[Trade Journal]
Publication: Journal of the American Institute of Electrical Engineers
New York, NY, United States
p. 571-, col. 1-2
Factors Controlling the Design and Selection of Suspension Insulators
BY W. D. A. PEASLEE
Electrical Engineer, Jeffery-Dewitt Insulator Co.
A discussion of the factors entering into the design and operating behavior of suspension insulators and the problems to be solved in designing a suspension insulator to overcome the objectionable features shown by experience to affect seriously the operation of the insulators in service.
Factors to be taken into consideration in the selection of suspension insulators for a given condition are given and a brief discussion of the general trend of future improvements is presented.
INTRODUCTION
IN the early days of electrical distribution of power the insulator problem was unimportant. The insulator gave more satisfactory service than the rest of the apparatus essential to the generation and distribution systems. As long as the voltages were low the dielectric field distribution was of relatively small importance. As the transmission distances and therefore the economic transmission line voltages increased the insulator problem became more Acute. The first attempt to meet the insulation requirements of these higher voltage lines was an increase in the physical dimensions of the lower voltage type of unit. No attention was given at this time to the distribution of the dielectric field or its shape although the laws governing the dielectric flux distribution in such cases were well-known.
As a result with the increased voltages came an increasing amount of insulator trouble until when the transmission voltage passed the 30,000 volt mark, the insulator problem became of greatest importance. Improvement in design through rational study of the problems had brought the reliability of other parts of the transmission and generating systems to a very satisfactory point. The insulator, however, had not made a corresponding advance and failures were encountered at a rate that for a time threatened the success of high-voltage transmission of electrical energy.
The attention of the insulator manufacturers was turned at once to the problem and many new designs were brought out as suggested remedies for this situation. Practically none of these was based on a rational study of the insulator as a dielectric problem, most of the improvements being made from the narrow standpoint of the small experience then available. The problem was attacked by manufacturers and research men of the country, but unfortunately from widely different points of view. The manufacturers being limited by manufacturing difficulties and the great cost of a radical change in methods, clung to small changes in existing forms and processes, while the research man attacked the problem from a scientific standpoint, based on a careful study of the dielectric and mechanical problems involved, but too often handicapped by a lack of knowledge of manufacturing processes and their limitations. For these reasons many excellent idea: coming from both sources were laid aside from lack of coordination of the two lines of study.
In the early insulator types, at times the flash-over distance was much greater than warranted by the thickness of dielectric and many failures by electric puncture were encountered, also the design was such that corona was formed at different places on the insulator at low voltages.
Gradual improvements in the design eliminated many of these objectionable features one by one, and improvements in manufacturing methods brought forth constantly improving grades of porcelain.
When the pin type insulator reached a limit set by size, weight and cost, the suspension type unit was introduced. This was a decided step forward in insulator practise, but unfortunately the designers of the suspension type unit still neglected a thorough consideration of the dielectric field of flux in the designing of their units, making them simply mechanical modifications of the existing types.
Thus a great many faults of the early pin type insulators were repeated in the first suspension units. Due to its small size the flash-over voltage of the suspension unit was practically always below the puncture voltage, though, as will be shown later, the margin was not sufficient, and, with the introduction of the electric tests on assembled units in the factory, very few direct puncture failures were encountered when the insulators were first placed on the line.
At this point, however, a type of failure appeared which may be classified as a deterioration failure, the insulator passing successfully severe factory tests, but failing after a period of service on a transmission line under conditions less severe than those successfully resisted in the course of factory testing.
The study and analysis of this problem has filled the pages of engineering literature during the past ten years and many divergent theories regarding the causes of and remedies for the various types of failure have been advanced. At the present time insulators successfully passing factory tests deteriorate in service at rates varying up to 20 per cent per year. As stated by a prominent transmission engineer quite recently:
"All insulators at present on the market seem to be subject to a steady depreciation that is too large to be ignored or accepted as an operating necessity." The conventional type suspension insulator unit, and also, to some extent, the multi-shell pin type unit, seem in general to be subjected to the types of failures indicated in the following table:
MECHANICAL FAILURES
a. Due to the use of materials having widely different coefficients of cubical expansion as in conventional cap and pin construction which causes enormous stress under temperature changes.
b. Due to mechanical overloading.
c. Due to shocks as shooting.
d. Due to lightning and power arcs.
ELECTRICAL FAILURES
a. Actual electrical puncture.
b. Leakage under adverse conditions followed by flashover and navy power arc.
e. Due to porosity.
In the conventional type of insulator three materials, porcelain, cement and steel, are tightly compressed in contact in an unyielding fashion. These materials have different coefficients of cubical expansion and the temperature variations, in many cases quite abrupt, met with in operation, seem to set up internal stresses which crack the porcelain, leading to electrical failure. Further the cement itself is subject to volumetric changes somewhat of cyclic nature and also of a crystalline growth character that contribute to these phenomena. Prominent engineers have expressed the opinion that 85 per cent of the failures of insulators of this type were preceded by mechanical failures of this class. The sun striking upon insulators on a frosty some has in many cases been the signal for some rather startling exhibitions of such failures. In connection with this, the internal stresses existing in the porcelain parts due to improper manufacturing methods and firing, have doubtless contributed to this condition. That the manufacturers recognize this weakness is well shown by the elaborate precautions that have been taken to reduce this effect through the medium of felt washers, lead thimbles, etc., appearing more recently in their designs.
That the transmission engineers of the country have realized the importance of the deterioration type of failure is indicated by the extensive study, which has been made of the various methods of testing employed by most engineers responsible for large transmission systems today, such as the megger and buzz stick methods. Reliance is placed on these methods, to detect the beginning of this deterioration permitting the removal of the affected insulator before it has dangerously weakened the string. Many engineers are also advocating the deliberate addition of several units to an insulator string above the number required for actual insulation purposes as an insurance against this deterioration regarded by them as inevitable.
Failures due to mechanical overloading are rare in modern lines as the lines are usually designed with proper consideration of extreme loading conditions and ample mechanical safety factors. The same remark may be applied to failures from shock and shooting, and although at one time about the most popular outdoor sport, in certain localities, for irresponsible people, was the shooting off of the power company's insulators, fortunately, this condition is no longer of very great importance. The failures due to lightning and power arcs are, however, at the present 'time rather large. It is doubtful if we could define exactly what might be considered a direct stroke of lightning, and probably such strokes on transmission lines are rather rare. Lightning flash-over of an insulator string is usually in itself rather harmless, but the power arc that follows the static flash-over is extremely destructive to any but the most substantial types of insulator. The thinness of the porcelain part of the conventional type insulators, combined with the abrupt changes in form and surface directions renders them susceptible to destruction under the action of the intense heat of such an arc. Any insulator with thin petticoats is very likely to be considerably damaged by power arcs as the temperatures and the mechanical stresses involved are very high. The chief requisite for an insulator in this regard is strength, gradually increasing thickness from the edge of the skirt inwards, and a high thermal capacity. Insulators so designed will successfully resist severe power arcs and lightning surges, especially when the system is equipped with the proper kinds and numbers of relays, to a remarkable degree.
An actual electric puncture is probably rare on any modern insulator that has been properly fired, most electrical failures being the result of previous mechanical failures.
Leakage is a problem that is to a large extent dependent upon localities and specific conditions. Smelter fumes, salt fogs, dust storms and many other causes tend to make the leakage effect vary and it has been generally conceded that in bad localities the only remedy is a periodic cleaning of the insulators. A few extra units added to a string will postpone the inevitable cleaning, but it is probably safe to say that under bad conditions no insulator string could be used commercially that would not require cleaning after a time. In connection with this, however, as leakage always culminates in a flash-over, it is important that the insulator be able to withstand power arcs, especially in regions subject to bad leakage conditions.
Porous porcelain absorbs moisture from the atmosphere, thereby decreasing its electrical resistance. The leakage current flowing through the porcelain under electrical stress tends to heat localized portions to a very high temperature. This local heating causes mechanical failure followed by the passage of a power arc through the porcelain, or due to the negative temperature coefficient of electrical resistance of porcelain, the leakage current may under certain conditions gradually increase with a concomitant increase of temperature, this action being cumulative until the porcelain is punctured. A good many instances of failure of this kind both in laboratories and under field conditions have been encountered. Good glazing postpones the deterioration of porous porcelain but cannot eliminate it.
Until recently the progress in the manufacture of suspension type insulators has been rather largely along certain detailed attempts at the improvement of certain specific faults such as the utilization of felt washers, lead thimbles, etc. in the conventional cap and pin type design. The problem had not been attacked from a sufficiently scientific standpoint and there is still great need of a scientific study of this problem based on an analysis of the dielectric field of flux around insulator strings, and the electrical and mechanical requirements of the units in relation to the limitations imposed by ceramic and manufacturing conditions.
The following discussion of the analytical and experimental work undertaken along this line from the electrical and ceramic standpoint, the progress that has been made and the results that have been secured in the form of rationally designed insulators will, it is hoped, be of some interest to the operating engineer and stimulate further study and advance in this vital subject.
FACTORS GOVERNING RATIONAL INSULATOR DESIGN
The requirements to be met in the design of suspension insulators may be broadly classed under two headings:
1. The insulator must support the line mechanically with adequate safety factors under the most adverse conditions.
2. The insulator must insulate the line with adequate safety factors under any electrical conditions not rendering other apparatus on the line inoperative.
It is obvious that any suspension insulator must be designed in the form of a unit that will meet widely divergent conditions. That is, from the manufacturing standpoint it is inadvisable to manufacture units of different mechanical strengths for different weights of conductors and climatic loadings. The design hinges then upon a unit that in the heaviest lines considered under the most adverse conditions of loading will give an adequate safety factor and will yet be cheap enough to be used on the less important lines.
The insulation afforded is obtained by building up strings of different lengths, but it is hardly advisable to attempt to insulate a line at great expense to withstand almost infinite voltages when, due to the limitations of other apparatus connected to the line, the system will be inoperative under extreme over-voltage conditions.
A study of existing lines and the probable limitations in conductor sizes and tower spacing of lines from 150,000 volts down, indicate that a mechanical strength of from 9000 to 10,000 pounds is adequate for a suspension unit, provided the unit is so designed that repeated stressing does not injure the unit electrically. The rational design herein discussed is, therefore, based on this mechanical strength requirement. The amount of discussion that has taken place recently regarding the use of porcelain in compression and tension makes it advisable at this point to discuss this matter a little in detail. The mechanical strength of ordinary porcelain in tension is in the neighborhood of 1500 lb. per sq. in., while the compressive strength is around 40,000 lb. in a porcelain having reasonable dielectric and temperature change resisting qualities. On account of the wide differences in these two figures many engineers have been dubious of the advisability of using porcelain in tension. The same argument might be used against the employment of cast iron in tension, and yet, although having very largely the same mechanical characteristics as porcelain, cast iron is consistently used in tension in the design of machines and structural members. As long as the unit stresses in the material are kept below the ultimate strength of the material with due regard to adequate safety factors, there is no rational objection to the employment of porcelain in tension any more than there is to a corresponding utilization of cast iron.
INSULATOR SHAPE AS AFFECTED BY THE DIELECTRIC FIELD OF FORCE
The dielectric field of force between similar electrodes is in general an ellipsoid of revolution though this is not strictly true, except between electrodes which are confocal hyperboloids of revolution, and no insulator electrodes are of this form. The agreement of the dielectric field with the ellipsoid is only approximate. However, the insulator should in general be symmetrical and conform as far as possible to the shape of the dielectric field. The placing of dielectrics of different specific inductive capacitances in series should be avoided, and therefore, the surface of the insulator should follow as closely as possible the lines of force in the field. In general the equipotential planes between the insulator electrodes should intersect for equal increments of potential, equal zone widths on the insulator surface. In connection with this point it is interesting to study Fig. 1. In this figure it will be noted that the conventional type unit does not conform to this requirement and the result of this lack of conformity is the appearance of corona on the unit at relatively low voltages, the corona appearing first where the equipotential planes are closest together. The requirement of a symmetrical shape introduces at once the problem of attaching the hardware to the porcelain in a different manner from that employed in the conventional insulator type. At the same time it becomes necessary to develop some form of hardware that will eliminate the terrific stresses imposed by the conventional type of hardware as previously discussed. Furthermore, in addition to the above requirements a large thermal capacity is necessary in a unit to enable it to resist power arcs and this demands a rather massive porcelain structure.
The design of the hardware presents a further difficulty that is solved only by a compromise between ease of assembly and security of the connection against actual failure or uncoupling. Furthermore, the hardware and shape design of the porcelain structure must necessary in a unit to enable it to resist power arcs and this demands a rather massive porcelain structure.
The design of the hardware presents a further difficulty that is solved only by a compromise between ease of assembly and security of the connection against actual failure or uncoupling. Furthermore, the hardware and shape design of the porcelain structure must be such as to resist to the greatest possible degree abrupt temperature changes.
DEVELOPMENT OF A RATIONALLY DESIGNED SUSPENSION INSULATOR
It is not generally appreciated by the high-tension engineers of the country that the electrical duty of the end unit is the basis of rational suspension insulator designs. Fig. 2 gives the distribution of voltage on the units of a string of suspension insulators, and it will be noticed that the conductor unit is carrying by far the greater proportion of the voltage stress. This unit is, therefore, the key to the design as, if it is so designed as to be safe, the rest of the units are obviously well within safe limits of engineering practise. The curves of Fig. 3 may be of interest, giving the percentage of the total voltage across an insulator string that is carried by the line and tower units respectively.
These figures bring out to a marked extent the advantage of a proper distribution of the equipotential planes as previously discussed in that such distribution produces a unit in which the corona voltage is very high. The reason for this unequal distribution of voltage has been discussed in engineering literature of recent years and will not be commented on here. In this connection, however, Fig. 4 is illuminating in the light that it sheds upon the distribution of the equipotential planes on an insulator string. This method of illustration is most graphic in showing the actual physical conditions surrounding an insulator string under operating conditions.
Referring again to Fig. 2, it is seen that under the conditions given with a five-unit string on a 110,000-volt delta-connected line (conditions which are being successfully met at the present time by the rational design under discussion), the conductor unit is subjected to a normal-frequency voltage of 33,000 volts. The maximum high-frequency transient that has been reported by writers and investigators as likely to be met with high-tension transmission lines is around 100,000 volts. It has been shown that the effect of the normal and high-frequency voltages combined in a circuit is to produce stresses which are the arithmetical sums of the normal and high-frequency voltages. This is readily understood as according to the law of probability, the high-frequency and normal voltage peaks will coincide in time relation a certain percentage of the time. The very high time-lag of such highly damped high-frequency transients as are encountered on transmission lines, renders possible the application of such combined voltage stresses to an insulator without flashing it over. In other words, the line unit in Fig. 2 might have impressed upon it a total stress of 133,000 volts and though the flash-over voltage of the unit is 100,000 volts this unit would not flash over under these conditions due to the large time-lag just mentioned. As insulators should operate with an adequate safety factor, it is obvious that under such conditions a puncture value of around 300,000 volts at 60 cycles is necessary.
In other words the puncture voltage of a rationally designed suspension unit should be in the neighborhood of three times the dry flash-over voltage at normal frequency and this is the fundamental basis of the design of rational suspension insulator units.
While leakage is a question very largely hinging upon particular climatic or other conditions a rational shaped design has been found to improve the ability of a given length of surface leakage path to limit the leakage current. Instances are known of ideal shape designs wherein the flash-over of the insulator was the same when previously cleaned, as when covered with a considerable coating of dust. This, of course, is an extreme condition but the fact remains, and has been demonstrated in the laboratory, that proper surface shape is much more efficient in this respect than surfaces wherein the divergence from the direction of the lines of force is marked. The length of the leakage path of a suspension type insulator is rather limited. The units are 10 in. (25.4 cm) or 11 in. (27.9 cm) in diameter, and if many thin petticoats are added to the unit to increase the leakage distance they are rendered much more susceptible to destruction by power arcs on account of the thin porcelain necessarily involved.
Porous porcelain is undoubtedly the cause of a great deal of insulator depreciation. One large insulator manufacturer has recently made the published statement that non-porous porcelain could not be made, stating, "a low moisture absorption is desirable, but it must not be assumed that any satisfactory porcelain can be made which will have zero absorption." This statement is absolutely challenged. One of the first objections to a rational insulator design was made by ceramic people who stated some years ago to the author that there was no doubt that such a design was desirable but that it was impossible to make a porcelain insulator of the shape, volume and thickness necessary without having it very porous. After a great deal of factory and laboratory research this problem has been solved and insulators can be made in practically any size or shape of absolutely non-porous porcelain as determined either by the psychometer or impregnation tests. This matter will be further discussed later in the paper.
It is well-known that the efficiency of an insulator string is a function of the ratio of the capacitance of the metallic interconnecting parts between the disks to ground to the capacitance of the insulator itself as a condenser and that the string efficiency is improved as this ratio decreases in numerical value. This point must be carefully considered in any rational design and hardware with a large surface between the units avoided as much as possible.
The advisability of a high impulse ratio has been admitted only quite recently by engineers in general, and this feature is of importance because the impulse ratio is a measure of the ability of the insulator system to withstand lightning frequency flash-over. The impulse ratio of a unit and of the string built up from such units is a very important feature of insulator design, and one which has not received the attention that it should have received from most manufacturers.
The voltage at which corona appears on a unit is of great importance as a reference to Fig. 2 will show, and it is important to have this corona-forming voltage as high as possible. On a rationally designed insulator this voltage should be considerably above 30,000 volts while in many conventional type insulators at present on the market corona appears rather decidedly at voltages from 12,000 to 16,000 volts. Fortunately the conditions necessary .for the attainment of the above features of insulators influences in the right direction the value of voltage at which corona will appear on the unit. Many lines are in operation with conventional type units wherein additional units have been added above the actual insulation requirements of the line to insure the operation of the string without corona on the conductor unit.
The conditions discussed in general require conflicting features of design and render the design of any insulator more or less of a compromise, and the skill of the insulator designer is tested in producing the particular compromise giving the best solution under the limitations of ceramic and manufacturing possibilities. The following brief description of some of the experimental work undertaken in the research laboratories of Jeffery-Dewitt Insulator Company in the study of the design and manufacture of suspension insulators may be of interest in showing something of the amount of work involved in such studies and something of the tendency and possibilities of future development.
DETERMINATION OF THE THICKNESS OF THE DIELECTRIC BETWEEN ELECTRODES
As previously discussed the puncture voltage of the high-tension insulator should be approximately three times the dry flash-over voltage. Having given then an acceptable dry flash-over voltage and mechanical strength, the first problem in the design of a rational insulator is the determination of the dielectric thickness between the electrodes. The curve 1 in Fig. 6 gives the puncture voltage of one type of porcelain against thickness. This curve was obtained with the form of electrodes shown in the figure, and is the basis of the design herein discussed.
The development of the hardware to meet .the requirements of symmetrical shape has been an interesting one. The first development in this design was approximately the insulator shape shown in Fig. 7, and the hardware was a solid cap at each end cemented into the porcelain. Electrically this was an excellent design, but on the application of the alternate immersion test wherein the units were immersed alternately in boiling and freezing water, it was soon found that the solid cap cemented into the porcelain was not permissible. The wide temperature variation imposed by this test damaged the porcelain rendering it mechanically and electrically unreliable. A gradual development towards flexibility resulted, after a great deal of experimental work, in the flexible spider shown in the design of Fig. 7. One of the features governing the development of this spider was the requirement that the plane of dielectric stress be maintained as near as possible normal to the plane of mechanical tress a condition which this method of attachment fulfilled admirably. The legs of this spider are fastened into the porcelain by an alloy having sensibly the same coefficient of cubical expansion of porcelain. By this means the well-known detrimental effects of cement are eliminated. The flexibility of the spider legs combined with this alloy give a unit that will withstand the alternate immersion test an indefinite number of times without any detrimental effect on the insulator. Tests have been made of this character up to 100 alternate immersions followed by high-frequency flash-overs and final breaking in a tension machine. All of the tests indicate that the unit as designed is free from the detrimental effects of wide temperature variations. After the development of this spider, curve 2 of Fig. 6 was made to determine the efficiency of the spider as a flux distributor. This efficiency is given in curve 3 of the figure and shows in the thickness of the porcelain used in the unit (57 millimeters), 2-1/4 in., a value of 72 per cent. This value is the ratio of the puncture voltages in the same thickness with the two types of electrodes, but as the puncture volt age is that at which the dielectric flux concentration at its point of maximum intensity exceeds the critical value, it is also a measure of electrode efficiency as a means of obtaining a uniform flux distribution.
The difficulty of making a porcelain insulator of this thickness has deterred manufacturers from progress in this direction, and we can substantiate a manufacturer most emphatically in this difficulty. A great many thousand dollars were spent before we discovered how to manufacture porcelain in this thickness without firing strains or porosity. It may be said here that the solution involved a radical departure from the prehistoric methods of porcelain manufacture that have been followed continuously for a long time by most porcelain factories. These changes are met throughout the process from the original handling of raw materials through the final firing processes, and it is only upon the development of these special processes, utilization of special drying methods and the use of the tunnel kiln for firing control that the problem has finally been solved. Fig. 8 gives the firing temperature porosity curve for one porcelain body and illustrates the narrow range over which this body may be fired to produce non-porous porcelain. In securing this curve the porcelain test pieces were fired to various temperatures and cooled, the test being made at room temperature, that is, the abscissas on the curve represent maximum firing temperature of the sample while the ordinates are the porosity of the sample after firing and cooling. A discussion of this characteristic has been given.(1)
TESTS AND INVESTIGATION OF THE RATIONAL INSULATOR AS DEVELOPED
As mentioned before, the impulse ratio of a string of insulators is of very great importance and with an insulator of rational design the impulse ratio should be high. The curves in Fig. 5 are very interesting in this connection and show the excellent results secured by this rational design in impulse ratio in the individual unit and strings. The unit as designed will, therefore, for a given number of units in the string, have a very much higher flash-over to lightning disturbance than the conventional type unit with a lower impulse ratio. This, combined with the large mass of porcelain, thick petticoats and general substantial character, gives the unit a remarkable ability to withstand lightning conditions and their resultant power arcs.
The claim has been made that repeated mechanical stresses will weaken porcelain, and also that porcelain in tension is weakened electrically when under stress. In investigating this, units of the type described have been stressed to 9000 lb. (4100 kg.) in the tension machine and subjected to dry flash-over at 200,000 cycles while under stress. This test was continued until the units had been under stress for several days and at the application of high-frequency flash-over for as long periods as 100 hours there was no indication that this mechanical stressing affected in any way the dielectric quality of the insulator. This is not surprising as the plane of mechanical stress is normal to the plane of dielectric stress as before mentioned. To study the effect of repeated or continued mechanical stress, strings of insulators have been hung out in the weather with a dead load of 5000 lb. (2270 kg.) each, and periodic tests are being conducted to ascertain the condition of these insulators, and the results so far have not shown any indication that this fear is warranted. Further, repeated shocks and tension tests on porcelain samples under various conditions indicate that porcelain is not injured in any way by repeated stressing, unless the applied loads stress some part of the porcelain beyond its ultimate strength. If this is done, porcelain will fail quite naturally, as will cast iron or any other brittle material, but the results of practise and continued and careful laboratory tests indicate that the previous fear of fatigue due to continued working of the porcelain mechanically is ungrounded. These tests are being continued and some rather interesting reports will be made to the Society in the future as to the results secured from laboratory and practical tests of this nature.
FACTORY TESTS
It is doubtful if any developed tests at the present time that can be applied to an insulator without injuring it will prophesy its operation when on the line as to depreciation and for that reason we have made a rather radical departure in some respects from ordinary factory testing.
The fuchsine method of testing for porosity(2) is used in our factory, one unglazed unit being placed in each car of 70 insulators that pass through the tunnel kiln. This unit is broken up immediately on the removal of the car from the kiln and the pieces subjected to the fuchsine test. On the slightest penetration the entire carload is rejected and scrapped. This test gives a very satisfactory control test for porosity and guards effectively against any errors in raw materials or reading of the pyrometers that might cause porosity through firing outside of the permissible range as indicated in Fig. 8. It insures the scrapping of the small number of porous units inevitable in quantity manufacture of porcelain. After the inspection of fired porcelain the hardware is assembled and each unit is subjected to a 5000-lb. (2270-kg.) mechanical load. After this load each unit receives a two-minute dry flash-over at 200,000 cycles. The units are then again inspected and turned over to the assembly department. It is believed that the best insurance the customer can be given as to the quality of the product he is buying is that the manufacturer started with a rationally designed product correctly proportioned and manufactured to fulfill the required conditions. A certain percentage of the finished product should then be tested to destruction to determine that the required standards of manufacture are being maintained. To this end a certain percentage of the product delivered to the shipping department is selected at random by the research laboratory and tested to destruction. The plotting of the data secured from these tests establishes a probability curve for the product considered, the study of which has revealed some very interesting things regarding manufacturing limitations.
Furthermore, when this probability curve is once established with accuracy any test falling outside the determined limits on this curve is a danger signal and further tests are at once made. If these tests confirm the first results an immediate investigation is made to determine wherein the factory processes are not maintaining the required standards. It is believed that this method of testing is better than the imposition of very severe acceptance tests on all the units, as such tests, unless carried to destruction, tell very little regarding the future performance of the insulator.
OPERATING CHARACTERISTICS AND SELECTION OF UNITS
The selection of proper insulator strings for any given transmission line involves a rather careful study of a good many conditions. Given a rationally designed insulator, the individual characteristics of which are accurately known, the selection of the number of units for a string for given conditions is a matter of the development of the proper safety factor for the right conditions, using as a basis the worst line conditions liable to arise. In this connection the curves of Fig. 9 are of considerable interest. Wet flash-over values are rather deceptive. Due to the leakage currents, the distribution of the voltage amongst the units of a string is very much improved. Furthermore, wet flash-over values are apt to be erratic unless conditions are very carefully controlled as to the purity of the water used, precipitation, size of spray, etc. In general, the selection of the proper strings involves a determination of the worst electrical conditions likely to be met on the line and the selection from characteristic curves of the insulator of a string that will give the desired safety factor under these worst conditions. The curve 3 of Fig. 9 is very interesting as a measure of protection afforded to disturbance of lightning frequency. In Fig. 10 some rather interesting data are given, that is too often ignored in the selection of insulators for a transmission line, especially when it is remembered that an increase of 45 deg. cent. (110 deg. fahr.) in temperature is equivalent to an elevation of 3000 ft. (914.4 m.) of the line. After a preliminary selection of a string has been made, a study should be made of the duty on the conductor end unit to determine whether or not from the characteristics of the insulator this unit is working within safe limits. If not, the string should be readjusted to operate this unit under proper conditions, and then the results examined on the basis of the margin of safety afforded on the failure of one unit. The readjusted values of the voltage on the different units and the resulting safety factors will give a very good idea of the advisability of further insurance against trouble by the addition of end units. This question is, of course, an economic one, and the amount of money that it is permissible to pay for such protection is a question that each engineer must decide for himself.
FUTURE PROGRESS
The insulator situation today is in a state of constant development and considerable progress may be expected in the near future. Certain recent investigations indicate' that piezo-electric effects may be of considerable influence in porcelain depreciation and recent developments indicate that this situation will soon be met in a very satisfactory manner. Also some rather interesting work is being done at present on the solubility of porcelain in water under the conditions existing in the capillary passages connecting the voids of porous porcelain. Investigations are under way using pressures around 10,000 lb. per sq. in. with very high and very low temperatures to accelerate this action and, by means of the microscope, determine from samples of porous porcelain that have depreciated in the field compared with the porcelain subjected to accelerated tests in this manner, to what extent this solubility may be responsible for increasing porosity. The problem of very high-voltage transmission systems is being studied and some new types of insulators made up of rather special porcelain bodies are being developed that will meet this situation without difficulty and by the time there is money available to build any of the large projected extremely high-voltage lines, insulator manufacturers will be ready to meet the problem.
To be presented at the Pacific Coast Convention, Portland, Ore., July 21-23, 1920.
(1) "High-Tension Insulator Porcelain," W. D. A. Peaslee, A. I. E. E. White Sulphur Springs, June 29, 1920. "Test of Electrical Porcelain in Factory & Laboratory", W. D. A. Peaslee, American Society for Testing Materials, Asbury Park, N. J., June 22d-25th, 1920.
2. American Society for Testing Materials, Asbury Park, N.J., June 22d-25th, 1920.
3. W. D. A. Peaslee, "High-Tension Insulator Porcelain," A. I. E. E. White Sulphur Springs, June 29, 1920.
