RECENT DEVELOPMENTS OF T H E KJELLIN AND ROCHLING-RODENHAUSER ELECTRIC INDUCTION FURNACES. BY J. H A R D ~ N . ( A Paper read before the Faraday Society, June 23, 1908, Professor A. K. HUNTINGTON, Vice-President, in the Chair.) The object of this Paper is to discuss the rapid development of the electric induction furnace during the last two years. Before proceeding to discuss the practical improvements in detail, it will be interesting to touch upon one or two novel points of a purely physical nature. We will consider one which will be of general interest- THE (‘ PINCH ” EFFECT OF AN ELECTRIC CURRENT. Many people experimenting with electric arc and resistance furnaces have observed that when a current is passed through a bath of molten metal or electrolyte the current density cannot be raised to a very high limit without distorting the bath.If the current density is raised to a certain point, it is plainly seen that the metal carrying the current contracts violently, or, in other words, becomes pinched, the effect of which is that the metal on either side of the pinch swells up above the.average level of the bath. If the current is still increased, the pinch effect becomes so complete as to actually cause separation in the bath, thereby breaking the current circuit and causing a violent flash, together with rapid volatilisation. This is only a momentary effect, as the separate parts immediately flow together again, closing the circuit, when the same phenomenon repeats itself. Thus it was found impos- sible to pass an unlimited amount of power into the bath, although one might be well below the boiling point of the metal.The explanation of this pinch- ing is quite obvious. We have only to remember that conductors carrying current in the same direction have a mutual attraction for each other, and if we consider that the molten bath consists of an infinite number of elementary paths of current, it is natural that, as each one is attracting the paths in its immediate neighbourhood, there must be a tendency to compress the full section of the bath. If now, therefore, the bath is slightly contracted at any one spot, the current density will here be raised and the pressure will in- crease until finally the bath becomes so pinched that an actual break in the circuit may take place. Correspondents of the technical papers have discussed this question, and the point has been raised whether this phenomenon can also be observed in the induction furnace.Theoretically it is obvious that it should, for although the induced current is an alternating onc, it has for a given moment of time the same direction in each part of the path. We have taken observations under working conditions in a 60-kw. Kjellin furnace erected in London for experimental work. A small charge of pig- iron, consisting of about one-third of the full capacity of the furnace, was placed in the bath, and a current of 20 kw. was employed. As soon as the r 20 Of course, this is only momentary.RODENHAUSER ELECTRIC INDUCTION FURNACES 121 charge was fully liquid the pinching effect commenced, and the metal was seen to contract at a certain spot, raising the level of the metal on both sides of the pinch, which was sufficient to break the circuit, causing a flash, imme- diately after which the metal flowed together again, closing the circuit. The level of the metal immediately on each side of the pinched area rose about I+ in.above the normal level of the bath. It was found, on examination, that a small piece of slag was burnt into the bottom of the hearth at the point where this phenomenon occurred, thus causing the original reduction of area. Pieces of pig-iron were added to the bath, and as the depth increased the pinching slowly disappeared. It was found that if the original charge was about half of the full charge no pinching effect could be observed.Another striking feature is the resistance curve of the melt-that is to say, the curve obtained by taking the voltage at constant power across the terminals of the furnace during melting. Supposing we are starting the charge by means of a cold ring of welded or cast iron. Putting on a full load and keeping the kilowatts constant, we first find that the voltage rises above the normal for full load, and as the ring gets hotter the voltage still gradually rises until the ring becomes a bright red heat. From that point a decided drop in the voltage is noticed, although the power is kept constant and the weight of the charge is kept the same. As soon as the ring begins to melt the voltage again rises, but does not reach the same value as before the ring was red-hot.Now let us consider the cause of this result. The furnace is, as you know, nothing but a transformer with a short-circuited secondary. This latter must be placed some distance from the primary, owing to the thickness of the lining, cooling chamber, &c. We therefore have a certain amount of magnetic leakage, not only around the primary, but also round the secondary. The secondary is of iron, with comparatively high permeability, and we have therefore intro- duced an easy path for the lines of the stray field, hence the increased in- ductive voltage across the terminals. This is to a certain extent compensated for by the lower ohmic resistance of the ring ; but as the resistive coefficient of the latter is such as to increase with the temperature, the total resistance, measured across the terminals, will increase with the temperature, but the power factor will be lower during this period, which shows that it is not only the ohmic resistance that is increased, but also the inductive resistance.This is due to the fact that the permeability of the iron is also increased to a certain extent with the temperature; in fact, it rises very rapidly up to a temperature of about 840°C., when the permeability begins to drop very quickly, and reaches zero at about 920° C. In this interval between 840° and 920' (this figure will vary with various kinds of iron) the inductive resistance is rapidly decreasing, because the easy path for the stray field is checked, and the voltage across the terminals is consequently lowered. But in the meantime the ohmic resistance is steadily increasing, but more slowly than the change in the permeability ; therefore, the voltage will again rise, though slowly, until the temperature is reached at which the loss by radiation and the heat introduced balance each other.It may even increase somewhat above this point, owing to oxidation of the charge, but this increase is very slight. This is the explanation why some people were misled into believing that the increase of resistance in iron due to heat was not a straight line curve, (It may, perhaps, not be so, but the effect shown on the furnace terminal is certainly produced in the way stated-which can be proved by the watt- and voltmeter readings.) Now, turning our attention to the more practical side of our process, we recognise a new feature, which is likely to prove a distinct improvement, viz., the Rochling-Rodenhauser modification of the induction furnace.122 DEVELOPMENTS OF T H E KJELLIN AND ROCHLING- In the original Kjellin furnace some disadvantages are experienced when dealing with material which has to be refined and treated in very large quantities. For instance, when a charge of three tons or more is to be treated, the section of the b;tth becomes very large, thus causing a low resistance, whereby the power factor is lowered.If we try to increase the resistance by making the ring wider in diameter and of smaller section, the distance from the primary will be greater and the power factor again lower. Thus it becomes necessary to emp!oy a generator of very low periodicity for such furnaces, which is, of course, undesirable.Also the processes of de- sulphurisation and dephosphorisation are very tedious, as it is difficult to keep the slag sufficiently liquid for such purposes. Neverilzeless, this class of .furlaace will still hold its own, as it -forms an almost ideal crucible steel furnace. In the case of crucible steel we have seldom to deal with more than onc and a half to two tons at a time, and as it does not pay to use impure raw material, no refining is required, but plain melting and “ killing,” and for this class of work plain induction furnaces can be provided, which will answer very well at 15 to 25 cycles per second. But when it is desired to refine, say, a material smelted from inferior ores and decarburised in a converter, but still containing up to 0.1 to 0.2 per cent.sulphur and 0.05 or more phos- phorus, in quantities of five to seven tons, this plain induction furnace would not be so satisfactory. This was what was required for making rail and other similar steel at the Rochling’sche Iron and Steel Works at Volklingen, Germany, and therefore the engineers at these works, Mr. Rodenhauser and Dr. Schonawa, set them- selves to adapt the Kjellin furnace, and so arrived at the ‘( combined furnace.” This consists of a transformer furnace with two ring-shaped baths adjacent and communicating with one another, in the case of a single-phase furnace, and three such baths in the case of a three-phase furnace, with a square or rectangular hearth in the centre between the rings, with doors in front and behind, in exterior appearance very much like a Siemens open-hearth furnace ; but the principal feature is a heavy secondary winding of copper cables, placed around and co-axial with the primary (one on each leg of the core), sur- rounded by the rings forming the charge.These copper secondaries, consist- ing of a few turns only, are connected to conductive plates-they can hardly be called electrodes, for reasons given below-built into the furnace wall, two in front and two.at the back for a single-phase furnace. These plates consist of corrugated cast steel plates, and a compound of magnesite, dolomite, and tar is applied firmly over the corrugation. The plates do not conduct well when cold, but as soon as the furnace is charged with molten raw material they will act as a (( conductor of the second class,” and readily allow the current to pass.Thus about one-half of the power is transmitted to the charge by induction in the rings and the rest of the power through the side plates. As the copper secondary is placed very close to the primary, the leakfield is very much smaller ; in fact, three furnaces for one to one and a half tons ‘are ,now in operation with 50 periods at a power factor of 0.7 to 0.85, a result which could never be obtained with a plain induction furnace of a similar size, in spite of (( bifilar ” baths and other devices which have been tried. But fhis is not the chief advantage, as the same result may be obtained by other electrical means. A far more important gain is to be found in the metallurgical possibilities obtained with the new design. We know that for carrying out any refining process in steel we need a sufficiently liquid slag and ways and means of handling the same.This is to a certain extent obtained in some (( electrode furnaces,” where an arc plays between carbon blocks and the slag “blanket.” This, however, in some cases, has122 DEVELOPMENTS OF T H E KJELLIN AND ROCHLING- In the original Kjellin furnace some disadvantages are experienced when dealing with material which has to be refined and treated in very large quantities. For instance, when a charge of three tons or more is to be treated, the section of the b;tth becomes very large, thus causing a low resistance, whereby the power factor is lowered.If we try to increase the resistance by making the ring wider in diameter and of smaller section, the distance from the primary will be greater and the power factor again lower. Thus it becomes necessary to emp!oy a generator of very low periodicity for such furnaces, which is, of course, undesirable. Also the processes of de- sulphurisation and dephosphorisation are very tedious, as it is difficult to keep the slag sufficiently liquid for such purposes. Neverilzeless, this class of .furlaace will still hold its own, as it -forms an almost ideal crucible steel furnace. In the case of crucible steel we have seldom to deal with more than onc and a half to two tons at a time, and as it does not pay to use impure raw material, no refining is required, but plain melting and “ killing,” and for this class of work plain induction furnaces can be provided, which will answer very well at 15 to 25 cycles per second.But when it is desired to refine, say, a material smelted from inferior ores and decarburised in a converter, but still containing up to 0.1 to 0.2 per cent. sulphur and 0.05 or more phos- phorus, in quantities of five to seven tons, this plain induction furnace would not be so satisfactory. This was what was required for making rail and other similar steel at the Rochling’sche Iron and Steel Works at Volklingen, Germany, and therefore the engineers at these works, Mr. Rodenhauser and Dr. Schonawa, set them- selves to adapt the Kjellin furnace, and so arrived at the ‘( combined furnace.” This consists of a transformer furnace with two ring-shaped baths adjacent and communicating with one another, in the case of a single-phase furnace, and three such baths in the case of a three-phase furnace, with a square or rectangular hearth in the centre between the rings, with doors in front and behind, in exterior appearance very much like a Siemens open-hearth furnace ; but the principal feature is a heavy secondary winding of copper cables, placed around and co-axial with the primary (one on each leg of the core), sur- rounded by the rings forming the charge.These copper secondaries, consist- ing of a few turns only, are connected to conductive plates-they can hardly be called electrodes, for reasons given below-built into the furnace wall, two in front and two.at the back for a single-phase furnace.These plates consist of corrugated cast steel plates, and a compound of magnesite, dolomite, and tar is applied firmly over the corrugation. The plates do not conduct well when cold, but as soon as the furnace is charged with molten raw material they will act as a (( conductor of the second class,” and readily allow the current to pass. Thus about one-half of the power is transmitted to the charge by induction in the rings and the rest of the power through the side plates. As the copper secondary is placed very close to the primary, the leakfield is very much smaller ; in fact, three furnaces for one to one and a half tons ‘are ,now in operation with 50 periods at a power factor of 0.7 to 0.85, a result which could never be obtained with a plain induction furnace of a similar size, in spite of (( bifilar ” baths and other devices which have been tried.But fhis is not the chief advantage, as the same result may be obtained by other electrical means. A far more important gain is to be found in the metallurgical possibilities obtained with the new design. We know that for carrying out any refining process in steel we need a sufficiently liquid slag and ways and means of handling the same. This is to a certain extent obtained in some (( electrode furnaces,” where an arc plays between carbon blocks and the slag “blanket.” This, however, in some cases, hasFIG. 3.-The Rochling-Rodenhauser 3-phase InductionFurnace-1 Ton. 60-kw. Experimental Kjellin Furnace-London.RODENHAUSER ELECTRIC INDUCTION FURNACES 123 proved troublesome, the drawback being that the temperature must be extremely and unnecessarily high at the spot where the arcs are playing, which may not be without certain disadvantages to some steels.If we try to dip the carbons direct into the molten metal, we find that they are consumed at once, in such case contaminating the steel, which is, of course, difficult to avoid. The conducting side plates before mentioned are of quite a neutral nature ; in fact, some plates, which had been in constant use, day and night, for three months were so little corroded at the end of that period that the loss of the plates, calcu- lated per ton of steel, was hardly determinable. Part of the power is induced in the rings, thus heating the charge, and the rest passes through the side plates, to such an extent only as experience has proved to be necessary in order to obtain a sufficiently liquid slag.The ring-shaped part of the bath is covered with bricks, at a height below the level of the charge in the centre bath. Thus no slag can enter into the rings, and as it is the slag which is injurious to the lining, the rings need hardly any repair during a long run, whereas the rectangular bath in the middle is easily accessible, and can easily be patched out. The lining is simply calcined magnesite or dolomite, mixed with tar, and stamped in hot. It has been %id that the use of these steel side plates would be equal to a return to the old system of electrodes, with all their disadvantages, but it is evident from what has been stated above that this is not so, as practically no consumption whatever of these plates takes place, and they can hardly be called ‘‘ electrodes ” in that sense of the word.After the lining is stamped in, the tar is burnt out (either by heating a cast steel ring or pouring a small quantity of pig iron into the hearth), leaving behind a sintered mass, forming a solid brick of basic lining. The pig iron is teemed for treatment in the Bessemer converter, and a fresh charge is given, which is tapped direct from the converter. It is more economical to burn out the carbon and the silicon in the converter, before refining from phosphorus and sulphur. The larger furnace at Volklingen will take a charge of four tons. Calcined lime is added to form a suitable slag ; this slag sometimes also contains about 6 per cent.of magnesia. In case of need, a small quantity of fluorspar is also added, to act as a flux, but this is not always nwessary. Plate scale from the rolling-mill is added for decarborising. In this condition the slag will take up the phosphorus very readily, after which it is made more viscous by applying cold lime and drawn off through the slag door by a slight tilting of the furnace. It is essential for a successful dephosphorisation that the charge should be what is called “ hot brittle”--i.e., have an excess of oxygen, in order to prevent the phosphorus wandering back into the charge again. After removing the slag which contains phosphorus, ferrosilicon or carbon is added, forming SiO, or CO, thus depriving the charge of the oxygen.It has been found that the adding of ferrosilicon will shorten the time of the de-oxidation ; thus, if power is cheap, the cheaper carbon may be employed, and in the case of dearer power it is better to use the ferro- silicon. As soon as the dephosphorising is effeoted this first slag is entirely removed, and a fresh slag of lime only is formed, which, when the temperature is raised, acts as a desulphuriser in forming iron sulphide, But this is not so in the case of the combined furnace. Let us now follow up the progress of the operation.124 DEVELOPMENTS OF THE KJELLIN AND RuCHLING- The oxygen is also driven out in this operation, probably partly by combus- tion of the ferro-silicon (or carbon), whereby the temperature is increased, thus forming calcium carbide, and partly by adding a small quantity of other reactive agents.After this, the maximum power is applied, in order to drive out the last trace of oxygen, and as soon as no more gas bubbles are seen to leave the charge a test piece is taken out and forged. If too soft for the purpose, some coke powder is thrown in until the right proportions are arrived at. As a rule the operation is finished in one and a quarter to two hours, but if necessary the steel can, without disadvantage, be kept in the furnace for ten hours or more. It is thus possible to treat a material which contains up to 0.1 per cent. phosphorus and 0.1 per cent. sulphur or more so that a product containing 0*006 per cent. phosphorus or less and o*oz per cent. sulphur or less and from 0.5 to 0.1 per cent.manganese and 0.01 silicon will be obtained. As to the power consumption, if the furnace is charged with molten material from the converter, the consumption is from 125 to 150 kw. hours per ton of finished material. This, of course, depends upon the quality of the raw material, but 130 kw. hours may be taken as a good average for rail steel. The finished product is especially distinguished by its great strength, equality, and homogeneity. In fact, rails have been made with a much higher bending and breaking point than ordinary Bessemer or Thomas rails, and these rails command from 25s. to 45s. per ton more than ordinary rails, owing to their greater durability. There is no necessity to give a large quantity of figures as to strength, &c., of this material, but a few may be given below.ANALYSIS AND PROPERTIES OF ELECTRIC RAIL STEEL. Exhibited on the Meeting of the Furaduy Society. NO. - I 2 3 c. 0'55 0'50 0'55 I I 0'0941 Oq30 Si. S. P. 0.30 0.03 0.05 0'25 0.025 0.05 0.29 0.03 0.04 Tensile Strength. Ton, Square Inch. 53'2 52.0 52'5 Elongation Per Cent. 8 Inches. I 8.5 I 9.0 I 7-0 Contraction. Per Cent. 31'4 26.7 30.8 Low Carbon Electric Steel. 0*086/ 0*024( Traces/ 23.22 I 36.0 1 71.5RESULTS OF REFINING IN THE COhfBINED FURNACE. MECHANICAL PROPERTIES. CHARGING MATERIAL. No. I 2 3 4 5 6 7 8 9 I0 I1 12 13 I4 15 16 17 18 I9 20 21 - C. 3.400 P. 0048 0.048 0.064 0.060 0'077 0.042 0.098 0.057 0.07 I 0.048 0.049 0.070 0.060 0.085 0'045 0.043 0069 0.060 0.050 0.08 I 0.074 hln. 0.440 0'520 0.528 0.580 0.592 I *050 0.488 0.420 0.580 0'540 0.450 0'340 0.440 0.460 0'496 0.476 0.464 0.460 0.500 0.460 0.540 S 0.137 0.08 I 0.089 0084 0.089 0.097 0.073 0'105 0'057 0.08 I 0'073 0.105 0.065 0'073 0.057 0'12 I 0.078 0.081 0.08 I Si. 0'15 C. 0.093 0.070 0.152 0'144 0'334 0'397 0.690 0.892 0.928 0'939 0.972 I -04 1.05 0'201 0.854 I '00 0*080 0.620 0.750 1-170 0'2 I2 P. 0'022 rraces 0.025 0'0 I 8 0.024 0'0 I5 0.023 0.016 0'0 I 5 0.016 0.016 0.0 I 8 0.026 0'022 0'0 I2 0'01 I FINISHED h1ATERIAL. b1 n. 0.420 0.420 0.420 0'435 0'495 0.540 0.667 0'347 0.352 0'376 0.360 0.3 I 8 0.300 0'352 0.300 0330 Steel AIloj 0.01g 0.016 0.025 0'022 0'022 0.253 0'505 0.880 0,283 0.320 S . 0.024 0.032 0.024 0.026 0'035 0.024 0.016 Traces 0.016 0'0 14, Traces 0.016 Traces 0.016 0.01; 0'020 0.032 0'02 8 I'races 0.016 0.0 I 6 Si , 0'014 0.24 0.29 0.3 I 0.38 0'34 0.19 0.3 I 0.13 0.25 0.16 0.18 0'010 0'12 0'22 0'20 0.14 0'26 0'79 0.4 I 0.30 Cr. Ni. Tensile Strength. Tons, per Square Inch. 22.5 23-8 31'5 3 1.0 32.8 39'75 41'4 45'2 40'3 61.5 65.0 72.1 Elongation per cent. Total length 8 inches. 37'0 40.0 33'5 28.5 31.0 27'5 26.0 I 9.0 17'5 I 2.5 7'0 10'0 Contraction per cent. 70.2 69.3 62'3 51'5 53'2 41'3 32'9 48.5 46.4 23'5 49'2 23'4