[Trade Journal]
Publication: Electric Journal
Pittsburgh, PA, United States
vol. 12, no. 6, p. 282-289, col. 1-2
A Comparison of Different Methods of Testing
DR. A. CHERNYSHOFF, Professor of Electrical Engineering, Imperial Polytechnic Institute, Petrograd and C. A. BUTMAN, Research Engineering Dept., Westinghouse Electric & Mfg. Company
UNTIL quite recently electrical porcelain has been tested at 60 or 25 cycles with voltages below the flashover point of the insulator. However, practice has shown that insulators which had passed such a test quite often failed in service; in particular, a heavy thunderstorm would break many insulators. Hence, to duplicate as nearly as possible the effects of a lightning stroke, Messrs. Imlay and Thomas were led to try a high-frequency impulse test.(1) Their work was followed by that of Mr. E. E. F. Creighton,(2) who called attention to the availability of the Tesla coil for high-frequency test of insulators. The present paper has to do with an investigation of the relative merits of the various methods of testing porcelain, and the development of a new method to resemble more closely the effects of a lightning stroke.
IN ORDER to make a comparison between all present methods of testing insulators, a series of tests was run following a method outlined by Messrs. Imlay and Thomas.(1) The arrangement of the apparatus for this test is shown in Fig. 1, and the diagram of connections in Fig. 2. The contact maker consisted of a lever on the end of which was a strip of copper which made contact for an instant with a fixed similar strip, when the lever was allowed to fall, thus completing the circuit; in this way the contact was always of the same duration. A choke coil limited the current when a spark passed through the gap and a resistance R, regulated the amount of energy in the primary of the 300,000 volt 10 k.v.a. transformer shown at the extreme left in Fig. 1. A sheet of asbestos R, was used to limit the energy in the secondary of the transformer. The condenser C consisted of a suspended iron plate ten feet eight inches long and five feet wide hung parallel to the wall as shown. The wall was covered with a grounded wire netting, and in most of the experiments the plate was about four feet from the wall. The diameter of the spark-gap spheres was almost eight inches. The voltage across the gap was measured by short - circuiting the asbestos resistance and then adjusting the spark-gap so that a spark would just occur at the required voltage as indicated by a voltmeter across the primary of the transformer. To avoid corona loss, the high tension connections were made with 15/16-inch brass tubing a n d the edges of the condenser plate were curved away from the wall. The insulator on which most of the tests were made is shown in Fig. 3; it was supported on a grounded iron pin.
The effect of the condenser shunted across the spark-gap is to superimpose a high frequency discharge on the ordinary 60-cycle discharge. In running this test ten insulators were chosen at random and each was subjected to 50 impulses(3) at the same voltage. The voltage was then raised by steps until the insulators failed, the test being made in air. The results of the test by the Imlay and Thomas method are shown in Table I.
A continuous series of impulses was obtained by keeping the primary circuit closed. The condition of the discharge under these circumstances is shown in Figs. 4 and 5. An oscillogram, Fig. 6, of the voltage curve of the primary shows no flashover with 150,000 volts maximum, although a flashover is obtained at this voltage with 60 cycles. The results of the test by this method are shown in Table II . The voltage was obtained from the ratio of the transformer and from the oscillogram, which was taken just at the flashover voltage, 105,000 volts effective.
An ordinary 60-cycle discharge was obtained with these same connections by merely short-circuiting the spark-gap. In Fig. 7, which illustrates the character of the discharge obtained in this way, the discharge takes the shortest air path to ground. The test at 60 cycles was first made in air, but as few insulators broke down and difficulty was experienced due to heating, the insulator was immersed in oil and the test repeated. The results of the tests by this method are shown in Table III. The insulators used in this test were chosen at random.
THE DIRECT-CURRENT SINGLE-IMPACT METHOD
With the short impulse test described above, a train of waves is set up which is of indefinite length, and there is no way of telling how many impacts are given the insulator. Furthermore, each successive train of waves may be different, as contact may be made at different values of voltage. Hence, to make the investigation more exact and to duplicate more closely the conditions during a lightning discharge, a method was developed whereby only one impact is given the insulator. This was accomplished by using direct current on the transformer. Inasmuch as the impact at the making of the circuit is definite, while owing to the arc at the break it is impossible to have a steep wave front, it was necessary that the impact should only occur when the contact was made. This was accomplished by slowly cutting down the current after the circuit was closed by introducing resistance so that no spark passed between the terminals of the series gap when the current was broken. Another condition which must be fulfilled is that the current through the transformer must be alternately reversed in order that the core will remain in the same magnetic state. In addition to the foregoing requirements it was thought desirable to be able to control the nature of the impact so that it should always be positive or negative, as might be thought necessary. These conditions were all fulfilled by the rotating device shown in Fig. 8. A positive or negative impact was obtained by placing a rectifier across the primary of the transformer. Hence, when the current flowed in one direction the primary was practically short-circuited through a series resistance and there was not sufficient flux to make an impact, while with the current in the other direction the flux was as usual and an impact occurred. By reversing the connections of the rectifier the impact could be made positive or negative.
The operation of the rotating contact maker is as follows: Suppose B to be positive. When the black end of the arm A is at H there is an open circuit; as the arm is moved to B contact is made and the circuit is closed; as the arm is moved to G the resistances are introduced until the circuit is broken beyond G. When the black end of the arm is at B, the current flows from B through the resistances r1, r2, r3, etc., then through the arm itself to the center, then through an ammeter and switch to the primary of the transformer, thence through the other side of the switch through a variable resistance to the inner ring of the rotator. The brush contact E touches the inner ring which is connected to the outer brush F, which makes contact with C, completing the circuit. When the black end of the arm is at C the current is reversed through the transformer, the same resistances, r1, r„, etc., being introduced into the circuit before it is broken. An electromagnetic counting device is placed across the resistances r1 and r4 in series with a lamp so that every time the current is made it operates. When the rectifier is working the number of impacts will be one-half the number shown by the indicator. A motor drives the arm of the contact maker at such a rate that when the rectifier is working there is about one contact per second. The rectifier is constantly excited by applying about five amperes direct current to maintain the arc. Resistance R2 regulates the amount of current flowing through the branch circuit, R, the amount of direct current supplied the rectifier, and R1 the amount of current flowing through the contact maker. The resistance R4 was short-circuited during this test.
Whether the impulse was positive or negative was determined initially as follows: Sphere 2 of the series gap was insulated from ground, a spark was allowed to pass, thus charging the sphere. The nature of the charge on the sphere was examined by bringing up a cork ball, suspended by a long cotton thread and charged positively by bringing it in contact with glass that had been rubbed with silk. If the charge on the sphere was positive, it repelled the cork, if it was negative it attracted it. With a given connection of the rectifier the charge was always positive or negative, as the case might be.
Tests were made by the direct-current impact method to see if there was any difference between the effect of a positive or negative impact on the insulator. The character of discharge obtained under these conditions is shown in Figs. 9, 10 and II and in the oscillograms, Figs. 12 to 21. The test was also designed to investigate the fatigue of insulators with positive and negative impulses, in particular to see whether it took a smaller number of impacts to break down the insulator at higher voltages and whether there was a difference between the number of positive and negative impacts required at those voltages, each insulator being subjected to 2,000 impacts unless it broke down with a less number. The results of these tests on the type of insulator shown in Fig. 3 are given in Table IV, and are summarized in the curves given in Fig. 22. Each point on these curves represents the average of from four to ten insulators. No great claim is made for accuracy but the curves may be taken as representing typical conditions.
As some of the insulators are capable of with-standing more than 2 000 impacts the averages given should be considered a s minimum values. The insulators were all taken from stock and had not previously been tested. The voltages were obtained from the oscillograms taken at the same time.
The voltage scale was obtained from the direct-current line shown in Fig. 20, and by reading the voltage with a voltmeter when a 6o cycle wave was taken. The oscillograms show the voltage wave of the primary of the transformer.
Further tests were made to determine the relative ease of flashover with positive and negative charges under different conditions. The following results were obtained with a six-centimeter sphere gap (90,000 volts) on the insulator shown in Fig. 3:
Case I—With the head subjected to positive impacts and the pin grounded, one hundred impacts resulted in ninety-six flashovers.
Case II—With no change in any part of the apparatus, the connections of the insulator were interchanged so that the head was grounded and the pin was positive. One hundred impacts resulted in nine flashovers.
Case III—The connections of the insulator were allowed to remain the same, but the terminals of the mercury rectifier were interchanged, thus making the pin negative and the head remaining grounded. Under these circumstances one hundred impacts resulted in fifty-nine flashovers.
Case IV—With the apparatus the same as in Case III, the connections to the insulator were reversed, making the head negative and the pin making connection with the ground. This condition only produced two flashovers in one hundred impacts. The foregoing results were repeated at will.
HIGH-FREQUENCY TESTS
In order to compare the effect of a high-frequency superimposed on a 6o-cycle wave, as obtained in the Imlay and Thomas or continuous impulse test, with high-frequency alone, a series of tests was made on the same type of insulator, with high-frequency obtained by means of a Tesla coil with the connections as shown in Fig. 23. The coil was designed to stand 250 000 volts between the primary and secondary under oil. The connections between the Tesla coil and the insulator were made by one-inch brass tube and were so arranged that the insulator could be exactly in the same relative position as in all the previous methods of test. The char acter of discharge obtained with high-frequency is shown in Figs. 24 and 25. Tests were made with the Tesla coil on ten new insulators chosen at random; they were subjected for one minute to various voltages as shown in Table V. until a voltage was found that would cause a breakdown at that time. Quite a number of new insulators were tested in addition to those recorded in Table V which confirmed the results given. No insulator was found which would stand 139 500 volts from the Tesla coil for one minute; the voltage being measured by means of the spark-gap in parallel with the insulator. Below flashover of the insulator, the spark-gap alone flashed over. Above flashover of the insulator care had to be exercised that the parallel gap sparked, as it was found that the insulator effected the spark-gap length in a marked manner. The high-frequency voltage was regulated by the distance between the zinc plates.
ADDITIONAL TESTS ON PORCELAIN TEST PIECES
Tests were made to determine whether the breakdown of electrical porcelain is due to the electrical stress or to the heating caused by the leakage current. Special glazed and unglazed cylindrical porcelain cups were used which had been designed especially for this length in a marked manner. The high-frequency voltage was regulated by the distance between the zinc plates.
ADDITIONAL TESTS ON PORCELAIN TEST PIECES
Tests were made to determine whether the breakdown of electrical porcelain is due to the electrical stress or to the heating caused by the leakage current. Special glazed and unglazed cylindrical porcelain cups were used which had been designed especially for this contact. The cups were tested for five minutes or until breakdown. Those which did not break down in five minutes were tested at higher voltages to see what voltage would break them down ; this was found in most cases to be 56,000 volts. As some time was used in adjusting for this voltage, the instantaneous voltage may be taken as about 58,000 volts. The results of these tests are shown in Table VI.
To determine whether there was any difference in the breakdown strength of glazed and unglazed porcelains similar tests were run with glazed cups, under oil, at 60 cycles, the voltage being increased until a breakdown occurred. Comparative results are shown in Table VII. According to these tests the glazed porcelain was 14 percent stronger than the unglazed.
An additional test w a s made by bringing the voltage up to a given value at 6o cycles and then quickly dropping it 700 or 800 volts, which was maintained for two minutes; at the close of this time the same process was repeated again, and so on until a voltage was reached which caused a breakdown. At 38,300 volts the cup broke down in 40 seconds, although it had momentarily been raised to 39 150 volts at the start.
A test was made on two glazed cups with the continuous impulse method, with the cup filled with oil, but with air outside. The flashover voltage was 43,000 volts. A continuous test was made on the first cup for a half hour at 30,360 volts with the cup in air but with oil inside. The temperature of the oil rose four and one-half degrees centigrade, but no breakdown took place. The cup was then placed in oil and tested at 39,900 volts, at which voltage it broke down in one and one-half minutes. A second sample, tested at 41,100 volts as above, broke down in three seconds.
With the Imlay and Thomas impulse method tests were made on samples of glazed porcelain cups under oil. The breakdown voltage and the number of impulses necessary to cause a breakdown at that voltage are given in Table VIII. Most of the insulators had been tested at several other voltages before that which caused the breakdown was applied.
RELATION OF THE THEORY OF IONIZATION TO THE TEST
RESULTS
The time of ionization plays a very important part in the breakdown strength of insulators. It had previously been thought that it was possible to design an insulator so that it would flash over rather than puncture. However, the experiments show that it is possible to have a breakdown below flashover.
This result was probably due to poor character of porcelain, although the samples tested were of average strength. A very important result is that it is possible to raise the voltage on the insulator above the normal 60-cycle flashover value if high frequency is used. The results indicate that with sixty cycles, no matter how high the applied voltage is, the voltage on the insulator is never much above the ordinary flashover value, as the insulator is practically short-circuited, hence few breakdowns occur. This is shown clearly in Fig. 7, where the discharge takes the shortest air path. However, the character of the discharge is entirely different at high-frequency as it clings to the surface and the voltage on the insulator can be raised a sufficient amount to cause a breakdown. The reason is that with high-frequency there is not sufficient time to ionize the air, so that the discharge flows over the surface, as shown in Figs. 24 and 25. The higher the voltage the less is the time for ionization and the greater is the stress on the insulator.
Figs. 9 and 10 indicate clearly that the negative potential ionizes the air better when on the pin than when on the insulator head. As it is hard to ionize the air with the negative impulse on the insulator head and thus relieve the pressure, it is possible with negative impulses to raise the voltage above the flashover value obtained for positive discharges. Hence, it is seen why at the same voltage, breakdown occurs more quickly with a negative impulse than with a positive one. The explanation of this is similar to that for the discharge between a point and a plane; if the point is negative a spark will pass at a lower voltage than if it is positive.
The Imlay and Thomas impulse consists in part of sixty cycles and in part of high-frequency discharges. This is shown in Fig. 4, where part of the discharge follows the shortest air path and part follows the surface. At voltages near flashovers the discharge is all on the surface as shown in Fig. 5. Figs. II and I show the alternate positive and negative discharge. To the eye, the discharge appears to go over the surface and then outside of it, following the shortest air path, or vice versa. The positive charge making the discharge on the outside and the negative the one that follows the surface.
THE RELATION OF THE RESULTS TO THE DESIGN OF INSULATORS
In order to design an insulator properly it is necessary to take into account the frequency at which it is to be used. Owing to the time for ionisation being different at different frequencies, the stress can be distributed differently. For instance, with the type of insulator shown in Fig. 26 at 60 cycles, the discharge will go as shown and the stress will be between the top of the insulator A and the top of the pin B. However, at high frequency, as the discharge follows the surface, great stress will occur between B and C. Hence it was found that with the Tesla coil and also with the Imlay and Thomas method, breakdown occurred between B and C. A very small insulator was tested with the Tesla coil a perhaps four times higher voltage than its flashover and a breakdown occurred in less than a half minute. Here, although the air path was very short, the frequency was so great that the stress was applied all the time without ionizing the air and relieving the strain. The weak place in the insulator shown in Fig. 3 was found to be between the cement and the pin. Of course, if the insulator is immersed in oil the distribution of stress is different, as it is in a different dielectric, also the time of ionisation is different. Hence, the breakdown at 6o cycles under oil for Fig. 3 was found to be from the base of the head to the pin. With the direct-current impact test it was observed that this insulator first broke through the lower petticoat at the cement to the pin and then broke down between the pin and the head, for both positive and negative impacts. Attention should be paid to the relation of the capacity of the upper petticoat to that of the lower, as the distribution of stress between them may be important.
PROPERTIES OF PORCELAIN
The tests made by the authors indicate that the breakdown of the porcelain tested was not due to heating, but solely to the electrical stress. In one case already cited the temperature of the oil in a porcelain cup rose only four degrees C. in a half hour, although the stress was not much below the breakdown value. In the direct-current impact test there is no opportunity for heating, as the voltage is applied only for an instant and then is removed for a relatively long time. About ten percent of the insulators tested seem to be considerably stronger than the average. The glazed specimens of porcelain tested required a higher breakdown voltage than the unglazed.
The character of the breakdown seems to depend on the frequency. With high-frequency a pinhole usually resulted which was often hard to locate. With 60 cycles under oil a disruptive discharge took place which many times left a large hole and scattered particles of the porcelain about.
In testing the type of insulator shown in Fig. 26 red streamers occurred in a permanent position beneath the glaze. The length of the streamers changed with the voltage. The cause of this phenomenon is not known but it may be due to gas in cracks covered by the brown glaze.
FATIGUE OF ELECTRIC PORCELAIN
The experiments show that under the conditions of the tests, the insulators were permanently injured by the testing. The curves in Fig. 27 show that a voltage which is only 4o percent of the instantaneous breakdown value caused the porcelain to be weakened, the amount of weakening depending on the time the voltage was applied. The form of the curve is the same as that for fullerboard.
The same conclusion as to the fatigue of porcelain is confirmed by the curve shown in Fig. 22, showing the results of tests made in air by the direct - current impact method. If there was no fatigue the insulator would stand an indefinite number of impacts at a given voltage if it would stand one. Hence the curve indicates that each impact injured the porcelain a certain amount, and that the injury is greater the higher the voltage. These experiments also confirm the opinion that breakdown is due to electrical stress and not to heating. A still further demonstration that breakdown is caused by fatigue is that in which the voltage across a porcelain cup was raised above that at which it finally broke down.
COMPARISON OF DIFFERENT METHODS OF TEST
1—The Sixty Cycle Test is fairly satisfactory below the flashover voltage. Still it is not searching enough, as quite a number of insulators were found which had been tested by this method which punctured through the lower part of the lower petticoat when tested at high frequency with a slightly greater voltage. It would seem of little value to test with 6o cycles under oil, as the stresses are applied differently than in air. A test in oil might be of advantage in a design test of insulators to determine the dielectric strength of the material.
2—The Continuous Impulse Test is better than the 60 cycles test as the impulse has a high frequency discharge in addition to the 60 cycles. This method could be easily set up in a factory by simply adding a large condenser and spark gap to the ordinary equipment. This method has the disadvantage that only a few insulators can be tested at one time, as the capacity of the condenser must be much larger than the capacity of the insulator tested. Fig. 4 shows that with this test near the flashover value the discharge clings to the surface, which means that the strain is on the insulator. Fig. 5 shows the form of discharge at a much higher voltage. Here it is shown that some of the dis-charge still clings to the surface, although the air has been ionized and part of the discharge is passing through the air. It has been proven that by this method the voltage on the insulator can be increased much above the ordinary flashover voltage measured at 60 cycles. Hence, this is a much better test than 6o cycles alone.
3—The Imlay and Thomas Impulse Test has the disadvantage that some mechanical device is necessary to make and break the primary current. A difference exists between the short impulse and continuous test in that the short impulse has to ionise the air anew each time. Hence, as strains are set up and puncture sometimes takes place before there is time to ionize the air, the short impulse is probably more severe. In addition to the foregoing, if there are any surges set up on opening or closing of the circuit these also will make the test more severe. The Imlay and Thomas test as at present set up has the disadvantage that the circuit of the primary of the transformer is made at different parts of the cycle of the alternating-current wave and with the core at different states of magnetization. Hence, the nature of the impulses is somewhat indefinite. The number of insulators which can be tested at one time is limited, as with all the other high frequency methods. With this method the number of impulses in a wave train is different at different times.
4—The Direct-Current Impact Method was devised to overcome the objections to the Imlay and Thomas method and to make the test more definite. The oscillograph curves show that the voltage is applied very abruptly, and hence reproduces the conditions of a lightning stroke. The charge can also be made positive or negative as occurs in nature in a thunderstorm. With this method the voltage can be applied so quickly that there will not be sufficient time to ionize the air of the spark gap. This occurred in the course of the experiments, and it was found necessary to introduce a choke coil into the primary of the transformer. In this way the steepness of the wave was reduced, and the time for ionization increased, causing the spark gap to work. By this method a voltage may be impressed on the insulator in excess of the 6o cycle flashover voltage, and by means of the oscillograph and counter it is possible to find out the number of impacts given the insulator.(4)
5—The Tesla Coil method has the advantage of portability and can be used for testing insulators on the poles. The severity of the test depends on the coupling of the Tesla transformers.
For factory use, the best single test is probably the direct-current impact method. If this method is to be used, ten impulses could be given at about five percent higher voltage than the flashover voltage at 60 cycles. Tt would seem to be of little advantage to test insulators below their flashover voltages. If the Tesla test is to he used the voltage should be at least five percent higher than the flashover voltage on 60 cycles. It is much better to test insulators for a brief time at a high voltage than for a longer time at a lower voltage. By using a high voltage test one is sure that the insulators will stand that voltage, also there is an economy of time.
CONCLUSIONS
These tests were made on a limited number of insulators and while the results that are of a laboratory nature are entitled to considerable confidence, it would require the testing of a much greater number of insulators to determine just what kind or method of test would be most suitable for commercial purposes. The tests indicate that the following conclusions are warranted:
I—By means of high frequency it is possible to raise the voltage on an insulator in air above the 60 cycle flashover voltage.
2—Inasmuch as the same breakdown voltage was obtained with the Tesla Coil, the Imlay and Thomas test and 60 cycles under oil it is concluded that the frequency of itself has little effect on the breakdown strength. The effect of high frequency is to raise the voltage which can be applied to the insulator in air. Hence, the breakdown is due to the voltage primarily and not the frequency.
3—The breakdown of porcelain is due to the electrostatic stress and not to heating.
4—The character of the breakdown at high frequency and 6o cycles was different.
5—It would seem of little use to apply a high frequency test of any kind to insulators below the 60 cycle flashover value.
6—Porcelain appears to be fatigued by the application of potentials above a certain unit stress.
(1) See high frequency test on line insulators by L. E. Imlay and Percy H. Thomas in the Transactions of the A. I. E. E. for 1912, p. 212.
(2) See "Electrical Porcelain," by E. E. F. Creighton, in the Proceedings of the A. I. E. E., May, 1915.
(3) The term "impulse" as used hereinafter in connection with the Imlay and Thomas test is defined as "the succession of impacts or impact trains produced during the brief interval in which the circuit is closed by the falling contact maker."
(4) For additional information concerning the different methods of testing electrical porcelain, see Mr. R. P. Jackson's discussion of Creighton's paper previously referred to.
