1991 >> November >> Porcelain Insulator News  

Porcelain Insulator News
by Elton Gish, NIA #41

Reprinted from "Crown Jewels of the Wire", August 1991, page 7

DESIGN OF HIGH-VOLTAGE INSULATORS IN 1908
(from lily-shell multiparts to early suspension types)

Collecting multipart insulators has been very exciting for many collectors since my book Multipart Porcelain Insulators was published in 1988. Now, over 100 of you have a copy of the book. Many of you are enjoying the discovery of "new" specimens abandoned long ago along power lines, substations, quarries, etc. (not unlike the 1960's for the glass part of our hobby), and the benefits of trading specimens using the M-numbers. A few of you are keeping a watchful eye on a scattering of old classic units that have been in service since before 1910, anxiously waiting for them to come down. After all, a multipart has never been stolen from an active 40,000-60,000 volt line. In my book, we discussed briefly the evolution of early multipart insulator design and how the beautiful lily-shell styles met with early failure.

Here is a little more information that I found in the February 8, 1908 "Electrical Review". The article was written by Walter T. Goddard of the Locke Insulator Manufacturing Co. and read before the Electrical Section of the Canadian Society of Civil Engineers on December 19, 1907. I will just pull out a few of the highlights rather than print the entire article, and include other information as well about early suspension types.

As a general rule [according to Goddard], two-piece insulators are used for 10,000 to 30,000 volts, three-piece insulators for 30,000 to 50,000 volts, and four-piece insulators for 66,000 to 70,000 volts. Above 70,000 volts, the underhung or suspension insulator is used with 4-10 shells of porcelain depending on voltage.

In designing multipart insulators for 1908, the first consideration was puncture strength. This was a critical error which I will explain later. Goddard recommended that porcelain shells not be made more than 5/8" thick. It was not possible at that time to reliably produce porcelain of greater thickness without internal flaws.

The second design consideration was to provide as much dry surface as possible under average rain storm conditions. This was the primary purpose of the lily-shell design. The upward curving shells would extend the electrical leakage length between the conductor cable and the pin, and the tall design would allow long, dry areas between the pin and the inner shell. In some designs, there was almost 12" of narrow, dry area between the bottom skirt and the pin. The M-3740's found by Barry Conolly (see April, 1991 CJ) had nearly this entire area next to the pin filled with Portland cement!

As opposed to straight sides (Fig. 8) where rain bounces away, the lily-shell design (Fig. 6), bounces rain water upward wetting the underside of the upper skirts. This action wets nearly the entire leakage surface of the insulator except for the surface of the inner shell next to the pin. Consequently, the inner shell is then subjected to almost the entire line voltage. It is no wonder why insulators such as M-3250, M-3740, and M-3890 were easily punctured. Remember that voltage is considered a force. During times of voltage surges from lightning, the electrical current often would find it easier to pierce one or more of the thin porcelain shells rather than travel the distance necessary to flash over or arc to the pin.

The single factor contributing to electrical puncture of these early designs was the height being greater than the diameter. The height of the insulator and outward flaring skirts provide a very long path for the electrical current to travel to reach the pin; thus, it inhibited the electrical arcing from crown to skirt edges to pin. During voltage surges from nearby lightning strikes, it is easier for the electrical current to force its way through 1, 2, or 3 thin porcelain shells resulting in electrical puncture. Puncture is the result of an electrical imbalance across the insulator leaving the inner shell to withstand nearly the full voltage surge. Puncture is usually evident by a permanent, tiny hole, and, in severe cases, the puncture hole was through the crown directly to the top of the pin causing the circuit to short to ground. Punctured insulators are impossible to detect from the ground, so, in severe cases, each unit had to be removed and tested. Also, voltage surges, as well as normal heating of the insulator from current leakage (or puncture), would cause the cement to expand and break the shells. Lightning arcs, too, could be very damaging to the thin porcelain skirts, often shattering them. 

The obvious solution [according to Goddard] to prevent puncture was to add a fourth recessed shell inside the insulator as illustrated in Fig. 7. This did reduce the potential for electrical puncture. However, it added another layer of cement. As the insulator heated up from the sun and/or electrical leakage and corona effects, the thick layers of cement would expand laterally and crack the shells. In some cases, the shells would crack so severely that they would break apart. By 1908, this was already becoming a major problem with three-part lily-shell designs such as M-3250, M-3740, and most notably with M-3890.


M-3890's damaged by lightning in 1908

The massive failure of M-3890's on the 160-mile, 60,000 volt Niagara, Lockport and Ontario line from 1906-1908 due to lightning strikes (see the photograph above) brought about a new realization of how to design multipart insulators. When 11,000 insulators were removed and tested, it was found that 40% of them had puncture damage of one or more shells. 82% of the puncture failures involved the inner shell with almost 56% of the failures involving just the inner shell. It was quite obvious that the inner shell was receiving the majority of the electrical stress from voltage surges. The M-3890' s that passed the test were placed on the bottom crossarm. The single, top insulator, which position received the most damage from lightning, was replaced with M-4338 early in 1909.

M-4338 was specifically designed, in the light of experience on this line, to be six inches shorter. They found that long, thin insulator parts, as in M-3890, could not resist mechanical shock of lightning. Moreover, they discovered that short shells would allow a more even distribution of voltage across the entire body of the insulator; therefore, reducing the electrical stress on the inner shell, and reducing the chance of puncture. The shells were made thicker, too, to resist breaking, and a fourth, recessed inner shell was added for additional protection against puncture; however, it probably was not needed. In short, the new four-part insulator was designed to flash over rather than to puncture. This was the first realization of the benefit to an insulator design with a diameter to height ratio of less than 1: 1.

The new designs incorporating the fourth, recessed inner shell and the diameter to height ratio of less than 1: 1 are M-4325C, M-4332, M-4338, M-4339, M-4415, M-4415A, M-4600, M-4605, M-4606, M-4608, and M-4800. We do know that M-4325C, M-4415, M-4415A made by Thomas and M-4600 made by Lima were probably made before 1908 (a few of these are still in use in California). These three styles are similar in that they all have ridges under the top skirt (more on that feature in a future PIN article). I find it puzzling that M-4338 was developed in 1908 on the east coast to meet a specific need, but other insulator companies, too, picked up on the design change to supply insulators to the west coast.

It should be noted that M-4338 is not the grey Ohio Brass crosstop (without recessed fourth shell) that several people have in their collection (It was assigned M-4395). Specimens of M-4338 have not been found. 

The benefits of the new, shorter designs were slow to become widely excepted. Along with many lily-shell styles shown in the 1917 Parker catalog which still exhibited long, thin shells, there was the outdated M-4390. This insulator still had a height to diameter ratio that was greater than 1. The use of the recessed fourth skirt persisted as late as 1939 with the Westinghouse M-4340. This was probably still a good idea since the cement expansion problem had been overcome in the 1920's.


M-3890

L. C. Nicholson, the electrical engineer working on the Niagara, Lockport and Ontario line, sought to remedy the problem of severely damaging lightning arcs on M-3890 by installing an arcing ring around the bottom of the insulator. This provided some relief by allowing voltage surges to arc around the insulator to the arcing ring, thus protecting the thin skirts from both breakage by the arc and from electrical puncture. The rings also reduced the burning damage caused by the arc on the surface of the insulator. The photo on the previous page shows M-3890 with a "Nicholson arcing ring". It probably was an expensive fix. Even when damaged units were replaced with M-4338, they were also fitted with Nicholson arcing rings (photo below). So, you can see that they still did not yet fully comprehend what was happening. With massive failures of M-3890, it was time for drastic action with little time for experimentation to better understand the problem. The rings would later be called "grading rings" and were used to equalize or grade the voltage over the suspension string. 


M-4338 fitted with Nicholson arcing rings in 1909

Nicholson obtained three patents covering the arcing ring design. Two were for multipart insulators and one for suspension insulators.

Goddard stated that the straight sided shells on the insulator shown in Fig. 8 would probably stay dry longer during a rain, but that the lily-shell design would provide a greater leakage path and offer greater electrical resistance. We now know that the leakage path should not be the entire consideration.

In an article in "Electrical World" dated December 22, 1909 titled "Present Status of European Practice in Transmission Line Work", it was believed in Switzerland that rounded skirt edges would cause the rain water to run under and drop on the surface of the lower skirts.

They favored a slightly upward curved skirt edge that would allow the water to drop off to a sharp point missing the skirt below. This idea may have some merit if it were not for the other problems with lily-shell styles. The Swiss attempted to solve the cement problem by using a mixture of sulfur and plaster which was made non-porous by coating it with shellac. Other Swiss experts had good results with screwing two or more shells together and then re-firing to fuse the pieces into a single unit. At least one 35,000 volt line in Germany utilized insulators made in one piece to eliminate the problems with cementing!

Pin type insulators have a practical limitation. Insulators with more than four shells were never shown to be practical. The only reference that I have found about a five-part insulator was in Brent Mills' book, Porcelain Insulators and How They Grew. It was rated at a working voltage of 120,000 volts and shown in the 1920 Pittsburg catalog! I wish I could find a copy of that catalog. Knowing Pittsburg's penchant for cataloging a multitude of odd, difficult to make styles, I seriously doubt that they ever made a five-part insulator [Pittsburg also cataloged a porcelain CD 147 in their catalog No.2 circa 1918!]. 

Adding additional shells compounded the cement expansion problem as well as making the insulators much too heavy to handle and mount on a tower. What would they have used as a pin for a five-part multi?! M-3490 already used a wooden pin that was almost as big as a baseball bat [Bill Rohde retrieved several of these heavy pins]. Therefore, the suspension insulator was the next logical step for transmission of voltages greater than 60,000 volts. They offered the advantage of size, cheapness of manufacture, and, most notable, by simply adding additional units to the string, you could easily increase line voltage. Suspension insulators allowed transmission of power at 70,000 to 1,000,000 volts, far surpassing the pin type practical limitation of 70,000 volts. Below 70,000 volts, pin types were cheaper. Also, if one of the suspension units was damaged, it would only slightly reduce the effectiveness of the entire string. It could be easily replaced at leisure without experiencing severe or total line failure. However, early suspension designs were rooted in thinking about problems with pin type insulators; namely, puncture resistance. Goddard and others believed that one should first design the suspension insulator against puncture and then leakage distance. 

To this end, Fig 13, with its' two-part suspension design, was the recommendation. Again, cementing of the two porcelain shells was a major problem. The cement expansion would soon crack the porcelain shells and result in insulator failure or worse, dropping of the conductor! These suspension units were made up to 14" in diameter! They were quite heavy and problems were experienced with both cement expansion, and cemented metal pins and hooks, pulling out of the cement. The mechanical strength of these early units became inferior after a short time in operation.

These early two-piece suspension insulators must now be very scarce. Ben Kirsten has found a specimen of a Thomas 14" two-piece suspension unit (see Ben's drawing on the next page). He says that it has a red-brown glaze color, and it weighs 23 pounds! A string of 6 of these units would weigh 138 pounds! A string of 6 modern 10" suspension disks would only weigh 63 pounds. 

The Duncan Two-Piece Suspension

 

This drawing was found in American Hy droelectric Practice published in 1917. This is similar to Ben Kirsten's insulator. The caption said that it was made by Thomas and used for 110, 000 volts on lines of the Southern Power Co. and those of the Georgia Railway and Power Co.

The two-piece suspension unit was patented by John Duncan on February 25, 1908 (see copy of patent drawing and photographs below). Duncan licensed his patent to the Locke Insulator Mfg. Co. as evidenced by the above two photographs taken from the Journal of Electricity, Power and Gas dated October 5, 1907, less than one month after the patent application was filed. Three sizes of the two-piece suspension units were shown in the 1913 Locke catalog (see end of this article) and two sizes in the 1916 Locke catalog. A 1920 Thomas catalog is the earliest one that I have, and no two-piece suspensions were shown. Also included at the end of this article is a photograph of several types of suspension insulators which appeared in the AIEE journal dated December 13, 1912 as part of an article by P. W. Sothman on testing suspension insulators. Suspension disks similar to Ben Kirsten's Thomas insulator is "Type D" in the photograph. Sothman's conclusions on testing "Type D" were:

"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 failure to be condemned, causing distortion of [electrical] field and premature discharge. As a two-piece insulator, electrical stresses of petticoats are not balanced."

The hook in those tests would pull out due to cement failure under the load of the conductor, but the unit was stronger than most of the other suspension types tested. The two-piece style was not very practical because the metal cap had to be larger to accommodate the extra thickness of two shells; thus, it was heavier and bulkier.

In 1908, Goddard and others correctly believed that an insulator should not be regarded as a barrier to current, but rather as a high-resistance path between the line and earth. By providing an insulator with additional shells and using lily-shell skirts, they believed that the high-resistance (and puncture) requirement was met. 

The electrical fields of high voltage transmission lines and their effects on insulators did not reach an adequate level of understanding until the late 1910' s. Up until this time, many engineers involved in high voltage insulator design were wandering aimlessly. Designs reflected the whims and wants of each purchasing agent or power company specifications (again reference the multipart styles shown in the 1917 Parker catalog in Brent Mills book). It was the Faradoid patent (1,374,998) granted to Fortescue and Gilchrist on April 19 , 1921 that educated the insulator industry to what was really causing insulator failures, and what could be done to provide a genuine improvement in multipart design.

The development of the suspension disk insulator was delayed many years until reliable cementing techniques could be developed for the cap and pin assembly. During this development period, the power transmission industry was forced to rely on the inferior Hewlett suspension disk. The Hewletts were manufactured originally by General Electric (holders of the patent). Later, due to demand out-stripping General Electric's production, the manufacture of the Hewlett was licensed to Locke and Thomas. This story will have to wait until another time. 

These suspension styles are from the 1913 Locke catalog. Multipart styles (below) are "new" M-numbers that are not shown in my book, Multipart Porcelain Insulators, but are referenced in this article.



Medium Image (79 Kb)
Large Image (207 Kb)

[From AlEE Journal dated December 13, 1912]

"Type D" is the Duncan two-piece suspension insulator of a more modern curved style. "Type A" is the Hewlett. "Type B" is the new "fishtail" Hewlett for strain service. "Type C" is an early one-piece suspension. "Type E" is a modern style of suspension disk with petticoats. "Type F" is of foreign manufacture which used plaster rather than Portland cement. to attach the metal pin and cap. This unit did not test well because of the poor holding power of the plaster.



Medium Image (109 Kb)
Large Image (214 Kb)




| Magazine Home | Search the Archives |