to \(-36^\circ\)C), this oil gives a 150-fold increase in viscosity. The form of the dependence of viscosity on temperature fits the formula

\[ \eta = 0.5 e^{-0.025 t}, \]

where \(\eta\) is the viscosity in Engler degrees. After water was introduced into this oil, its viscosity at a temperature of \(-15^\circ\)C did not change appreciably, although the breakdown voltage at room temperature fell from 28 kV to 23 kV according to a standard spark gap.

The viscosity of the remaining oils changes with temperature in a different, more complex manner.

In the temperature range investigated, for some oils a change in viscosity by a factor of 200,000 was observed. Fig. 3 presents the dependence of the logarithm of viscosity on temperature for one such oil. A strong dependence of oil viscosity on temperature is known for paraffin-base oils. For these oils the viscosity increases very rapidly as the temperature is lowered near zero. The oils we used in the experiments give, in this region down to temperatures of minus 10°–12°C, a linear dependence of the logarithm of viscosity on temperature. At lower temperatures a very rapid increase in viscosity begins, strongly dependent on prior stirring of the oil. A series of experiments with stirring of the oils by the falling ball, by a stirrer, and by other means show that the high value of the viscosity of the oils obtained is due, in its origin, to the presence in the oil of a certain structure, which is destroyed by stirring, leading to a decrease in viscosity. In Fig. 3 the crosses indicate the value of the oil viscosity after stirring. If stirring is carried out only by the measuring ball itself, then a viscosity value is obtained that decreases after each fall and becomes established only after the eightieth fall. As an example of the influence of stirring, it may be noted that for one of the dark oils the viscosity at a temperature of minus 40°C before stirring was 200,000°E; after stirring, the viscosity at the same temperature proved to be 10,000°E. For another light oil, the corresponding values were a viscosity of 1200°E before stirring and 500°E after stirring.

Fig. 3.

In the temperature interval from −40° to −50°C the viscosity of oils increases greatly, including, for the above-mentioned light oil, from 1200°E to 10000°E.

It is possible that the formation of structure in oil at low temperatures occurs in the same way as dendritic crystallization of metal in molds.[^6] It is known that when metal cools in a mold, crystals grow from the walls of the mold toward the center, owing to the presence of a temperature gradient in this direction. When oil is cooled in a glass cylindrical beaker until a temperature is established that is the same at all points along the radius, a certain temperature gradient will exist. In this case the long hydrocarbon molecules line up in the direction of the temperature gradient; owing to the high viscosity of the oils at these temperatures, the molecules retain their arrangement. Pseudo-crystallization takes place; a lattice is formed, which is disrupted by a body moving through the oil. Cooling of oils below the temperature at which the stirring phenomenon manifests itself is not accompanied by the precipitation of any sediments. Attempts to detect the presence of structure by means of light scattering in the oil were unsuccessful.

It is known from the literature that, when determining the pour point of an oil, it must be at rest. With stirring, the pour point ...

the oil may decrease by 3–5 degrees. Thomsen^7) believes that this phenomenon is connected with the destruction of the network of solid hydrocarbons formed in the oil at low temperatures. In particular, the paraffin contained in the oil crystallizes out in clusters and increases the overall viscosity of the oil mass.

To clarify the possible role, in the formation of structure in the oil, of charged and polarizable molecules, N. A. Prikhod’ko carried out experiments to determine the viscosity of oils at low temperatures in the presence of an electric field. The measurement was performed in a flat capacitor with a distance between the plates of 2.5 cm; a direct voltage of 10–12 kV was applied to the plates.

The results of these measurements are presented in Table No. 1.

Table No. 1

As can be seen from Table 1, measurement of viscosity in the presence of an electric field immediately gives high viscosity values, higher than measurement in the absence of an electric field. Moreover, viscosity measurements in an electric field made immediately after stirring give high viscosity values, which indicates the high speed of the processes occurring in the oil in the presence of an electric field and leading to high viscosity. The influence of the field on viscosity is manifested at temperatures below −18° and −20°C, i.e., approximately at the same temperatures at which structural viscosity begins to appear. At lower voltages, for example 80 V, the influence of the electric field was not noticeable. Further experiments showed that dissolving certain substances in the oil leads to the elimination of structural viscosity at low temperatures^8).

There are indications in the literature of the possibility of reducing the viscosity of oil at low temperatures by heat treatment^9) of oils. It is considered that the main factor influencing the freezing temperature of oils is paraffin, which is present in almost all mineral oils. Therefore, depending on the amount of paraffin in the mineral oil and its state, different oils will have different freezing temperatures. In addition, the remaining components of oils will also affect the freezing temperature, since to a first approximation it is the arithmetic mean of the freezing temperatures of the components. The freezing temperature is also strongly affected by the presence of surface-active and resinous substances which, owing to adsorption on the surface of the paraffin crystallization centers, can influence the freezing temperature even when present in the oil in small quantities. For such oils there is a dependence of the viscosity near the freezing point on the prior

heating; moreover, it manifests itself the more strongly, the lower the temperature at which the viscosity measurement is made. Preliminary heating of oils can reduce the viscosity by a factor of two. This phenomenon, associated with the simultaneous presence in oils of solid hydrocarbons of the paraffin series and resinous substances, is explained by the different forms of paraffin crystallization after heating and by the ability of resins, under certain conditions, to retard it.

N. A. Prikhod’ko investigated the influence of preliminary heating on the viscosity of oil at low temperature. Before measurement the oil was heated to \(80^\circ\text{C}\) and then cooled to minus \(29^\circ\text{C}\). The viscosity of this oil at a temperature of minus \(29^\circ\text{C}\) before stirring proved to be \(350^\circ\text{E}\), and after stirring, respectively, \(240^\circ\text{E}\). If the oil, after heating, is held for a day, then the viscosity values for such an oil turn out to be the same as for unheated oil, namely, respectively, \(220^\circ\text{E}\) before stirring and \(350^\circ\text{E}\) after stirring; this, obviously, was to be expected, since the oil undergoes such heating repeatedly during its production and operation. Thus, heat treatment of oils gives only a temporary effect of lowering viscosity.

Attention is drawn to the fact that the pour point of the oils investigated, determined according to the standards, is minus \(48^\circ\)—\(55^\circ\text{C}\) and satisfies the standards. However, beginning at a temperature of minus \(35^\circ\text{C}\), owing to the presence of structural viscosity, these oils have such a high viscosity that they can no longer be used for filling oil circuit breakers. The currently accepted methods of measuring viscosity, based on observing the outflow of oils from tubes under pressure, give an incorrect idea of the true viscosity of oils at low temperatures. In these measurements, stirring of the oils occurs, leading to the destruction of structural viscosity, which, according to our data, can reduce the viscosity by a factor of 10—15. Therefore it seems to us advisable to introduce into the oil-testing standards a viscosity test by a method that makes it possible to measure the full value of the viscosity, including the structural viscosity as well. The pour point of oils, determined by the test-tube method recommended by the standards, cannot characterize the suitability of oils for use at low temperatures.

DEPENDENCE OF THE BREAKDOWN VOLTAGE OF OILS ON TEMPERATURE

The temperature dependence of the breakdown voltage of oils has been well studied for temperatures above zero\(^{10}\). Data on the breakdown voltages of oils of domestic origin at low temperatures have been contradictory. One may cite, for example, the statement of I. A. Glozman\(^{10}\) that there are isolated data on a decrease in the electric strength of transformer oil at low temperatures.

Breakdown of technically purified transformer oil is a complex phenomenon and may be caused by ionization phenomena in air inclusions, by thermal processes arising in oil situated in an electric field, and by a number of other causes\(^{12}\).

When the temperature of the oil is lowered and its viscosity increases, its electrical conductivity should decrease; at the same time, heat losses in the oil will decrease and the development of the thermal form of breakdown will be impeded. Therefore, as the temperature is lowered, an increase in the breakdown voltage should be expected.

In our experiments, breakdown of the oil was carried out between standard electrodes with a distance between them equal to \(1\ \text{mm}\). The data obtained—

…are shown in Fig. 4. Curve I represents the dependence of the breakdown voltage under impulses on temperature for oil No. 12, which has no structural viscosity. Curve II gives the same dependence for dark oil possessing structural viscosity. The course of the dependence for both oils is exactly the same. Curves III and V respectively represent the results of tests at alternating voltage for oil No. 12 and oil No. 11. Curve IV refers to the breakdown at alternating voltage of oil No. 11 that had been preliminarily dried. For it there is no jump in the breakdown voltage on passing through zero.

Thus, measurement of the temperature dependence of electric strength showed the following: on passing through \(0^\circ\mathrm{C}\), the electric strength of oils increases abruptly; moreover, in breakdown under impulses the magnitude of this increase is smaller, while at alternating voltage the increase is relatively greater. The jump in electric strength on passing through \(0^\circ\mathrm{C}\) is connected with the freezing-out of emulsified water. For well-dried oil this jump is absent.

Fig. 4.

After the freezing-out of emulsified water, different oils nevertheless have somewhat different strengths, which is evidently connected with dissolved water present in a state of molecular dispersion.

When the temperature is lowered below minus \(25^\circ\mathrm{C}\), the electric strength again begins to increase. This increase is hardly entirely due to viscosity, as is thought, for example, by Germant.\(^{13}\)

In general, attention is drawn to the rapid increase of the breakdown voltage of oils with decreasing temperature in the low-temperature region. It is known that in thermal breakdown we have a linear relationship between the magnitude of the logarithm of the breakdown voltage and the temperature of the dielectric.

Fig. 5.

• ● — At impulse voltage
• × — At alternating voltage

In Fig. 5 is shown the dependence of the logarithm of the breakdown voltage on temperature for the oils represented in Fig. 4 by curves I and VI. In Fig. 5, the circles indicate the values of the logarithm of the breakdown voltages

on pulses with a tail length on the order of \(10^{-4}\) sec.; the crosses indicate the values of the logarithm of the breakdown voltages under alternating voltage. As can be seen from Fig. 5, for both types of voltage the relationship between the indicated quantities proved to be linear. For impulse voltages, the relationship between breakdown voltage and temperature for the oils investigated at temperatures below zero is well expressed by the following formula

\[ U = 36 \varepsilon^{-0.1t}. \tag{4} \]

In testing under alternating voltage, the dependence of the magnitude of the breakdown voltage on temperature may be represented either in the form of a two-term formula or in the form of two separate formulas, namely, in the interval from zero to minus \(30^\circ\)C by the formula

\[ U = 25 \varepsilon^{-0.01t} \]

and for temperatures below minus \(30^\circ\)C by the formula

\[ U = 28 \varepsilon^{-0.291t}. \tag{6} \]

Judging by the form of the dependence between breakdown voltage and temperature, we must assume, in this interval of low temperatures, a thermal form of breakdown for oils.

Measurements of dielectric losses, carried out on the Schering bridge by N. A. Prikhodko\(^{13}\) for various oils, showed a decrease of the loss angle according to an exponential law as the temperature was lowered.

Such a change in the loss angle with decreasing temperature indicates that in this temperature region, at industrial frequency, the dielectric losses in the oil are due to electrical conductivity.

Thus, the study of the temperature behavior of the breakdown voltage of oil and of dielectric losses gives results indicating that, at low temperatures, dielectric losses due to conductivity and thermal breakdown occur in oil.

It is interesting to note that for some oils, in the temperature range of minus \(50\)—\(60^\circ\)C, a maximum of the dielectric loss angle appears. If the magnitude of the loss angle at \(0^\circ\) is equal to 0.08, then at this maximum \(\operatorname{tg}\delta\) has a value of 0.068, i.e., reaches 85% of the value of \(\operatorname{tg}\delta\) at a temperature of \(0^\circ\)C.

M. Mikhailov and G. Soya\(^{15}\), who studied the temperature dependence of \(\operatorname{tg}\delta\) for certain dielectrics, obtained a maximum of \(\operatorname{tg}\delta\) at a temperature of minus \(60^\circ\)C for pure ice, dry pressboard, paper, shellac, cellophane, and for a mixture of rosin with lycopodium. The measurements were carried out at a frequency of 50 Hz. In the opinion of these authors, it can now be asserted that this maximum is characteristic of all substances containing hydroxyl groups. A maximum of the loss angle will therefore also occur for plastics containing cellulose.

At high frequencies this maximum, caused by dipoles, shifts into the region of normal temperatures.

BREAKDOWN OF SOLID DIELECTRICS AT LOW TEMPERATURES

The experimental data presented above show that the breakdown voltage of oils, when the temperature is lowered below zero, always increases, and the relationship between these quantities obeys the equation:

\[ U = U_0 \varepsilon^{-at} \tag{7} \]

The constants entering into formula (7) are different for oil breakdown under impulses and under alternating voltage.

For a more complete judgment about the operation of apparatus under oil at low temperatures, it is necessary to know how the properties of the other dielectrics used in these apparatus will change.

N. A. Prikhodko\(^ {14}\) carried out breakdown tests of high-voltage porcelain at different temperatures in a uniform electric field under impulses. As the data of these experiments, presented in Fig. 6, show, the breakdown voltage of porcelain in the investigated temperature interval does not depend on temperature.

Fig. 6.

V. A. Vorob’ev.\(^ {16}\) studied the breakdown of paper insulation as a function of temperature. In these experiments the following materials were taken:

1. Presspahn 0.8 mm thick, with specific weight 1.38, hereinafter designated presspahn A. This presspahn was taken out of an oil transformer after long service.
2. Presspahn 0.56 mm thick, with specific weight 1.32, hereinafter designated presspahn B.
3. Cable paper KV—17, 0.17 mm thick.
4. Yellow varnished cloth, 0.20 mm thick.
5. Pertinax.

Measurements were made under alternating voltage and impulses with a tail length of the order of \(10^{-4}\) sec.

Fig. 7.

Pertinax was broken down under conditions of a uniform field, which was obtained in specimens having a spherical recess.

In breaking down paper, presspahn B, and varnished cloth, the electrodes used were a plane (a disk 25 mm in diameter with rounded edges) and a hemisphere 25 mm in diameter. Presspahn A, under alternating voltage, could not be broken down between the indicated electrodes because of large surface discharges; therefore it was broken down in an electric field—point against plane. Before testing, the specimens of paper and presspahn …

were first boiled in well-dried transformer oil at a temperature of 80–90°C for 4–5 hours. Cooling of the specimens and maintenance of the low temperature were carried out with the aid of a cryostat, the design of which was developed in our laboratory ^17). Breakdown was produced in transformer oil. The results of these measurements are presented in Figs. 7 and 8.

Figure 7 shows the dependence between the magnitude of the breakdown voltage and temperature under alternating voltage. Figure 8 presents the dependence between the same quantities under impulse voltage.

As can be seen, in both cases the breakdown voltage increases with decreasing temperature for all the materials tested. For presspahn A in oil, in the practically interesting range of low temperatures from 0° to minus 60°C, the breakdown voltage increases, in comparison with the breakdown voltage at 15°C, by 15%; for presspahn B by 40%; for pertinax by 9%; for paper by 30%; and for varnished cloth by 20%.

The breakdown voltage of fibrous insulation is determined, first, by the nature of the distribution of the electric field in the fiber and in the pores. Under alternating voltage the field distribution will depend on the ratio

Fig. 8.

of the dielectric constants of the substance of the fiber and of the materials filling the pores. Under direct voltage the field distribution will depend on the ratio of the specific conductivities of the substance of the fiber and of the material filling the pores. Secondly, it is determined by the effect of an increase in the breakdown voltage of the air or oil filling the channels in the fibrous insulation, as compared with a solid layer of the same thickness of these same materials.

In order to clarify the nature of breakdown, a study was made of the dependence of the logarithm of the breakdown voltage on temperature. It turned out that, for presspahn A under alternating current, the logarithm of the breakdown voltage increases linearly as the temperature is lowered, i.e., it satisfies a dependence characteristic of the thermal form of breakdown. For presspahn B, up to a temperature of 55°C, the dependence between the magnitude of the breakdown voltage in kilovolts and temperature is described by equation (7), where \(U_0 = 48\), \(\alpha = 0.077\). At temperatures above 55° the form of the dependence is preserved, but the values of the coefficients change and the dependence becomes stronger. For pertinax, up to a temperature of 55°C the breakdown is electrical, while at higher temperatures it is thermal.

In impulse breakdown, the breakdown voltage also proves to be related to the temperature of the dielectric by equation (7). In this case, for presspahn A the value \(U_0\) is equal to 40, and for presspahn B, respectively,

43 and for pertinax 37. The value of \(\alpha\) for all dielectrics proved to be the same and equal to 0.01.

If one compares the dependence of the breakdown voltage on temperature under impulse breakdown for oils, on the one hand, and for presspahn and pertinax, on the other, this dependence turns out to be identical and is expressed by formula (7). This circumstance makes it possible to consider that the breakdown of the indicated fibrous dielectrics is associated with breakdown of the oil in the pores of these dielectrics and with surface discharges in the oil.

For all the dielectrics investigated, the basic dependence characteristic of thermal breakdown is fulfilled; therefore one may think that for all the dielectrics investigated, at low temperatures, we are dealing with a thermal form of breakdown, with the exception of pertinax, for which a clearly expressed thermal breakdown begins only at temperatures above 55°C.

Fig. 9.

Proceeding from the data presented on the study of discharge voltages in oil as the temperature is lowered, one may expect that the initial voltages in oil as the temperature is lowered, as well as the discharge voltages over the surface of solid dielectrics, will increase.

Experiments on measuring the initial voltages in oil at low temperatures were carried out by N. F. Ryzhkova \(^{18}\). Using the photographic method, she determined the voltage at which corona appeared at a point in an electric field, point against plane, when the electrodes were immersed in oil. The results of these measurements are presented in Fig. 9, where the ordinate axis gives the voltage at which corona appears, and the abscissa axis gives the oil temperature. In these experiments alternating voltage was used. As can be seen from Fig. 9, as the temperature is lowered, the initial voltages in oil increase.

The above results of investigations of apparatus insulation show that the electric strength of fibrous materials impregnated with oil, and of the transformer oil itself, increases as the temperature is lowered; the electric strength of porcelain does not depend on temperature. Considering the problem of operating apparatus at low temperatures, we may say that, from the standpoint of the electric strength of apparatus insulation, such operation is entirely possible.

FROST RESISTANCE OF CASTING COMPOUNDS

For casting the bushings of apparatus and cable couplings, as is known, casting compounds are used.

Our electrical industry uses four grades of casting compounds: MB—80, MB—100, MK—55, and E—3. Compound MK—55 is manufactured on the basis of rosin, while the remaining compounds indicated are based on bitumen \(^{19}\).

We studied in detail^20) the properties of mass E–3 in the temperature range from 150°C to minus 60°C. This compound, produced by the Elektroapparat plant, is used for potting bushings VM–35 and MKP–76. According to our determinations, the cracking temperature \(Tg\) of this mass proved to be about minus 37°C.

It is known^19) that in the region of \(Tg\)—the temperature at which the compound transforms into a solid body—there is a sharp change in the dielectric permittivity in a constant electric field, in the heat capacity, and in the coefficient of expansion.

We believe that the disruption of the continuity of the substance that occurs in the region of the temperature \(Tg\) should also cause a change in other dielectric characteristics of the substance, namely: breakdown voltage, electrical conductivity, and dielectric losses.

Ionization processes that will take place in the air inclusions formed during cracking of the compound may lead to a significant reduction in the breakdown voltage of the compound and to an increase in electrical conductivity and dielectric losses. Measurement of the temperature dependence of the breakdown voltage, carried out by student N. A. Snegirev^20), showed that in the region of \(Tg\) there is indeed a sharp decrease in the breakdown voltage. Fig. 10 gives the dependence obtained by him for breakdown under alternating current.

Dependence of the breakdown voltage of compound E–3 on temperature under alternating voltage

Fig. 10.

The dependence of the breakdown voltage on temperature for a distance between electrodes equal to 2.5 mm is plotted in Fig. 10 by crosses; the analogous dependence for a distance between electrodes equal to 1 mm is plotted by circles.

The presence of a sharp change in the temperature dependence of the breakdown voltage in the region of \(Tg\), observed during breakdown under alternating and impulse voltages, makes it possible to determine accurately the cracking temperature of the compound.

Usually the \(Tg\) of a compound is determined either visually, by observing the appearance of cracks on the surface of the cooled compound, as we also did in our determinations, or by ear. One of the simple methods for determining \(Tg\) is Tammann’s method. It consists of the following: 1–2 cm³ of a substance melted in a test tube is subjected to slow cooling. Upon reaching the temperature \(Tg\), a light touch of the compound with a glass rod causes cracking of the substance throughout its entire volume, which can be noticed by the characteristic sound. Thus, the methods used for determining the temperature, although fairly simple, are subjective. Determination of the temperature \(Tg\) from the temperature dependence of the breakdown voltage is considerably more complicated, but at the same time it also gives us a quantitative result showing the degree of reduction

of the dielectric properties of the compound when the temperature is lowered below \(T_g\), which is necessary for practice.

The data presented above show that compound E—3, considered frost-resistant, proves to be so only down to a temperature of minus 35°C.

For filling cable sleeves of 6-kilovolt cable lines, a potting compound consisting of 70% rosin and 30% autol is used. The electrophysical properties of this material were studied in detail in our laboratory by A. K. Krasin and V. F. Bozhko \(^{21}\). In particular, the frost-resistance test of this compound carried out by them showed that, when tested in a porcelain vessel, cracking of the compound begins already at a temperature of minus 20°C. When tested in a cast-iron vessel, the compound cracks at the walls and across the vessel at a temperature of 22–25°C. Thus, this potting compound as well, despite a number of positive properties, cannot be recommended for use at low temperatures.

The results presented above from the investigation of two potting compounds used as frost-resistant show that their frost resistance is practically unsatisfactory. These data confirm the decision of the All-Union Conference on Electrical Insulation, in whose resolutions we read: “Special attention must be given to the question of developing and organizing the production of frost-resistant compounds” \(^{22}\).

OPERATION OF A TRANSFORMER UNDER LOW-TEMPERATURE CONDITIONS

The question of transformer operation under low-temperature conditions is closely connected with questions of using the full capacity of transformers and the possibility of overloading them, to which much attention has recently been devoted.

Putting a transformer with solidified oil under load may cause overheating of the oil near the windings, since heat removal, owing to the absence of convection, will be small in the first moment after switching on. Thus, the insulation will be in the same conditions as during transformer overload. On the other hand, at a low ambient-air temperature there will be good heat removal from the transformer tank and, naturally, the question arises of increasing the permissible load for the given apparatus. As for the question of overheating of the insulation when heat removal is impeded by thickened oil, the few data in the literature \(^{23}\) are contradictory, and the corresponding investigations were carried out on models under artificial conditions. On the second question posed, concerning overload, there are instructions in OST 2524, according to which, for regions with an average annual temperature below 4°C, transformer overload of 1% is permitted for each degree, starting from +35°C, but not more than 140% in all. A load of 140%, according to OST, will be reached at an ambient-air temperature of minus 5°C.

What is the permissible long-duration and short-time overload of a transformer under the conditions of a Siberian winter, when for a prolonged time the temperature remains below minus 30°? The existing OST and the draft of a new OST \(^{24}\) do not give exhaustive answers to these questions.

By agreement with a number of Siberian organizations (Kemerovo ATZ, Kuzbasset, and others), the high-voltage laboratory of SFTI carried out an experimental investigation of the thermal regime of an oil transformer \(^{25}\); below we briefly present the principal results of these investigations.

The measurements were made by research associates of the laboratory, E. K. Zavadovskaya and A. K. Krasin, with a transformer of 100 kva capacity.

of type TM 100/6 with 1000/400-volt windings. Under conditions of artificial and natural cooling the following observations were made: 1) heating of the transformer when connected in short-circuit mode; 2) heating of the transformer connected in normal operation; 3) heating of the transformer connected with a 160% load; 4) heating of the transformer connected under a 160% load with solidified oil; 5) thermal conditions during prolonged operation of the loaded transformer, connected under load at a low ambient-air temperature; 6) investigation of the thermal conditions of transformers on models.

The existing formulas for calculating the thermal conditions of a transformer^26) could not be used in our case, since we were interested in the true temperature at a number of points of the transformer, the heat-removal conditions from which are very complex and could not be taken into account with sufficient accuracy. It was therefore necessary to resort to an experimental investigation of the thermal conditions of the transformer under various conditions.

Observation of the temperature in the transformer was carried out by means of iron-constantan thermocouples connected by the differential method. The following points of the transformer were under observation:

1. The transformer rib.
2. The oil at the height of the middle of the coil, at a distance of 20 mm from the HV winding.
3. The gaps between the HV and LV windings.
4. The upper turn of the LV winding.
5. The lower turn of the LV winding.
6. The upper part of the core.
7. The oil at the height of the middle of the coil, at a distance of 10 mm from the HV winding.
8. The oil at the height of the middle of the coil, at a distance of 10 mm from the HV winding.
9. The surface of the HV winding at the height of the middle of the coil.
   $t_1$ — the air temperature at the height of the upper edge of the rib at a distance of 50 cm from the transformer.
   $t_2$ — the air temperature at the height of the lower edge of the rib at a distance of 50 cm from the transformer.

The arrangement of the thermocouples is shown in Fig. 11.

It is necessary to point out that items 1, 2, 3, and 4 of the measurements indicated above were carried out with the transformer placed in a special chamber made of boiler iron, the walls of which had good thermal insulation from the surrounding air. In this chamber (cryostat) the transformer was subjected to artificial cooling. Observation under item 5 was carried out on a transformer under conditions of natural cooling.

1. Heating of the transformer when connected in short-circuit mode. Drying of transformer TKZ continued for 14 hours; temperature observation was conducted for 9 hours. The highest temperature measured in the transformer did not rise above 47°C; the upper turn of the low-voltage winding heated up to this temperature. After six and a half hours the upper layer of oil reached the same maximum temperature. Then the temperature became established and subsequently did not increase. Thus, during drying of transformer TKZ no dangerous overheating of the insulation was observed.

2. Heating of the transformer when it was connected in normal operation. After switching on, for 50 min. the turn of the low-voltage winding had the highest temperature. Then the core reached the same temperature and for three hours had the highest temperature. By the end of the observation the highest temperature was on the surf—

surface of the high-voltage winding and was \(65^\circ\mathrm{C}\). The observation lasted 6 hours; during this time the temperature in the transformer became steady.

1. Observation of the heating of the transformer when it is switched on with a load of 160%.

A transformer load of 160% was chosen for purposes of testing the transformer under operating conditions more severe than those permitted according to the OST. At first these observations were carried out when the transformer was switched on at normal temperature. As in the preceding experiments, the most heated point proved to be a point on the surface of the high-voltage winding. However, it must be noted that, according to the data of P. M. Tikhomirov\(^{26}\), who observed the heating of the winding in oil, the most heated place in the winding is the region located inside the winding at a distance of \(2/3\) from the outer surface of the winding. This region has a temperature \(25^\circ\) higher than the temperature of the outer surface of the winding. Taking this circumstance into account, it may be considered that 1 hour 55 minutes after the transformer was switched on under overload, the limiting temperature had been reached, since at that moment the temperature of the surface of the high-voltage winding was \(70^\circ\mathrm{C}\); consequently, inside the winding the temperature was \(95^\circ\mathrm{C}\). The temperature in the cryostat (at mid-height) had by this time risen to \(34^\circ\mathrm{C}\). Then the transformer was cooled on average to a temperature of minus \(50^\circ\mathrm{C}\) and, with the oil solidified, was switched on under a load of 160%.

LOCATION OF THE THERMOCOUPLES

1—in the transformer rib; 2—at a distance of 20 mm from the HV winding; 3—between the HV and LV windings; 4—upper turn of the LV winding; 5—lower turn of the LV winding; 6—transformer core; 7—at a distance of 10 mm from the HV winding; 8—at a distance of 5 mm from the HV winding; 9—surface of the HV winding.

Fig. 11.

After the load was switched on, rapid heating of the oil began. The temperature rose especially rapidly in the ducts between the insulating cylinder and the winding. The heating of the transformer when it was switched on with a load of 160% can be seen in Figs. 12 and 13. Twenty minutes after switching on, the highest temperature is reached by the upper turn of the low-voltage winding, which subsequently remains the most heated place in the transformer. As can be seen from Figs. 12 and 13, the greatest heating when starting the transformer under these conditions was reached 40 minutes after it was switched on; the maximum temperature was \(47^\circ\mathrm{C}\). Then, after a certain decrease, the temperature becomes steady and remains constant. This established regime is disturbed when the air in the cryostat is heated to a temperature close to \(0^\circ\mathrm{C}\). After this, a general rise in temperature begins, analogous to that which we had in the preceding experiment when the transformer was switched on under a load of 160% at normal temperature.

Fig. 12.

Fig. 13.

From the data of the experiments described above, it may be concluded that, when a transformer with solidified oil is switched on under overload, there is no danger of dangerous overheating of the insulation; therefore oil-filled transformers may be switched on under load from cold standby, without preliminary measures for heating them.

Since the thickening of the oil in a transformer at low temperature does not create a danger of overheating of the insulation, and the heat evolved in the winding rapidly warms the oil, insulating oils of ordinary grades, rather than especially cold-resistant ones, may be used for filling transformers intended for operation in regions with low ambient temperatures. Such are the results of certain experiments with a transformer whose tank is provided with a ribbed surface.

As for transformers with other types of coolers (tubular, coil-type), it should be assumed that their warming-up will proceed with greater difficulty, and for them experimental studies analogous to those described should also be carried out.

From consideration of Figs. 12 and 13 it is evident that a transformer with a 160% overload can operate for a considerable time. However, under these experimental conditions the heat removal from the tank was considerably less than under conditions of natural cooling at low ambient-air temperature.

To resolve the question of the thermal regime of a transformer operating with overload under conditions of natural cooling at low ambient-air temperature, corresponding measurements were made with the transformer described above. The transformer was placed in a spacious chamber and was freely washed by air. During the experiment the air temperature varied from minus 35°C to minus 38°C. At first the transformer was switched on at the greatest load permitted by the OST, equal to 140%. When switched on under load, the internal parts of the transformer had a temperature from minus 4° to minus 12°C. The heating of the transformer in this case is shown in Figs. 14 and 15. The arrangement of the thermocouples was the same as in the preceding case.

As is evident from Figs. 14 and 15, the temperature in the transformer became established 9 hours after switching on, without reaching the limiting value; therefore the load was increased to 155%. Four hours after the load was increased to 155%, the temperature stabilized at the level of 70°C. After 7 hours of operation at a load of 155%, the load was increased to 177%. After 2 hours of operation at this load, the highest temperature in the transformer reached 105°C, after which the load was removed.

The experiment described shows that, at low ambient-air temperature, prolonged operation of a transformer with substantial overload is possible. In particular, at an ambient-air temperature equal to minus 35°C, for a TM 100/6 prolonged operation is possible not only with an overload of 40%, as permitted by OST 2524, but also with an overload of 150–155% of nominal. The winding temperature at a load of 155% was 70°C; allowing for P. M. Tikhomirov’s correction, the highest temperature in this case may be taken as 95°C. According to calculations by P. M. Shnitser[^27], at such a temperature the transformer can operate for 16 years, i.e., a normal service life. These results make it possible to raise the question of prolonged operation of transformers in the winter months with a considerable normal overload.

The possibility of increasing the utilized capacity of transformers by means of intensified heat removal from the tank is also indicated in articles by engineer V. V. Rubtsov and engineer P. D. Aizenberg[^27]. They obtained an increase in transformer loading above the rated power by 20–40% through intensified heat removal when its tank was sprayed with water. However, engineer P. A. Aizenberg draws attention to a significant increase

Thermal regime of an overloaded transformer operating at an air temperature of \(-35^\circ\text{C}\)

In the graph:

• \(140\%\)
• \(155\%\)
• \(177\%\)
• Time, in hours
• Temperature, \(^\circ\text{C}\)
• \(7\) and \(8\) are located on \(1\) and \(2\)

Legend:

• \(1\) — transformer rib
• \(2\) — oil at the height of the middle of the coil, at a distance of 20 mm from the HV winding
• \(6\) — upper part of the core
• \(7\) — oil at the height of the middle of the coil, at a distance of 10 mm from the HV winding
• \(8\) — oil at the height of the middle of the coil, at a distance of 5 mm from the HV winding
• \(t_1\) — air temperature at the height of the upper edge of the rib, at a distance of 50 cm from the transformer
• \(t_2\) — air temperature at the height of the lower edge of the rib, at a distance of 50 cm from the transformer
• ↓ — application of load

Fig. 14.

Thermal regime of an overloaded transformer
operating at an air temperature of \(-35^\circ\mathrm{C}\).

Graph labels:

• \(140\%\)
• \(155\%\)
• \(177\%\)
• Time in hours
• Temperature, °C

Legend:

③ — gap between the HV and LV windings.
④ — upper turn of the LV winding.
⑤ — lower turn of the LV winding.
⑨ — surface of the HV winding at the height of the middle coil.
⑩ — control thermometer in the transformer.
\(t_1\) — air temperature at the height of the upper edge of the radiator, at a distance of 50 cm from the transformer.
\(t_2\) — air temperature at the height of the lower edge of the radiator, at a distance of 50 cm from the transformer.
↓ — load steps.

Fig. 15.

losses in the copper when the transformer is overloaded, since they are proportional to the square of the load. Therefore the additional transformer capacity obtained under overload will give greater losses than in the case of installing additional transformer capacity; taking these circumstances into account, overloading transformers when they operate in the cold cannot be recommended without consideration of the economic factor. In the absence of the necessary installed transformer capacity at temperatures below minus 30°, prolonged operation of transformers of the above-investigated type with an overload of 50% may be permitted.

OPERATION OF OIL CIRCUIT BREAKERS UNDER LOW-TEMPERATURE CONDITIONS

Oil circuit breakers of outdoor substations, and sometimes also of indoor switchgear installations, under the conditions of a Siberian winter are subjected to the action of low temperatures; for example, in the winter of 1938/39 in Kemerovo, when the air temperature fell to minus 57°C, the temperature in indoor switchgear installations at that time fell to minus 36°C. The data presented in § 4 show how strongly oil thickens at such low temperatures; therefore the question naturally arises as to how the movement of the rod of an oil circuit breaker will take place in thickened oil, and how greatly the speed of this movement will be reduced. The available data in the literature\(^{28}\) show that at a temperature of minus 30°C the reduction in the speed of movement of the crosshead reaches 25%. These measurements were made on apparatuses of German substations, filled with oils having comparatively good low-temperature properties. Similar data were obtained in the Mosenergo system\(^{29}\) down to temperatures of minus 20°C, but, as the authors themselves indicate, from these data it is difficult to draw definite conclusions about the influence of temperature on the time of movement of the rod of an oil circuit breaker.

Fig. 16.

A. K. Krasin\(^{30}\) in our laboratory measured the travel time of the BM—14 rod during disconnection as a function of temperature down to temperatures of minus 44°C. For the experiments a BM—14 rated at 600 amperes with GP—125 was used. The apparatus was installed in the open air. Its drive was installed in the laboratory room.

To determine the travel time of the rod, a loop oscillograph was used, and the position of the rod during motion was record—

...at five different points by means of a contact system mounted on the drive shaft.

All measurements were carried out with the circuit breaker running idle, with 110 volts DC applied to the drive terminals. The temperature was monitored by a thermometer installed near the circuit breaker.

Figure 16 presents data from one of the measurement series. Along the abscissa axis are plotted the temperature values in °C, and along the ordinate axis—the time of the full travel of the VM rod in periods of alternating current. The light circles correspond to measurements made during cooling of the circuit breaker, and the dark ones to measurements during heating.

From Fig. 16 it is evident that down to temperatures of −30°C the increase in the rod travel time does not exceed 50% of the initial value (at a temperature of 0°C). At temperatures below minus 35–37°C, a strong increase in the rod travel time begins. At a temperature of minus 42°C, a case was even recorded in which the rod motion came to a complete stop.

Similar measurements of the travel time were carried out three times with two different grades of oil. The results were repeated.

Simultaneously with the measurement of the rod travel time, the viscosity of the oil was measured, using a modified Stokes method. Figure 17 gives the resulting dependence of oil viscosity on temperature. From Figs. 16 and 17 it is evident that the temperature variation of the rod travel time coincides with the temperature variation of the oil viscosity up to temperatures of minus 37°C. Below this temperature begins the region of unstable operation of the oil circuit breaker. It turned out that in the region of unstable operation the motion of the VM rod is substantially affected by the load from the drive after the tripping coil is switched on and the drive core is released.

Fig. 17.

The obtained data make it possible to conclude that, in the region of stable operation of the VM, the time of the full travel of the rod is determined by the viscosity of the oil and, for the oils investigated, can be calculated from the following equation

\[ \tau = 0.675 e^{0.84t} + 10, \tag{8} \]

where \(\tau\) is the time of the full travel of the rod in periods of alternating current, \(t\) is the oil temperature in °C. As for the region of unstable operation of the VM, no analytical dependence of the time \(\tau\) on the temperature \(t\) is observed.

It seems to us that the results obtained, which indicate an unambiguous relationship between the quantities \(\tau\) and \(t\), can be extended to circuit breakers with a large volume of oil; for them, from the curve of the temperature variation of the vis-

…of the oil viscosity, to calculate the value of the time \(\tau\) for various temperatures down to the temperature corresponding to the bend of the curve of the temperature dependence of viscosity.

In the region of a sharp increase in the temperature dependence of the oil viscosity, the oil circuit breaker cannot operate. For normal operation of the VM in this temperature region, heating of the oil is necessary.

The measurements carried out do not claim to be complete in the sense of describing the operation of the VM under low-temperature conditions. On the contrary, they emphasize the need to carry out research work in this field as soon as possible. At present it is necessary to clarify the following questions connected with the operation of the VM at low temperatures:

1) measurement of the time of motion of the rod as a function of temperature when current is passed through the contacts of the VM;

2) obtaining the dependence of the speeds and times of switching off and switching on on temperature at various voltages at the drive terminals;

3) carrying out a complete study of drives under low-temperature conditions, determining the minimum closing and tripping voltages for the entire range of temperatures of practical interest;

4) investigating the behavior at low temperatures of the insulation of all elements of the control circuit and of the VM itself, as well as of the various kinds of gaskets and seals used in these apparatuses;

5) investigation of arc extinction in solidified oil;

6) the obtained unambiguous dependence of the time of motion of the VM rod on oil viscosity makes it possible to raise the question of reproducing the operating conditions of the VM in the low-temperature region at high temperatures, if the breaker tanks are filled with viscous compounds.

For oil circuit breakers intended for operation in regions with low temperature, it is necessary to use special cold-resistant oils. When ordinary oils are used, knowing the temperature dependence of their viscosity, heating should be switched on at the temperature corresponding to the beginning of a strong increase in viscosity. The corresponding heating standards must be specified.

CITED LITERATURE

1. N. T. Biryukov. Elektrichestvo 14, 3, 1937. A. A. Vorob’ev and N. A. Prikhod’ko. Zhurn. Tekhn. Fiziki 9, 1969, 1939.

2. Resolutions of the enlarged conference on electrical insulating materials. May 1940.

3. Proceedings of the conference on combating energy losses at thermal power stations and accidents in the networks of Kuzbass. Tomsk. 1940. TII (in press).

4. Shteger. Technology of Electrical Insulating Materials. ONTI, Energoizdat. 1934.

5. A. A. Leont’ev. Uspekhi Fiz. Nauk 23, issue 2, 131, 1940.

6. V. D. Kuznetsov. Proceedings of SFTI. 4, 1, 1936.

7. T. K. Tomsen. Lubrication Practice. Moscow. 1932. Ivanov and Guttsait. Report on the work of VTI. Gromenitskii and Luchinskii. Grozny Oil Economy, 1922.

8. A. A. Vorob’ev and N. A. Prikhod’ko. Elektrichestvo 8, 66, 1940.

9. Ivanov and Guttsait. Loc. cit.

10. F. F. Vol’kenshtein. Breakdown of Liquid Dielectrics. ONTI, GTTI, 1934.

11. Eng. I. A. Glozman. Assembly of Oil-Filled Bushings of 110 and 220 kV. ONTI, NKTP USSR, 1936, pp. 1–19.

12. E. F. Komarkov. Electrical Materials Science. ONTI NKT USSR, 1935, pp. 78–96.

13. F. F. Vol’kenshtein. Loc. cit.

14. N. A. Prikhod’ko. Report on the work. Tomsk. SFTI. 1939.
    The same. Report on the work. Tomsk. SFTI. 1937.

15. M. Mikhailov and G. Soya. Zhurn. Tekhn. Fiz. 6, 786, 1936.

16. V. A. Vorob’ev. Elektrichestvo 9, 53, 1940.

17. V. F. Bozhko. Proceedings of SFTI (in press).

18. N. F. Ryzhkova. Report on the work. Tomsk Industrial Institute. 1940. Elektrichestvo (in press).

19. V. A. Baev. Insulating Materials. Gosenergoizdat. 1940, p. 97.

1. A. A. Vorob’ev and A. N. Snegirev. Change in the electrophysical properties of compound E-3 as a function of temperature. SFTI. Work report. 1940.

2. A. K. Krasin and V. F. Bozhko. Report on testing a casting compound. SFTI. 1937.

3. Resolutions of the expanded conference on electrical-insulation materials, May 7–10, 1940. Moscow. 1940. P. 12, item 11.

4. E. K. Zavadovskaya. Electricity. 8, 66, 1939. Proceedings of SFTI (in press).

5. L. M. Shnitser. Electricity. 1, 8, 1938; 12, 10, 1938; 12, 9, 1938.

6. E. K. Zavadovskaya, A. K. Krasin, and Lab. O. G. Nonchukov. Report on work on the thermal regime of transformers, parts I, II, and III. SFTI. 1938–40. A. K. Krasin. Electricity. 7, 60, 1940. Proceedings of SFTI (in press).

7. P. M. Tikhomirov. Electricity. 12, 24, 1938.

8. Eng. V. V. Rubtsov. Bulletin of Electrom. 4, 29, 1939. Eng. V. A. Aizenberg. Bulletin of Electrom. 10–11, 43, 1939.

9. A. M. Zalesskii. Oil Circuit Breakers. Kubuch. Leningrad, 1932.

10. V. P. Bur’yanov. Electricity. 4, 21, 1937.

11. A. K. Krasin. Electricity. 12, 52, 1939.

Editorial board: Professor M. K. Korovin (deputy responsible editor), Prof. I. N. Butakov, Prof. F. N. Shakhov, Prof. G. E. Bakanov, Prof. A. S. Betekhtin, Prof. A. N. Dobrovidov, Doc. A. M. Rozenberg, Prof. A. A. Vorob’ev, Prof. I. V. Gebler, Doc. M. E. Efendi, Doc. A. A. Belitskii.

Responsible editor—Prof. K. N. Shmargunov.

MN15271

Submitted for production 29/XII-1940.
Signed for printing 21/V-1941.
Price 12 rubles.

Typ. chars. in 1 printed sheet 62016
Volume 12 printed sheets, 18.5 author’s sheets.
Order No. 2496-1940.
Print run 500 copies.

Tomsk, printing house of the publishing house “Krasnoe Znamya”—Sovetskaya St., No. 47.