[Trade Journal]
Publication: American Institute Of Electrical Engineers
New York, NY, United States
p. 2143-2168, col. 1
COMPARATIVE TESTS ON HIGH-TENSION SUSPENSION INSULATORS
BY P. W. SOTHMAN
This paper presents an account of work done in connection with the selection of a suitable high-tension insulator for a transmission line operated at 110,000 volts. It is a report of an investigation giving a faithful account of the motives calling for the same, the method adopted and used to carry out the work, and the line of reasoning followed in classifying the results obtained. It is not intended to be a criticism of any individual design or of the valuable work which has been done by others in the same direction. The problem which had to be solved was well defined, requiring no more difficult task than to select from a number of insulators the one best suited for certain predetermined conditions. From the first to the last, the question was one concerned with engineering only, in which biased opinion or partiality was to be absolutely absent. How well this problem has been solved may be judged from the following account and perhaps more so by the tangible results obtained in the years following, during actual operation.
When it was found that the line losses of the proposed power transmission could not be kept at a reasonably low figure unless the system was operated at 110,000 volts between conductors, the question of insulation became at once of greatest importance. Unfortunately, at that time, very few reliable data were available with regard to the operating experience with potentials above 60,000 or 80,000 volts. Notwithstanding the lack of such practical experience, every manufacturer was ready to offer and guarantee an insulator for a transmission line operating at 110,000 volts. Before an attempt was made to draw up specifications for these insulators, a thorough canvass of the situation was made. The different insulator factories were visited to ascertain the manufacturing facilities of the firms and their ability to turn out a rather large order within a specified time. Tests on the proposed insulators were witnessed at the works of the manufacturers and all available information and data bearing on the subject were collected.
While visiting these factories, one could not help being most peculiarly impressed by the widely varying methods of testing employed by the manufacturers to demonstrate the merits of their insulators. This applies especially to the application of artificial rain and to the facilities afforded for observing the effect of the test. As a matter of fact, every manufacturer had his own way of applying rain, and of interpreting the effects observed. It can easily be understood why tests on one and the same type of insulator would show two entirely different results, depending upon where and by whom they were tested.
In view of the seemingly erratic behavior of the insulators during these manufacturers' tests, it was impossible to arrive at a definite conclusion. It became apparent that tests of this character should be performed under absolutely unvarying conditions in order to arrive at reliable figures, and arrangements were at once made to duplicate and elaborate these tests under conditions which could be controlled and changed at will to suit certain predetermined requirements.
The testing equipment of which use was made in the following tests, consisted of a large platform over which was placed a gas pipe, resting at each end upon 60,000-volt pin-type insulators. The insulators were tested one at a time, the small trolley from which they were suspended being moved opposite a mark made in the pipe midway between the supporting insulators, while all other insulators were crowded to one side and out of the way. The test on one insulator completed, it was moved to one side, and the next insulator placed in the proper position. This method proved to work out very well, especially in connection with the rain test described later.
The electrical apparatus consisted of two 50-kw. 2200/150,000-volt transformers in series, fed from a 25-kw. 220/2200-volt transformer.
The maximum voltage which could be safely obtained with this combination was slightly above 330,000 volts, with the two transformers in series, the neutral point being grounded, and 225,000 volts with one transformer alone and ungrounded. The voltage was controlled by means of a water rheostat in the low-tension circuit of the high-tension transformer. The readings were taken on an alternating-current voltmeter previously calibrated with spark gap in accordance with the Standardization Rules of the A. I. E. E. (1907). The voltmeter was connected across the low-tension side of a one-kw., 2200/110-volt transformer, the latter being connected across the low-tension side of the high-tension transformer.
All tests were performed at night in complete darkness. In order to have a permanent record for comparison, photographs were taken of each insulator during the several tests. The time on the clock dial appearing in the illustrations was used as a means of identification.
The tests were applied in the following order:
1. Dry test.
2. Wet test.
3. Parallel test, dry and wet.
4. Puncture test, under oil.
5. Mechanical test.
The Dry Test consisted of
a. Flash-over test on each section in order to exclude weak or punctured units.
b. Potential test on each complete insulator, also on smaller number of sections. Voltage was applied and raised by successive steps and photographic records were taken while the test was progressing.
Records were kept of the time on the clock dial, which was set for each new test, voltage applied, time of exposure, number of sections, and such other observations as were made during the test which could not be recorded on the photograph.
In this manner, each type of insulators submitted was subjected to the same series of tests under exactly like conditions.
The Wet Tests consisted of applying rain at 45 deg., the precipitation varying from 0.25 to 0.35 in. (6.35 to 8.39 mm.) and finally to 0.53 in. (13.46 mm.) of water per minute. Accordingly each insulator was subjected to three series of tests, in which voltage and precipitation were thevariable quantities. The execution of these wet tests was very similar to that of dry tests. Photographs were taken and records made of each test and observations were carefully noted.
For these rain tests, which were the most important of all, the following method was adopted. A number of nozzles of the type used for spraying trees were secured to the ends of pipes cut to suitable length, and these, in turn, were connected to the water mains by means of a rubber hose and arranged to slide along a vertical post. Two groups of nozzles were used, and by means of this arrangement, it was possible to adjust the angle of precipitation, and by moving the nozzles closer or further away from the insulators, to adjust the amount of rain supplied per minute. It was found more expedient to entirely open the valve in the mains, thus operating with full pressure of the standpipe, and to regulate the amount of water by adjusting the number of working nozzles, their heights and distance away from the insulator under test. The amount of precipitation was determined by means of a specially constructed rain gage, consisting of a funnel-shaped vessel with a cover, provided with an aperture five in. (12.7 cm.) in diameter. The edges of the opening were slightly raised to prevent the water from spilling and splashing over the top of the funnel when striking it at an angle of 45 deg.
The gage was held in the rain at the points where the several sections of the insulators would be located, and the quantity of water was measured with a graduated glass. As a rule the gage was operated for four minutes and the fall of water determined from the amount gathered during that time. By setting the nozzles and adjusting the spacing, the correct amount of precipitation could be obtained, and this setting was left undis-turbed during each series of tests. One insulator after the other was moved to the mark on the pipe and voltage applied, beginning low and increasing by successive steps.
It was found that although water flowed freely over all the sections, wetting an area of 7 ft. (2.13 m.) in diameter, on the platform, and thoroughly flooding the top of the sections, the inside of the insulators remained practically dry except for a few drops. For those reasons, it was necessary to apply the rain for some time before voltage was applied in order to arrive at reliable and unvarying results. The water was turned off when insulators were changed and turned on again when ready for test. A long series of check tests showed that the precipitation was constant within the error of observation after turning water on and off, provided the valve was always turned on wide open.
In most cases, the same underhung suspension insulators, strung horizontally, were subjected to a series of rain tests to ascertain their performance when used as strain insulators, but on account of this horizontal position, required a somewhat different method of supporting. The insulator was strung between two well-braced upright wooden posts. In order to prevent leakage to ground, a number of units were inserted between the posts and the insulator under test. The nozzles, twelve in number, were located directly above at a distance of 25 in. (63.5 cm.) from the center of the insulators, directing the spray of water, which could be raised from 0.22 to 0.5 in. (5.58 to 12.7 mm.) and even 0.75 in. (19 mm.) per minute at an angle of 45 deg. either toward the inside or the outside of the sections. The water was measured with the same rain gage used in previous tests, at eight different points within the space occupied by the insulator when in place, allowing the gage to remain thirty seconds in each of the eight positions.
Parallel Test. A most interesting series of tests with all insulators connected in parallel was made later, in order closely to follow their performance simultaneously at different voltages. The insulators, composed of a proportionately smaller number of sections, were supported from the pipe, equally spaced, their lower ends connected by a common bus. Voltage was applied and gradually raised as in previous tests. As soon as a voltage was reached at which one of the insulators would show signs of distress, a photographic record was taken of the whole set, after which the failing insulator was disconnected from the bus and the voltage increased until one of the remaining insulators would fail, and so on. A similar series of tests was performed with the insulators subjected to rain. Each insulator had its own set of nozzles and the flow of water was regulated to be the same for every string. The test was made with 0.15 in. (3.8 mm.) of water per minute at 45 deg.
Puncture Test. Under ordinary conditions, it is almost impossible to puncture an insulator, in dry air, since a well-proportioned insulator will flashover at a voltage well below its puncture voltage. To obtain values for the puncture voltage, it is necessary to immerse the insulator under oil and to take a number of other precautions, like the protection of leads, etc. Following this plan, a series of tests was performed in which this voltage was determined for all the different types of insulators.
Mechanical Tests. The testing device used to determine the breaking strength of suspension insulators, consisted of a frame-work in which the insulator was fastened by links and steel cables and the tension applied by means of a screw acting on a lever. A robust dynamometer indicated the maximum pull exerted by the screw and that pull multiplied by three, the ratio of the lever arms, gave the actual tension on the hook of the in-sulator. Voltage was applied across the insulator while pulling; but it was found that the insulator punctured always at the moment of fracture, so this method was discontinued on subsequent tests. A number of tests were made with each type to arrive at a fair average figure, and photographs were taken of the appearance of the fractures.
The above is a brief outline of the apparatus and methods used in making the tests on high-tension suspension and strain insulators. A number of post type insulators were also tested in a similar manner. As there were but two types offered, neither of which met the specified tests, considerable development work was necessary until fairly satisfactory types were evolved.
After completion of all design tests and before the final selection of the insulator best fitted to fulfil the specified require-ments, one week was set aside for witness tests. This was done to demonstrate to the manufacturers and their engineers the method which was followed in making these tests and to give them an opportunity to make their own observations with regard to the results obtained under conditions controlled in accordance with the tests specified. These conditions, as mentioned before, were kept unaltered during all tests and were constantly checked and adjusted, if this was found necessary.
The tests performed in the presence of the manufacturers were really nothing else but a repetition of the tests already made, and incidentally, served the purpose of furnishing an additional set of confirming results. In every case, these results checked closely with those obtained during previous tests, as the conditions under which each test was made could easily be duplicated.
TEST RESULTS
SUSPENSION INSULATORS
As to the method of comparing the performance of the different types of insulators under test, it became apparent from the start that no standards existed which could be followed or used as a guide. From an academic standpoint it would, perhaps, have been of importance to measure the watts lost for each type of insulator under varying conditions. This method may give results which would allow of direct comparison, but the difficulty of measuring power accurately under the conditions imposed by the test, and at such high voltages, appears to be out of proportion with the expected accuracy of the results.
It was, therefore, decided to compare qualitative rather than quantitative results. Under the assumption, which should not be far from correct, that the power loss of an insulator would make itself manifest in a proportionate display mostly of luminous character, the direct comparison of this visual display with the voltage required to create it should give a fair means of judging the relative insulating value of two insulators, provided all other conditions remained the same and unaltered. In applying this method in practise, there are, of course, a number of other considerations requiring attention. For instance, the display of luminosity may appear gradually in direct proportion with the voltage applied, or it may appear rather suddenly after a certain limit of voltage has been reached; or else, the display may appear to be localized at some parts or points, which, though a portion of the insulators, are of no value to its insulating quality and merely show faulty design. As will be explained later, the presence of such parts is always the cause of failure, regardless of the quality and design of the porcelain parts themselves. Taking into consideration the many sources which contribute towards the discharge of an insulator under potential, and by following the system of comparison outlined above, it was possible to classify the insulators according to certain well-defined merits and demerits. After balancing all merits of an insulator against all its demerits and by succes-sively eliminating those insulators possessing the greatest number of demerits, it was possible to arrive at one type which had the least number of disadvantages and the most of the advantages.
In the discussion of the actual test results obtained, the different types of insulators are designated with the letters A, B, C, D, E, and F, in accordance with the half-tone illustrations representing the different makes. From the results of the dry test, it can safely be said that all insulators with the exception of types A and B withstood the tests of three times line voltage more or less satisfactorily. The following table gives the actual results in condensed form:
From the wet tests, which were the most significant of all, the final results of the series executed with 0.5 in. (12.7 mm.) water per minute are given below:
The first visible discharge occurs invariably around the top section, in the form of streamers radiating in a more or less oblique direction away from the edge of the top skirt. The subsequent breakdown of the insulator appears to grow gradually with increasing voltage. This is especially noticeable on type A, whereas it is less prominent on other types; i. e., they may hold out fairly well until a critical voltage is reached. Above this voltage, the insulator will fail rapidly with relatively slight increase of voltage. As a rule, more or less active discharge always takes place around the pin of the insulator within the hollow of the petticoat on all insulators designed along the orthodox lines of a pin insulator like types C, D, and F. This discharge is practically absent in the one-piece insulator E, which is not provided with an inner petticoat. Any discharge which occurs at the point where the pin issues from the porcelain disk is effectively broken up and confined within a small, concentric corrugation.
The character of the breakdown is different for each type of insulator. Type A breaks down on account of the excessive leakage; the whole insulator becoming conducting, as it were. The breakdowns of the other types have more the character of a flash-over from one section to the next, the moment the voltage is high enough to break through the wet and conducting air enveloping the sections. This breakdown voltage could be ascertained with fair accuracy, and these figures were used as merits or demerits in accordance with their relative values.
From these records one feature is especially worthy of note. Almost in every case the discharge of the insulator was started by a sharp corner or point of the metal fittings by means of which the sections were held together. In all cases except A, B and D, the static field around the insulator sections was uniformly distributed in consequence of the almost symmetrically arranged parts occupying a space within this field. The word "almost" is used, as the presence of even slight projections like the head of a cotter-pin in the bolt linking the two sections together was enough to break up the air at that point after a certain voltage was reached. In case D this phenomenon was particularly noticeable. The metal parts in the shape of two prominent hooks were so large that the field was excessively distorted, creating highly uneven stresses in the air. The highest stresses are localized at the sharp point of the hook, as is apparent the moment the voltage is raised above a critical value. That this distortion of the field is always accompanied by a premature failure of the insulator can be proved by eliminating those unsymmetrical iron parts, covering them, for instance, as was done in some experiments, with a cylindrical metal shield. Although the striking distance is thereby somewhat reduced, the insulator is capable of withstanding a voltage at least 10 per cent over and above that which it was able to withstand with the hooks bare. In case of A the distortion of the field is especially prominent. As beautiful as the link feature appears from a purely mechanical viewpoint, it creates most unfavorable stresses in the air between the disks, likewise in the holes within the disks. The stresses in the porcelain cap are more uniformly distributed in all cases except A and B. In these latter, the dielectric is strained the most at that point where the interlinking metal parts have their least separation from each other. In all other cases in which use is made of a metal cap and pin, the stresses in the porcelain are higher closer to the pin, and decrease gradually and uniformly towards the cap. As long as the highest value of this stress is well below the safe working limit the insulator is not endangered. But in every case, the diameter of the pin, together with the voltage it assumes, remain the determining factors for the highest stress of the porcelain within the cap. For this reason, it seems that no advantage is gained by the use of a two-piece insulator. Theoretically correct. the idea of using two thicknesses of porcelain would appear to offer a larger margin of safety. In practise, the idea cannot be worked out to its full efficiency, for the size of the pin cannot be increased without correspondingly increasing the size of the cap, making an insulator of this sort too bulky and altogether impractical.
From all these considerations it was found advisable to have the metal parts of the insulators as symmetrical as possible, presenting a smooth and even surface void of any projection whatever. In accordance with this suggestion the ball and socket type connection was subsequently devised by one manufacturer to meet this contingency.
Another feature which militates against the use of a two-piece insulator is the fact that it is impossible to equalize the stresses in the two pieces under all and any conditions. In a dry condition, the porcelain of the inner petticoat is far less strained than when the top section is wet and conducting. The working efficiency of the material is bad and the cost of a two-piece insulator is necessarily high.
Outside of the design determining the electrical efficiency of the insulator, the method of mechanically connecting the different units to a string is of no little importance. The practise of cementing a pin into a porcelain shell and subjecting the pin to a strain may, at first sight, be regarded by many engineers as a doubtful proposition, and it was in that light that the inter-linking feature of types A and B was devised. As already mentioned, this link feature has proved to be a failure, at least electrically, and the cemented pin has so far given no cause for complaint. The breaking strength of the cemented pin, on the other hand, had been found to be far superior to the wire link type, in some cases being nearly twice as strong. The device of type C is unquestionably a most splendid solution of the problem, as for an actual holding power, this type can hardly be excelled.
As stated elsewhere, it was possible to tabulate the test results and to classify the insulators according to their merits and demerits. How this was done will appear from the two following tables. The first table contains a summary of characteristics, i. e., a tabulation of all features which can be measured and expressed in one or another unit. The first column of this table contains characteristics like diameter, spacing, number of sections, length over-all, open spacing between sections, widths, etc. A number of other characteristics relating to design are also added, like number of pieces, method of connecting the sections, material, etc., and finally, characteristics bearing directly on their electrical efficiency, like leakage distance, thickness of shell, dry surface, etc. Opposite each column representing the actual figures corresponding to each type of insulator were placed certain comments, indicating observations or deductions with regard to those particular characteristics.
The second table is really a condensed statement of all results from actual tests, both electrical and Mechanical. The table also contains the specification requirements. In compiling this table, a certain assumption was made which, although not absolutely correct, was justifiable in the light of the present comparison. For instance, in reference to the number of sections used, it was assumed that the share of line voltage per section was in direct proportion with the line voltage and number of sections.
According to the values given in the table, each section of insulator is subject to a voltage of approximately 22 kv. in all cases, except in type E, where this voltage drops to 15.7 and even 13.8 kv. per section, according to whether a complete insulator is made up of seven or eight sections, respectively. With the flash-over voltage per section known, the ratio of flash-over voltage to share of line voltage can be determined, this figure being equivalent to a safety factor against flash-over for the individual section. From the table it becomes at once apparent that type E has a very high ratio in comparison to type A, which has the lowest. Likewise, with the puncture voltage per section known, the ratio of puncture voltage to share of line voltage represents another safety factor against puncture, which, as in the former case, is the highest for type E and lowest for type C. The voltage per inch leakage distance has been found to be highest with type A, and lowest with type C, type E being next highest.
The table also gives the approximate percentage of sections puncturing. During the long run of the test it was found that insulator sections would puncture for no apparent cause and a record was kept of all these failures. At the end of the test it was considered of importance enough to compare these percentages with each other, assuming that these values could be taken as a fair indication of the superiority of one insulator above the other, with reference to its dielectric strength. From the table it will be found that types F and A both had exceptionally high percentage of puncture as compared with the low percentage of type E.
The average breaking strength of the different types of insulators as found from numerous tests are tabulated in the last-named table under " Mechanical Tests." The highest values were obtained with type C, the lowest with types E and F, all three types being cemented insulators. It must be said in defense of the last two types, that subsequent tests on regular stock insulators showed a breaking strength of not less than 8000 lb. (3628 kg.), the relatively poor results obtained by the former tests being due solely to the cement which had not properly set.
From all observations and test results, the following conclusions were drawn up on which a classification of the various types of insulators, in the following order, was based:
1. Type F. This type meets electrical requirements but not the mechanical tests. Design, however, can be readily modified to meet mechanical tests and incidentally, improve the insulator electrically. Percentage of puncture can be kept down by rigid inspection. Insulator shows high class of workmanship and material.
2. Type E. This type meets electrical tests with eight sections, but not the mechanical tests. Insulator should, without material modification of design, be able to come up to the required mechanical tests. Slight increase in diameter should also increase electrical efficiency of insulator. Large number of open spaces between units is of advantage. Insulator is strong, durable, light and compact. Method of connecting units should be modified so as to present symmetrical and smooth surface to prevent premature discharge.
3. Type D. This type meets electrical and mechanical tests. Insulator has, however, very faulty design. Diameter too large; weight and bulk too high. Inefficient cementing of hook. Hook feature to be condemned, causing distortion of field and premature discharge. As a two-piece insulator, electrical stresses of petticoats are not balanced.
4. Type C. This type meets the mechanical but not electrical requirements. The insulator is far too fragile, causing excessive breakage in ordinary handling. Sections are too close upon each other, leaving too small a clearance between units. As a two-piece insulator, electrical stresses of petticoats are unbalanced.
5. Type A. This type meets mechanical but not the electrical requirements.
Final selection of the type E insulator was made in consequence of various favorable considerations. Type F is of European design and manufacture, and its selection would have entailed several difficulties, especially in regard to delivery. Next to type F, type E was found to be the most suitable and practical insulator, both from an engineering and a commercial point of view, and this consideration, together with the outlook for better deliveries, determined its adoption. It must be mentioned that the diameter of the insulator was subsequently changed from 10 in. (25.4 cm.) to 11 in. (28 cm.), and that the ball and socket type connection was universally adopted.
STRAIN INSULATORS
No special insulators were offered for use as strain insulators excepting the one designated as type B, which is but a variation of type A. In order to increase the efficiency of the suspension insulator for use as a strain insulator, one or two additional sections were added by the manufacturers. From numerous tests similar in character to those performed on suspension insulators, it was found that none of the different types recommended by the manufacturer met the requirements of the specifications for wet test. Excessive leakage at voltages below the standard fixed in the specification (220 kv.) made their use as strain insulators prohibitive. As one exception, type E, using as many as ten sections instead of seven, showed some advantage over the others, but even at its best was found to be not entirely satisfactory. In every case, failure of the insulator did not occur suddenly, but very gradually. Distress begins to be visible at voltages as low as 110 kv., this distress increasing in almost direct proportion with the voltage.
After considerable experimenting with new designs and numerous combinations, it was found that the use of ten sections of the adopted insulator type E gave the least unsatisfactory results of all. With a modification of the design of the cap, increasing the breaking strength of the insulator, this type was finally adopted for use as strain insulators.
The preceding sections of this paper dealt with the investigation only in so far as it covered the selection of a suitable insulator. With this question settled, there remained one not less important part of the work, viz.: the supervision of the factory tests on some 140,000 insulator sections. The specifications called for distinct electrical and mechanical tests on each unit, and the acceptance of the insulators was based on their ability to pass these tests. Outside of these specified tests, the insulators had to conform to certain well-defined standards as to shape, quality, finish, etc. The whole inspection and supervision of tests was comparable to a weeding-out process, and it was the duty of the inspectors to see that this process was carried out in conformity with the specifications. After successful completion of all factory tests, the insulators were packed and shipped to the nearest railroad siding where they were delivered to the contractors.
Even though the specifications were drawn up with the utmost care, taking into consideration every phase of the work involved, it was found during the course of this investigation, and especially during the subsequent work at the factories, that they did not meet every contingency. In a number of instances it was found almost impossible to hold the manufacturer down to the terms of the specification, but that he had to be allowed a considerable margin in his favor. Although the manufacturer guaranteed to furnish insulators in accordance with samples submitted and approved, in the regular course of manufacture it was found to be a commercial impossibility to keep the standard at par with the hand-picked samples. In prescribing limits between which variances were allowable, bah mechanically and electrically, it proved to be a very difficult matter to draw a distinct line. After the contract was let and the manufacture was progressing, difficulties were encountered in determining when an insulator had successfully passed certain inspection or tests, requiring several conferences between manufacturer and engineers in order to come to a definite understanding. From all these experiences and observations, it was found that specifications for high-tension insulators were susceptible of a number of amendments, which, if properly worked out, would go far towards minimizing possible misinterpretations and misunderstandings.
In viewing this work now, after a number of years rich in experience have passed, and in the light of all after events, it must be admitted that the problem of insulating high-tension transmission lines is yet far from being solved. Much valuable experience has been gained which in the course of time will undoubtedly be utilized to improve methods and means of effectively insulating and protecting a transmission line. With special reference to the question of insulators and their future development, it will be understood by all, that work in this direction can be carried out successfully only with a close co-operation between the ceramic and operating engineers. The question of properly designing and loading an insulator is one which presumes a thorough knowledge of transient phenomena occurring on a transmission line and their proper interpretation with regard to the effect on the insulators. Once these phenomena are known and their effect thoroughly understood, the drawing up of specifications for:high-tension insulators will become a matter less open for conjecture. For it is quite probable that precautions, now taken in one direction, are often unwarranted and uncalled for, whereas, on the other hand, liberal allowances made in other directions may be of the greatest detriment to the line and insulators.
In summing up the experience gained during the foregoing investigations, especially with regard to testing, the following points are presented as worthy of future consideration and discussion.
They are given in the form of an itemized list of headings or questions to which are added a few remarks, commenting on certain experiences gained either in the field, in the factory, or in the testing room.
Design Test. What design test should be specified for insulators intended to work at a certain voltage?
In the present case a dry test of three times line voltage was specified. Experience. however, seems to indicate that even though the insulator may meet this arbitrary condition, its safety against failure in actual operation is not thereby assured. It is a well-known fact that an insulator is never endangered by the steady static forces but rather by those sudden and transient movements appearing in a system and caused either by external or internal disturbances. It is not the steady dead-load which is dangerous to a bridge or structure, even though it may assume a value two or three times higher than the load for which it was designed, but those moving loads which will set up vibrations and surges in the structure, especially if they are rhythmical in character and coincide with the natural swing of the bridge. For this reason, soldiers are generally not permitted to cross a bridge while marching in step. Although the actual forces coming into play are insignificant in such cases, their effect may, under certain conditions, become disastrous. It is without doubt that the insulators of a transmission line are very susceptible to similar phenomena, and to guard against failure from these causes, it will be necessary to impose tests of an entirely different character.
Method of supporting insulator during test. Should insulator support be grounded and voltage applied to groove, or should voltage be applied between groove and pin, both ungrounded?
At first sight, it may appear as if the manner in which the insulator is supported during tests is of no importance. As a matter of fact, the proximity of large grounded or ungrounded bodies close to the insulator under test will materially affect the distribution of the static field around the insulator, especially when these tests are performed with one side of the potential grounded; the best method of supporting an insulator and applying a test would undoubtedly be the one which closely approximates conditions under which the insulator works in actual operation.
Capacity of testing transformer and generator. Method of regulation of voltage. Determination of correct voltage during test at any time. Should spark gap be used or static voltmeter, or should step-down transformers in connection with voltmeters be used?
The kilowatt capacity of the testing outfit cannot be too large, for the puncture of a weak insulator may never be discovered but for the power back of the transformer.
As to the method of regulating the voltage: It must be accomplished by means which do not alter the shape of the alternating-current wave form and the latter should be a true sine curve. From the different means employed today, like water rheostat, induction regulators, auto-transformers, etc., the method of regulating the voltage of the alternator by controlling its field current seems to offer the most advantages.
In reference to the determination of the voltage, several methods are at present in vogue. The most common of these methods involves the use of a properly calibrated spark gap. An ordinary voltmeter in connection with a step-down transformer is also used, and finally, in some instances, static voltmeters have given excellent satisfaction. Each of these methods, however, has its drawbacks. The spark gap setting is susceptible to atmospheric conditions. It may also introduce undesirable oscillations at the instant of discharge. The breakdown voltage of an insulator cannot be determined by means of spark gap alone, which in this case must be supplemented by a voltmeter reading. Another feature is the burning off of the points, each time the gap discharges, a matter which cannot always be avoided. The method employing step-down transformers is not altogether reliable and should be used in connection with a gap from which the voltmeter readings are calibrated. Undoubtedly the best method to ascertain the value of the testing potential is by means of a static voltmeter of suitable design.
Frequency, permissible distortion of wave form, effect of harmonics and high frequencies.
The effect of the frequency upon the results is a matter which is very seldom fully appreciated. The value of the charging current increases in direct proportion with the frequency, and the effect of this current will naturally follow a similar law. An insulator tested at 60 cycles will show different results than when tested at 25 cycles, the potential being the same in both cases. If for any reason, the wave form of the alternator is not a true sine curve, the results may become extremely misleading, to say the least. In one case which is on record, a porcelain transformer bushing was tested at two different places under apparently identical conditions, and yet the results differed by nearly 40 per cent. The tests were checked and repeated several times with no better results until, finally, the wave form of one of the alternators was found to have a very pronounced 13th harmonic. Immediately this harmonic was suppressed, the tests could be duplicated at both places without difficulty. The smaller the number of insulators tested, the smaller also the capacity of these insulators, the more pronounced will be any effect caused by higher frequencies appearing in the electrical system used for such tests. With a large number of insulators, and consequently, with a large capacity available, these higher frequencies will cause relatively little trouble, provided the amount of energy they represent is small. But in all cases where the capacity of the insulator tested is small, the wave form of the alternating current should be a pure sine wave.
As to the number of insulators which should be tested in order to arrive at a fair average value, this is a matter left open for discussion.
Effect of power factor upon test.
The effect of the power factor on insulator tests is also left open for discussion. When a large number of insulators are tested simultaneously, the available load of the transformer is utilized to charge that large capacity and there will exist considerable lead between this charging current and the impressed e.m.f. Whether or not this power factor has any influence on the test results is left open for discussion.
What wet test should be specified? Should it be artificial rain, dew, salt water spray, etc.? Amount of precipitation per minute? Character of precipitation and means for applying the same? Angle at which this precipitation should be applied?
Several means for approximating the conditions found in the open air are used at the present time. Artificial rain is applied which may vary between wide limits from a downpour to a mist; it may be applied vertically or at an angle, usually 45 deg. Or else the insulators may be confined within an air-tight room in which steam is left to escape until the insulators are completely enveloped in an atmosphere of steam and covered with a film of condensed vapor. Each test will yield certain results, but no two tests can be compared unless the conditions governing the tests are the same in both cases. Which of these methods is the most effective remains to be determined. It should always be chosen with regard to the facility for duplicating it at any time. In the present instance, the specifications called for 250 kv. with 0.5 in. (12.7 mm.) rain per minute vertically applied, or 220 kv. applied at 45 deg. These figures may seem arbitrary, and far above the standards commonly used, but on the other hand, they also include a safety factor higher than it is customary to allow. The above rain tests are easily made or duplicated, which is a great advantage. On the other hand, the distribution of the water needs considerable improvement to approximate more closely real rain.
What should determine the failure or the success of insulator under test? Should it be the luminous display when test is performed in absolute darkness, and if so, what should be the limit of intensity? Or, should the ratio of flash-over or breakdown voltage to voltage at which first sign of luminosity appears, be considered?
With all conditions of test fully determined, and agreed upon, there remains the most difficult task of all; namely, to judge the performance of the insulator under test. The method followed and described elsewhere in this paper was the only one which promised to yield comparable results. This method, however, has the disadvantage of being a purely subjective matter. Even the comparing of photographic records is susceptible to that personal element always present. That the method is not free from objections has been realized from the start, but in the absence of some better way, it had, at least, the advantage of simplicity. It is quite evident that there must be other ways of determining the efficiency of insulators than by merely comparing their luminous display either among themselves or with that of a standard. What this method should be, is an open question. Undoubtedly, the determining of watts lost would yield results free from the personal element if a reliable method could be devised. In regard to the other method mentioned, in which the ratio of breakdown voltage to voltage at which first brush discharge becomes visible, is made the basis of comparison, it is likewise not always an easy matter to determine the exact value of this voltage. The breakdown voltage of an insulator, as a rule, is fairly constant, but the voltage at which the brush discharges become visible depends largely upon various accidental conditions and eventualities which render its determination extremely difficult. Consequently, this method should be viewed with the utmost caution.
Puncture Test.
Method of applying and performing test. Method of apply ing electrodes. Number of samples to be tested in order to arrive at a fair average value.
As a well-proportioned insulator will flash over before its puncture voltage is reached, it becomes necessary to test the insulator under oil. In this test, the most important feature is the application of electrodes. Unless the area presented is of sufficient size, erratic and unreliable results are obtained. The cemented cap and pin of sections of the suspension-type insulator form ideal electrodes in a test of this character, inasmuch as they distribute the stresses in the porcelain evenly and uniformly, also in exactly the same manner as obtains in actual use.
Mechanical Test.
What should this test be? Method of subjecting insulator to mechanical test according to whether pin insulator, suspension or strain insulator is tested. Method of applying load. Method of recording load at any instant. Should mechanical test be performed with insulator under voltage?
Routine Test and Inspection.
Inspections for physical defects. What are the limits to be observed in rejecting insulators on account of mechanical imperfections?
There naturally exists considerable difference of opinion between manufacturers and engineers as to the insulators which should be rejected. In most instances, the porcelain manufactured in this country will show an imperfect surface. This imperfection is caused by warping of body, discoloration, small cracks, flaws, grooves and foreign material adhering to glaze, bubbles underneath the glaze, etc. As a rule, foreign and especially German porcelain is faultless in those respects and there is never the slightest difficulty in rejecting insulators in these factories.
What routine tests should be specified? Method of applying such tests. Number of insulators tested simultaneously. Method of applying voltage. Should these tests be continuous or should test be executed in stages, allowing for the removal of insulators failing during test? If insulators are subjected to time test, should the test be continuous? Is it good practise to subject insulators to flash-over test for any length of time, with regard to possible deterioration or fatigue of the porcelain? What conditions will determine the success or failure of test? Should insulators failing, be cut off automatically from the rest of the insulators under test?
Two-Piece Insulators.
Where insulators are made up of several parts, cemented together, should cement preparation and method of cementing be specified? Length of time allowed for setting? Should insulator be tested over electrically after cementing is done?
Considerable difficulties were experienced in Germany with cemented insulators. The cement used in that country is either plaster of paris prepared in a special way, or litharge and glycerine and several other cements of secret composition. It has been found that after some time, the shells would crack, due—as was inferred—to the working of the cement used, and for that reason the cemented type insulator has been abandoned in favor of the single-piece type.
While the insulators which were selected as a result of these tests have proved to be highly satisfactory throughout a period of two years' operation, there have been nevertheless a few characteristic failures. In most cases the ultimate failure of these insulators was due to puncturing, although there exists strong evidence that this failure was preceded by the cracking of the petticoats, through no apparent cause. In most cases when puncturing takes place, it affects all sections of the insulators, with the curious but nevertheless logical result that holes are burned through the insulator cap opposite the point of puncture in the porcelain and fusing the metal surrounding it. The size of the hole in the cap depends to a great extent upon the time setting of the circuit breakers at the power station. If the circuit breakers trip out instantaneously after puncture has occur-red, there may be no burning of the cap whatsoever. On the other hand, if the circuit breaker holds on for three or four seconds or even longer, the current to ground, which is limited through a resistance of large heat capacity connected between the neutral point of the transformers and ground, fuses the porcelain and adjacent metal parts. In one of these cases, where the holes through the caps were quite large, the circuit breaker did not trip out at all and one of the insulator sections, under the excessive heating action of the current to ground, finally came apart, thus automatically interrupting the circuit.
It is a rather difficult matter to determine the primary source of these failures. It may be due to cracks in one or two of the sections, which naturally tend to lower greatly the total insulating capacity of the insulator. If then, during a lightning storm, or during switching, surges are set up in the line, the weakened insulator becomes punctured as a direct result of any discharge occurring in the vicinity along the line. Considering the large capacity involved in the system and the power back of the transformers, this localized discharge in the insulator may immediately set up powerful commotions and oscillations which may affect the adjacent insulator and finally puncture one after the other in rapid succession.
The cracks inside the insulator cap and also the falling apart of the insulator sections with no apparent cause may have been due to faulty insulators having accidentally passed inspection, or it may have been brought about by the sub-sequent expansion which takes place in the cement within the cap and around the pin. This cracking of the cemented insulator has been experienced in Germany, as mentioned elsewhere in this paper, although in these cases the insulators in question were pin insulators only. It is almost impossible to determine whether the cracking of the porcelain within the cap is actually due to the uneven expansion of the porcelain, cement, or metal cap under temperature changes, or whether they were a result of the puncturing of the insulators. The inspection of some 140,000 sections of insulators was a task requiring much endurance, and although the work was carried out with great conscientiousness, it is quite possible that a few sections with a weakness in the porcelain cap which would not be detected by the ordinary routine test, may have passed by. There is also to be considered the theory of electrical and mechanical fatigue in the porcelain and cement respectively, which has already been discussed by several authorities. There seems to be no doubt that some such effect takes place, but the data which have been collected so far on the subject are not sufficient to permit of definite conclusions being drawn. To be able to insure absolutely continuous service over any transmission system may necessitate the sectionalizing of the line, where each section can be periodically tested at much higher voltage, or else it may be necessary to remove the insulators in batches and test them individually, as has already been done by one of the large operating companies.
At the present time there appears to be a rather unwarranted competition by the different manufacturers and operating companies to use excessively high voltages. There is a system already in operation at 145,000 volts and quite recently another company has contemplated the use of 180,000 volts. In view of the fact that operation at 110,000 volts has not yet reached a stage of maturity, and the fact that phenomena which were not anticipated, occur on such lines, and which, even now, are far from being fully understood, considerable caution should be displayed before attempting the use of still higher voltages. A few years ago the suspension type of insulator was heralded as the solution for line insulation up to any voltage at which it would be practicable to operate for many years to come. The factor limiting the use of high voltages so far as the line was concerned was then considered to be the effect of corona and leakage into the atmosphere. But from past experience it is almost certain that these views will need revision and that a systematic and thorough study of the properties of insulators is urgently required.
Examples of the rather uncertain conditions manifesting themselves in a high-voltage power transmission system are mainly the behavior of oil circuit breakers when large amounts of power have to be handled, the lightning arrester problem, and even the high-tension transformers. Most all of this apparatus, as will be admitted by the manufacturing companies, is yet in the stage of development, and it is very gratifying to see that a large amount of study is being devoted at the present time to rendering these devices more reliable in service.
The above criticism should not be taken as an indication of extreme conservatism or as tending to block the way to progress, but under the prevailing conditions, it is almost imperative that a word of caution should be spoken to prevent the somewhat extravagant use of the higher voltages when the use of lower voltages would answer the purpose equally well, and especially when the difficulties which are encountered with these extreme voltages may endanger the financial prospects of a particular power proposition.
I cannot close this paper without mentioning the Ontario Power Company at Niagara Falls, in whose plant these tests were made. The president of the company, Mr. J. J. Albright, and vice-president, General F. V. Greene, and their engineer in charge, Mr. V. G. Converse„ have given their heartiest support and assistance to the furtherance of this work in gratuitously supplying all necessary testing equipment, power and the help of their personnel for these tests. I welcome this opportunity to personally and publicly thank these gentlemen for the interest they have taken in this work, for their generous help and friendly cooperation.
A paper presented at the 278th Meeting of the American Institute of Electrical Engineers, New York, December 13, 1912
