Publication: Electrical World and Engineer
New York, NY, United States
The Composition of Porcelains for Electrical Purposes.
The current issue of the Transactions of the American Ceramic Society contains a paper by Mr. Arthur S. Watts, of Findlay, Ohio, giving the results of exhaustive tests on electrical insulating porcelain. The tests, which extended over several months, were extremely complete, covering almost the entire range of earth insulating material. The paper includes numerous tables, giving full data of the tests. Below we give an abstract of the more general parts of the paper:
The leading materials used in porcelain manufacture are certain kinds of kaolin, which are very slightly plastic, quartz and feldspar. Of course ordinary feldspar displays basic properties in the fire, uniting with silica to form silicates, thus producing an intimate mixture of all silicates. Therefore, the contents of kaolin must be in a definite ratio to the quartz and feldspar in regard to the insulating properties.
Too much quartz makes the body too little plastic, thus making it difficult to work, and after burning, the ware shows impressions and some warping due to the tendency of the clay substance to the contract. Too high a content of feldspar makes the body soft and liquid, and increases the tendency to crack. A body low in feldspar does not produce sufficient vitrification, and hence decreases its insulating efficiency.
A porcelain body may shrink 15 to 18 per cent., and it is very important that the exact degree of shrinkage be known. No crazing or shivering is allowable in the glaze used on ware for electrical materials. If the glaze is too infusible the ware will be dull, and if too fusible it is likely to be covered with blisters. Investigation has proven the condemnation of glazes containing metallic oxides to be unfair. Metallic oxides, when used in glazes combine with the silica, forming silicates.
Probably in no line of electrical investigation have greater progress and more varied experiments been made than in the shape and size of electrical insulators. The plain bell which is used for low potentials, and which is also practical and satisfactory for alternating currents up to 1,000 volts, does not meet the requirements of modern high voltage transmissions.
The first step in the way of improvement was made in 1857, by Borggreve. The product was satisfactory so far as electrical and mechanical requirements were concerned, but failed owing to the fact that the interior space was square, which gave rise to uneven tensions in the body and frequent fractures.
The second step in the way of improvement was the production of a very long insulator having a very narrow cross-section. It was the idea of the inventor to increase the resistance by means of a long arcing distance. The insulator produced was of clear glass and suffered from the contraction and expansion of the iron bolt used to connect it to the cross-arm, and which became intensely hot from the sun's rays on the insulator. This condition caused the introduction, in 1895, of the double bell insulator of porcelain. In this, the interior cylindrical part of the insulator is covered by a second bell, which prevents the radiation of heat and also the formation of dew.
Up to a tension of 3,000 volts, the ordinary double bell, 75 mm. wide and 100 mm. high, suffices, but above this voltage the question of further improvement arises.
In 1869 Lenoir and Prudhomme obtained a patent for the use of oil in insulator bells. Insulators of this type were used in constructing the famous line between Lauffen and Frankfurt. They proved impractical, as the oil must be replaced frequently, and in case of rain it was often replaced by water.
In constructing the Kaiser Wilhelm Canal Line, a triple-bell insulator, without oil, was used, and proved satisfactory even when exposed to salt spray. This type is now commonly in use for high voltage transmissions.
At first the insulators were glazed and burned, resting on the bottom edge of the bell, but the roughness of this edge seemed to cause the collection of drops of water and to prevent this, the insulator is now glazed all over, the glaze removed from the head and the insulator burned inverted. Very recently it has been discovered that by glazing the inside of the insulator in the threaded hole, the insulator is made much more resistant and need not be so thick.
The weight of the porcelain used in an insulator is not of the greatest importance, but rather, the quality of the porcelain and the kind of glaze.
In Europe, insulators are made by all five of the clay working processes, viz., turning, jollying, pressing, casting and dry pressing. Dr. Seger, in his work, shows very conclusively that a semi-dry process, in which the material to be pressed contains the highest per cent, of water allowing of good pressing, gives the densest and strongest body.
Turning of large insulators is done similarly to all clay turning, the product being made mouth up, and the top finished when the ware is sufficiently dry to permit of its being inverted. In this class of ware, much trouble is experienced in making the screw-thread for the pin which is to support the insulator. This thread is made by means of a threaded steel tube, which is provided with a handle on one end, and which is slowly turned in and out. In place of a full steel tube a segment may be used. Automatic boring apparatus is now being applied for this work.
Jollying is now largely replacing turning although it does not insure as dense and uniform a body as the former process. Pressing is done in molds. The clay is placed in the mold in a lump and by means of a plunger is forced to assume the shape of the inside of the mold for its exterior while the plunger shapes its interior.
Dry pressing produces more exact shapes than any other method. The molds used are of steel and are very expensive. The body must be in such shape that it will not stick to the mold. To obtain it in this condition some factories use the following process: The clay after drying and being thoroughly pulverized is placed in wet cement or plaster boxes and allowed to absorb 10 to 20 per cent, of moisture. The mass is then sieved thoroughly. Ten kilograms of the body will contain about 1 to 2 litres of water. Of course this proportion must be maintained exactly in order to insure a uniform shrinkage. The molds are sometimes complicated and are often composed of insertions. The piece of ware produced is threaded by means of a screw inserted in the bottom before the clay is introduced, and which is screwed out rapidly by means of a crank. The tendency is to make insulators without threads, as this has always been a great source of trouble and loss in manufacture.
Mr. Watts said that in outlining his tests, it seemed necessary to choose some type or standard of porcelain which could be used as a basis from which to map out varying series. The wisest course seemed to be to procure a large number of standard porcelains, and to obtain from these by comparison the most characteristic type of porcelain now in use. Acting on Professor Binn's suggestion,* Al2 O3 was taken as unity in the formulas.
An examination of the chemical formulas of 36 specimens at first sight appeared to indicate a hopeless conglomeration, ranging to all extremes; but closer study disclosed the fact that, while the RO varies from .0467 in the Japanese to 5.43 in the old English, and from .114 in modern Belgium to 1.84 in modern English ware, the proportion of SiO2 is comparatively constant. Only in two out of 36 cases does the SiO2 run below 3.7, and in only two cases of modern porcelain does it run over 6.0 SiO2. The average of the list was between 4 SiO2 and 4.4 SiO2, hence 4.2 SiO2 was taken as the standard.
Next a standard for RO was chosen. To obtain this by the same method as pursued with the SiO2 would prove unsatisfactory, since to take an average of the entire list would result in a RO of 1.00, while to take an average of the 30 modern porcelains would give about 0.3 RO. The first case would give about a cone 6 product, while the second case gives very nearly the type of an American porcelain, maturing at cone 12. As a choice between such extremes, could not be satisfactory, a series was made, covering each extreme and a third about half-way between, as follows:
Series A 1. RO, 1. Al2O3, 4.2 SiO2 Cone 6 type.
Series B. 0.5 RO. 1. Al2O3, 4.2 SiO2 Cone 9 type.
Series C. 0.3 RO, 1. Al203, 4.2 SiO2 Cone 12 type.
The various porcelain trial bodies were produced in the following manner:
The two extremes of each set were weighed out and ground wet in a ball mill. The intermediate numbers were obtained by liquid blending of the extremes. After the blends were thoroughly mixed, they were poured into plaster molds to harden by loss of their water. Of each porcelain body the following ware was made:
Two brickettes made in standard cement—brickette molds—and having a central cross-section of 1 inch in the green state. These were made by the plastic process. Two cubes of 50 grams weight, made also by the plastic process. Two tiles, 4 1/2 inches x 4 1/2 inches, in green state. These were made by the dry-press process.
The material for the dry-pressed tile was prepared as follows: The dry body was thoroughly pulverized and was then dampened with water by means of a brush and the hand, thus obtaining a fairly uniform dampness. The damp clay was allowed to stand about two hours, and was then put through a 20-mesh screen and promptly pressed into the desired shape by means of a hand tile press. About 12 to 14 per cent, of water was in the clay w-hen it was pressed.
Upon the various forms made up, there were made an electrical resistance test, tensile test, abrasion test, absorption test and shrinkage test.
While no great obstacle lay in the way of the tensile, abrasion, absorption or shrinkage tests, the difficulty was great with respect to testing the electrical resisting properties of the porcelains. In the first place, it was not known how much electrical resistance a given thickness of porcelain might display. In the second place there could be pressed only a tile of one size, viz., 4 1/2 inches x 4 1/2 inches. It was thus necessary to obtain, if possible, some insulating material in which to imbed the tile, and thus increase the arcing distance of the current, since with 4 1/2 inches arcing distance 55,000 volts could not be used without the current arcing around the tile. The idea of using Portland cement, however, suggested itself, as it seemed plausible that this might be a good insulating material.
A semi-vitrified tile, 4 x 4 x 3/8 inches, was incased in a slab of Portland cement (Dyckerhoff). The cement was carefully tamped around the tile, so that a good attachment might result. Several other slabs were also made of cement, both German and American. In these no tiles were encased. All the cement slabs stood six days in water, and were then taken out and carefully dried. Their resisting properties were then tested with poor results. Each slab showed only so much electrical resistance as a like thickness of pure air would show, thus proving that the small insulating power displayed by the cement was due to the fact that its thickness held the electrodes apart and made an arc necessary.
Next the tile was tested encased in cement. The following insulating materials had also been provided: One sheet window glass, one sheet plate glass, one porcelain tea plate, one tile 4 x 4 x 3/8 inches unincased. The results of the test of the above was as follows:
Tile 4 x 4 x 3/8 inches, incased in slab of Portland cement, punctured at 17,000 volts.
Tile 4 x 4 x 3/8 inches, unincased, punctured 20,250 volts
Sheet window glass 6 x 6 x 1/8 inches, punctured at.... 40,000 volts
Porcelain Tea Plate, 6 inches in diameter x 1/8 inch, punctured at 49,500 volts
Sheet Plate Glass, 6 x 6 x 1/4 inches, stood 60,000 volts, and then the current arced around sheet and could not puncture it.
The tile incased in cement and the unincased tile were both of the same composition, and as both were porous to a degree, it is to be presumed that the incased tile which had been soaked for six days still retained some moisture when tested, which accounts for itŁ lower puncturing point than was displayed by the unincased tile. The facts brought out by this preliminary test seem about as follows:
First.—That Portland cement does not possess any more insulating power than would be displayed by a like thickness of dry air.
Second.—That porcelain, if not perfectly vitrified, does not posses; any great insulating efficiency, even when perfectly dry.
Third.—That glass ranks next and lies between semi-vitrified and vitrified porcelain.
Fourth.—That thoroughly vitrified porcelain, even though only 1/8-inch thick, possesses as much insulating strength as is necessary, since 40,000 volts is the highest voltage that can be transmitted without arcing around a 4 1/2 x 4 1/2 inch tile.
Fifth.—That plate glass possesses an especially high electrical resisting property.
We know that the composition of glasses has a great influence on their ability to resist electric currents. Also, that the temperature at which they are used is a factor.
The search for an aproning material to use with the tiles was continued, and it was found that a gutta-percha plate could not be cemented to the tile for each test and then removed, and to furnish a separate sheet of gutta-percha for each tile would mean entirely too great an expense. Selenium also was out of the question owing to its cost. It was concluded, however, to try sulphur, which is a very similar material to selenium, and to do this two bottomless flasks, by means of sulphur in the melted state, were cemented to two tiles. Two large plates of sulphur were also cast with tiles imbedded in their centers. One tile was completely imbedded, leaving only a hole 1/2 inch in diameter on each side to admit the electrode. The second tile was only imbedded on one side in the sulphur, the other side was left exposed. There was a 1/2-inch hole on the imbedded side of the tile to admit the electrode. The tiles were vitrified. A plate of sulphur, 6 x 6 x 1/2 inches, was also cast. The results of the electrical test on sulphur was as follows:
The plate containing the half imbedded slab stood about 58,000 volts, and then the current arced through the sulphur. The plate containing the completely imbedded tile stood about the same, the current going around the tile through the sulphur. The tiles cemented to the flasks were not satisfactory since the current found its way between the glass and the sulphur. The slab of pure sulphur 1/2-inch thick stood about 13,500 volts and then punctured. On its being punctured the slab caught fire from the spark.
The results showing that at least none of the materials available could be successfully used as an aproning material, the idea was discarded of aproning the tiles, and the thickness of the pieces to be tested electrically was reduced to an amount which would puncture before the current would arc around the tile.
The formulas of the three porcelains selected as fairly illustrating the types now in actual use have been given. Taking each type, as a basis, three separate sub-series of bodies were made, as follows, each sub-series consisting of a set of five porcelains:
I. Maintaining RO and SiO2 constant and varying the proportion of AW*.
II. Maintaining the Al2O3 to SiO2 constant and varying the RO in kind but not in quality.
III. Maintaining the RO and Al2O3 constant and varying the proportion of SiO2.
PREPARATION OF THE REGULAR TRIAL PIECES.
The preliminary tests having been completed, the various bodies were prepared on sufficient scale, and were worked into the shapes heretofore decided upon, and after drying each lot thoroughly, they were fired. It was impossible to obtain in the small kiln used an absolutely uniform temperature over the entire kiln, but except in one case, all received an equal period of firing.
The paper discusses in detail the results of each set of tests, consisting of tests for electrical resistance, abrasion, absorption, shrinkage and tensile strength. Following are the conclusions arrived at:
The object of the investigation was to prove, if possible, the proportions of the essential constituents which go to make up the best insulating porcelains. The tests showed that porcelains of equally good appearance may possess greatly varying values when subjected to various tests. The essential properties of an electrical insulator are, that it possess, first of all, a high efficiency as a non-conductor; second, that it stand, without snapping, any reasonable strain put upon it by the exigencies of wiring; third, that it withstand the blows of missiles, even bullets, without breaking. The electrical, the tensile and the abrasion tests, respectively, were made to show these properties in the porcelains tested.
The electrical resistance tests show that every specimen standing a superior test has a high silica and moderately low alumina content. Unfortunately in each series showing the variation in proportions of CaO and K2O, the content of Al2O3 was so high that no exceptionally good porcelains were obtained, even with high K2O content. However, the tables show that a marked improvement is apparent upon increasing the K2O and decreasing the CaO, except at the extreme of the series where we see the K20 used alone, and here the specimens show weakness due to brittleness. This indicates that had the K20 been somewhat increased, and the CaO proportionately decreased, the result would have been much better.
From a study of the abrasion tests, it is very apparent that with K2O alone as a flux, and even with a moderately high clay substance, we obtain a brittle and weak porcelain. When as low as 20 per cent of K2O is replaced by CaO, we find the product changed to a tough but well vitrified body, although the translucency is reduced by any additions of CaO. To just what extent this is the case is not a matter of importance in this work. In the series showing the effect of Al2O3 content, no great difference was noticed in the ability of the specimens to resist abrasion. Much the same statement would apply to the series showing the results of varying of SiO2. However, the medium high SiO2 specimens seem to clip worse than the extremely high SiO2 specimens, while the low SiO2 specimens show a la