comparative adjectives, it would be more to the point to describe two furnaces for comparison as being "twice the length," "one-half the depth of ore column," "onethird the width," "one-quarter the volume," etc.; these statements then being based on any dimensional component of the furnace volume: its electrical equipment, electrode sizes and areas and their relation to furnace crucible area, the rate of power input, and metal production, etc.

During the past eight years the writer has been actively and intensively engaged in the electric smelting of manganese, silicon, chromium, tungsten and molybdenum ores for the production of their ferro-alloys, also the electric smelting of iron ores for basic and foundry pig-irons, and has continually made a careful study of the furnace conditions on each, endeavoring to determine those factors which under some instances gave failures, while in others marked success. These conditions were encountered when smelting the various ores and in various designed and proportioned furnaces, with varying electrical capacities. I use the term "proportioned" in preference to "size." It was demonstrated that the proper proportioning of the several dimensions of a furnace, to do certain reduction work on ores by electric heating, must be determined by careful study and observation, and one must build the furnace and equipment into those conditions which exhibit themselves during operations.

E PERIENCE WITH A LARGE FURNACE

I desire to narrate an experience while operating two 15-ton rated ferromanganese furnaces in 1914. The one furnace was constructed after an ideal, or rather a dream. I use the term "constructed" because the furnace was not "designed," but the structural details were drawn up and the job assembled, with the result that there was on the ground a steel shell brick-lined, electrodes and holders, buses and transformers, and a great variety of electrical paraphernalia. The furnace was erected when I arrived, and it was ready to operate. The furnace shell was made of -in. steel plating, heavily reinforced and braced, and was a good piece of work. The shell measured 27 ft. inside in its length, with curving side walls 12 ft. across at the center, and 10 ft. at the ends; and from the bounding angle around the top of side and end wall plates to the bottom plate approximately 12 ft. 6 in. This was indeed a tank. An 18-in. brick lining was put in all around and an 18-in. bottom. This lining and bottom in the shell gave a crucible of 24 ft. by (average) 8 ft. by 11 ft., having a cubical volume effectively filled with charge of 1900 cu. ft. The four electrodes were 12 in. in diameter Acheson graphite, with threaded cone joints. The transformer installation consisted of six 750-k.v.a. single-phase General Electric transformers, two on each phase connected in parallel. Copper buses and electrode holders were sufficient for maximum loads carried. The electrical equipment on this installation was well laid out, and caused no interference, but it was not possible to get any good measure of this electrical capacity into that furnace full of charge material.

This furnace was considered a "large" size furnace. From the standpoint of smelting capacity it was a "small" size furnace. Here we have two "sizes" for the same furnace, each dependent on particular features of design, production and the ratio of certain component parts to one another, but such statements are vague and ambiguous when speaking of "size."

From the above figures we may note, the furnace, filled 90 per cent of its effective volume with charge, amounts to 1900 cu.ft. Then from power input we may calculate 10 amp. per cu.ft. of charge, or approximately 10 amp. per 180 lb. of charge. If it were possible to distribute evenly the heating effect of this current throughout the furnace, it can be readily appreciated that a minimum amount of work would be done. The net result would be small volumes of fused charge probably reduced near the electrodes, and large volumes of unfused charge, not melted or reduced, spread throughout the volume of the charge at low levels in the furnace. This anticipated result was verified. In proximity to the electrode tips high temperatures were developed. As the furnace hearth was conducting, the path of least resistance to current flow was straight down to the bottom, consequently rapid and excessive heating occurred in the areas directly under and around the ends of the electrodes. It was therefore only possible to carry loads (electrical) when fused and reduced material made contact between the electrodes and the furnace bottom. The electrodes were spaced 66 in. center to center. This apparently was too great a distance, at the then operating voltage, for a continuous passage of current from one electrode to the next, and to provide a continuous heated path for current to pass from one phase to the next, to give proper phase balancing, and also to develop the temperature for smelting. The conditions, then, represented a molten pool under each electrode, contact with the hearth, passage of current to the next pool through the hearth and to the electrode, completing the circuit as the current travel may be. But areas midway between the electrodes were cold, semi-fused non-conductors, and insulated each electrode, electrically and thermally, from the next.

THERMAL INFLUENCE OF THE ELECTRODE

Here, then, was a phenomenon exhibiting the heating influence which an electrode has over an area affected by it. In order to classify and distinguish this phenomenon I have termed it the "thermal influence of the electrode." I would define the expression as being the effect of the current at the electrode and contact, and the influence which the electrode has in distributing this heat to a surrounding area. This may be determined experimentally for any voltage range, with any nature of materials constituting the charge, and for any power loads, but depends principally on the voltage between the electrode and the charge or the hearth, and the nature of the charge, electrically and physically. With furnace conditions so abnormal that only light and uncertain loads could be maintained in the furnace,

and a low ore column being necessary, the loss from volatilization was certain to be high. Figures from periodic capitulations showed slag and volatile losses of manganese to run from 22 to 32 per cent. These figures show very poor recoveries compared to others on the same ores. The reasons therefor are apparent.

INEFFICIENCY DUE TO IMPROPER PROPORTIONING The various related dimensions of this furnace lacked the proper proportioning to develop the best efficiencies in the operations. As the results aimed for by a proper mixing of the charge were lost, due to the frozen areas in the furnace, the proper reductions did not take place, accounting for the high manganous oxide content in the slags. Some of the slags ran as follows:

zero.

These represented weekly averages, and running on irregular operations. These erratic smelting conditions caused high volatile losses, running as high as 9 to 14 per cent. But volatile losses may be controlled only to a moderate figure. It is not possible to reduce them to This furnace was run for 8 days only, and produced 24 tons of 80 per cent ferromanganese alloy. The recovery was 62 per cent of metal. Maximum load for 11 per cent of the time was 1800 kw. Eighty-eight tons of manganese ore averaging 44 per cent Mn and 18 per cent SiO, were smelted. This was a “large” size furnace in all respects except production and power input. It had large dimensions only. But this was not its real and only fault. The real and primary cause may be attributed to the design and lack of care and attention to the proper proportioning of its component parts. Such an inefficient piece of apparatus makes a poor but effective comparison.

The second furnace was of similar constructional details, but in its design there were incorporated those metallurgical principles of detail which the apparatus was required to carry out. Careful attention was given to the details of properly proportioning those component parts which were to make up the furnace. It had developed in my experience to regard carefully certain essential ratios in furnace details as all important in the design. I had also observed the effect of electric heating in straight resistance type ore smelting furnaces, and had learned to appreciate the effect of heat, caused by the current from one electrode, and heat caused by the current from an adjacent electrode. This particular relationship as before mentioned I have termed the thermal influence of the electrode. This heat influence may be large or small depending of course on the voltage drop between the electrode and the charge or the furnace hearth, the power input and the nature of the materials in the charge. As the mathematical value may be expressed in terms of the electrode diameter, this is primarily a factor. From the value of the thermal influence of the electrode, under certain conditions, we may find a value representing the ratio of electrode area to area of furnace crucible.

AREA OF THERMAL INFLUENCE OF ELECTRODE

To illustrate this statement: It has been determined by observation and measurement that the area of thermal influence of a 12-in. diameter graphitized electrode in an ore smelting furnace, running at 60 v. on

Considering the plane of the thermal influence area to be horizontal, it is now possible to determine the proper spacing of the electrodes, and also to fix the area of furnace crucible and the ratio of furnace area to electrode area. We now have a starting point for assembling the several factors entering into the design of the furnace. As the energy consumption per ton of product, and desired production must be known, these factors must be balanced against the former for electrode area and furnace crucible area. The proper spacing of the electrodes is an important feature and directly concerns efficiency of operation in any "size" of furnace. It is primarily essential to have a furnace operate with a cool top and a hot hearth. To obtain this the electrode spacing must be such as to utilize the thermal influence of each electrode at the plane of electrode contact, in the smelting zone of the furnace. Also the electrodes should be sufficiently far apart to increase, as far as possible, furnace capacity for charge. They must also be properly spaced to eliminate any "jumping across" of current from one electrode to another in the upper part of the furnace and in the unsmelted portion of the charge. By considering these factors, electrode spacing will be ultimately fixed.

EFFECT OF DEPTH OF ORE COLUMN

The proper depth of ore column and length of electrode to carry are very important factors, often given little thought and consideration, because their causes and effects are not understood. Many furnaces which I have examined and studied showed no reasons for making these dimensions as they were. The proper depth of ore column must bear a definite relation to the material being smelted and to the metal produced. The height of ore column directly bears the same relation. In smelting chrome and manganese ores lower ore columns are used than are in smelting iron ores. This is due to the difference in reducing reactions and slag purifying of the metal; also in manganese and chrome smelting no well of metal is maintained as in iron smelting.

In the electric smelting furnace, reducing gases are not depended on or utilized for ore reduction in the descending column. But as reduced gases are not utilized for reduction long ore columns are not necessary, and also too short ore columns are not desired. But in the "small" capacity furnace operated by hand, shorter ore columns may be used, because the accumulation of heat on the hearth is not so great as in the "larger" mechanically operated furnaces. We must also be guided by

the mechanical strength of carbon and graphitized electrodes, and the economy of carrying average lengths of either in the furnace.

The depth of ore column may not always be determined solely by metallurgical conditions. The designer may desire to carry a height of ore column inconsistent with strength of electrodes or length of electrodes obtainable from the manufacturers. He must be guided by a combination of conditions.

In the case of graphitized electrodes, good mechanical joints may be made and a joint having at least 85 per cent of the strength of the solid body of the electrode may be obtained. This refers to transverse strength of the joint and the material. Ungraphitized carbon electrodes do not exhibit such good mechanical properties in the joints. The pressure of the charge against the electrodes and the movement in the charge largely determine length of electrode, by gaining the requisite mechanical strength.

RELATION OF FURNACE DESIGN TO LIFE OF REFRACTORY LINING

A most important subject to a smelter is usually linings, pertaining to the life of the refractories in the furnace. The proper design of furnace has considerable effect on refractories, and we have another important component of furnace design. It is the distance from the front and back walls of a furnace to the electrodes (considering the electrodes in a straight line with the furnace long axis). This proper proportioning calls for a close study of past furnace operations on various loads, metals, ores, charges and rate of smelting. My experience has shown that a furnace inner-wall-line should be just close enough to the electrodes as to allow charge to fuse on its surface, but yet be far enough from the electrodes to interfere with fluxing or melting of the wall refractory. This may be controlled to a large extent by using water-cooled wall plates so that the fusing may be stopped by the rapid conduction of heat as soon as this condition is approached. In ferrochrome smelting this practice is of considerable benefit in maintaining the linings. The bad results of having the walls too far away are also numerous. The bad results may be exhibited by a cold tap-hole; running of the unfused charge into the tap-hole at time of tapping metal; inactive charge in the furnace not working down and being smelted, causing scaffolds and hangups; excessive heating of the one wall. On the side of the furnace from which tapping is done, the cooler wall conditions should result from the unfused charge constantly moving on this wall and the charge being shifted when tapping of metal takes place. On the other hand, it may also be the hotter wall, because the hot charge would follow the flow of metal, which would run to the tapping hole or notch, The opposite wall, then, would now exhibit the hotter conditions due to the charge lying dead on this wall.

If the imagination may be used, we can say the following conditions exist: As before stated, each electrode has a certain area over which its heating influence is exhibited. Consider three or four electrodes in a straight line, along the long axis of the furnace shell. Consider the thermal influence areas to be circular or nearly so, as the electrode is circular. We would then have under proper conditions three or four circular (or nearly so) areas, tangent to or interlocking one another; in other words, a hot area continuous through the center of the furnace, due to overlapping of heated areas.

As the thermal influence areas also extend at right angles to the long axis of the furnace, its effect is extended to the front and back walls, and tangents to such areas would be the wall line. The width of the furnace may be so determined, and the ultimate results on the wall refractories may be determined. It must not be considered that the location of front and back walls from the electrodes is as vitally important along the top of the charge. Here no fusion or reduction takes place, but the plane of proper fixing of these distances is in the fusion zone only. But, as will be mentioned later, straight walls are desirable, and not walls giving enlarging or reducing forms to the furnace shell.

DETAILS AND PROPORTIONS OF ANOTHER FURNACE

This second furnace was designed with the consideration of points such as those previously mentioned. It was a "large" size furnace, in cubical volume, in smelting capacity, metal production and power input. I desire to attach some of its details and proportions:

This second furnace, then, both in features of design and in operation, was much superior to the first one. The current density if evenly distributed throughout the furnace calculates to 27 amp. per cu.ft. of charge, while in the first furnace this value showed only 10 amp. This figure, while not being of any great value in this significant use, means considerable when it is considered that every cubic foot of material is acted upon by about two and one half times the current and heat that the first furnace shows. These values must not be taken in their mathematical value, but in their indicative terms. It also gives a fair value for the comparison of two fur

naces, to arrive at a relation for heat input. In the second furnace there was accomplished a greater concentration of heat, rather than a greater distribution of heat. The value for current density interprets this condition. Even heat distribution and also heat concentration may be accomplished with the proper design of the apparatus and proper proportioning of its component details.

The conditions governing voltage ranges in an ore smelting furnace are not a direct comparison to those encountered in the steel furnaces using arc and arc-resistance heating. The smelting furnace here referred to uses both arc and arc-resistance for the generation of its heat. It may be desired to change the voltage during the operations; this would necessitate variable voltage regulation on the transformers; or it may be desired to use a fixed voltage on the transformers and vary the resistance in the furnace. The subject of the proper voltage for best recoveries and proper operating conditions does not appear to be the question to settle. The question appears to be: With any voltage on the secondary side of the transformer, within reason of the points previously referred to on this subject, what are the conditions which can be obtained to get the best results from the given voltage? The matter of a few volts variation is nothing here nor there. Usually an approximate voltage is determined on to build the equipment to, often determined from some past experience, etc. It is not necessary to pay great attention to voltage range, and to spend large sums of money to build special expensive equipment to get a certain desired voltage. This may and can be taken care of by other means, and the subject of adjusting the furnace conditions to a transformer's secondary voltage may be taken care of. This of course refers to the moderate and not the extreme.

INFLUENCE OF REDUCING MATERIAL USED

In using the various reducing materials-coal, coke, charcoal, petroleum coke, gas carbon briquettes, etc.— and using any one of them in a range of sizes from 1-in. up to 21-in. ring size, it will be noted that a varying set of conditions exists for varying size of the same material and for each different material. To eliminate unnecessary repetition here, I refer to my article on "Coke as Reducing Agent in the Electric Smelting Furnace," METALLURGICAL & CHEMICAL ENGINEERING, Vol. 14, No. 12, June 15, 1916, Page 691 et seq. Here only two reducing agents are compared, coke and charcoal. But the conditions produced by the use of any of these materials bear directly on furnace operations. Such conditions may be defined as varying voltage by varying the resistance, power efficiency in the furnace, and recovery of metal in the alloy, and manganese content in slags.

It might be stated here that a transformer with the voltage ranges 30-60 or 30-60-90 would meet all requirements on a smelting furnace. A fixed voltage may be used on the secondary buses, and internal resistance of the furnace varied. By a change in the reducing material used, a change in the nature of the charge may be effected, consequently a change in the electrical characteristics of the charge is effected. Also a change in the size of the material effects a change in the electrical conditions.

In one furnace producing ferromanganese, fine coke (1) in.) continually produced carbides, while coarse coke (2 in.) gave good reductions and produced a minimum of carbides in the slag, with increased alloy production. Lump charcoal also promoted better conditions. In an

other instance in producing foundry-pig, 1-in. coke made good foundry-iron (2-21 per cent silicon). When using coarse 24-in. lumps, a low silicon pig resulted, under 1 per cent silicon. These operations took place in making electric furnace pig iron on 30-ton furnace operations direct from ore in the furnace. The component details of furnace design also played a large part here, as the electrodes were probably too close together, allowed a large part of the current to jump across, and made the furnace top very hot. The fine coke did not exhibit these details, but it caused a very hot furnace and caused the formation of considerable carbide in the bottom of the furnace. These carbide accretions soon built up and caused hang-ups, with the result of burning out side and end walls. Water cooling was resorted to, and while it made considerable difference on the side walls, did not help the interior of the furnace. One reason for the accumulation of these accretions at the point where they did occur was a contraction in the area of the furnace, causing excessive heating to a small volume of charge here and producing compounds of high heats of formation. Restrictions in shape and dimensions are bad features in an electric furnace and should be avoided unless for vital reasons. COKE AND CHARCOAL IN DEEP AND SHALLOW FURNACES

In two different furnaces running on ferromanganese, having the same bus bar voltages, same ore and reducing material, neither furnace had the same height of ore column, or the same spacing between electrodes. The deeper furnace-9-ft. depth of ore column, 60-in. spacing of electrodes-worked better on coarse coke. The other furnace-7-ft. depth of ore column, 40-in. electrode spacing-worked better on coarse charcoal. From my article referred to this becomes apparent. The shallow furnace worked well on 1-in. coke, while the deeper furnace could not be run on 1-in. coke. By increasing the voltage on each furnace and connecting the deeper furnace in star, the only alternative was a higher ore column. With higher voltages, my opinion leads to the use of higher ore column. Higher ore column means more electrode length to carry, with increased breakages. I once decided to investigate the relation of voltage to electrode breakage, and now believe that high voltage meant high ore column and consequently more electrode breaks. Undoubtedly the resistance of electrode material has something to do with electrode weakness, but in such conditions as an electrode smelting furnace runs under, this would be a minor cause.

Another point which has considerable bearing on furnace operations and recoveries is the method of charging the furnace. As referred to in the beginning of this article, a furnace controlled and worked entirely by hand can be more easily charged than a furnace requiring mechanical appliances for loading and charging. This is of course due to the better manipulation of the charge by the operator. One of the essential conditions to be desired in charging a furnace is to maintain the charge level on top of the ore column. This may be obtained in two ways, one by distributing the material by hand-rabbling, and one by properly spotting or placing the charge when dumping.

ADVANTAGES OF HAND-CHARGING AND RABBLING The operation of spreading the charge by hand-rabbling becomes increasingly difficult as the areal dimensions of the unit and size of the charges increase. This is apparent. If mechanical pushers are used on the top

of the charge, other mechanical difficulties arise. Also the desirability of performing hand operations on a furnace is of considerable advantage. Most of the charging may be done by a shovel and hand labor. In this way 20 lb. may be the maximum load thrown on the charge at one time and place. This load may be placed exactly where desired. The small unit charges may be well mixed beforehand. The charges do not fall from a great height onto the ore column, and do not develop a tendency for the charge to pack. Light loads of charge and low drop make a loose ore column and good smelting conditions. Stoking is another essential. The handoperated furnace can be more successfully stoked to prevent crusts from forming and causing the charge to keep moving down and not be allowed to become dead in the furnace. The accumulation of dead charge in a furnace will at once cut down production and recoveries. As the metal bearing materials incipiently fuse and crust the furnace, loss of metal results, and all calculations are thrown out. Heavy wall crusts affect linings, by the more rapid deterioration of side walls. In all the above respects the hand-controlled furnace shows points of superiority in operation, recovery and production.

In furnaces having areas and volumes where hand labor for charging has to be eliminated, mechanical appliances must be used. The operation of some of these mechanical appliances has considerable to do with production and recoveries in the furnace. As the majority of such methods dump the charge from above and usually from 8 to 12 ft. heights and require rather large unit charges, a decidedly different set of conditions exists from that of charging small mixed charges by the shovel from a low fall. The large charge (500-1000 lb.) falls with considerable impact. The heavier material of the charge, falling with greater impact and settling first, upsets the proper feeding of the materials. This condition usually upsets the careful mixing which may previously have been accomplished. Also ore or limerock may fall around the electrode, when the opposite may be desired, and carbonaceous material falls next to the walls. By the carbon material falling next to the walls the temperature of the walls is considerably increased. This also results in areas in the center of the furnace being impoverished in reducing material. Also one part of the furnace may be running faster than another part and it may be necessary to place charge where most needed. This necessitates some rabbling and distribution of the material over the furnace top. With an 18 by 10-ft. furnace top area, this is inefficient at best. In rabbling, a workman usually draws or pushes more of the light material than heavy material, consequently uniformity throughout cannot be obtained. These conditions are prime effectives on production and recovery in an electric smelting furnace. These condi

tions also eliminate some of the smooth operations, such as occur on furnaces of lesser dimensions, volume and capacity. They will often account for otherwise unex plained factors.

COMPARISON OF METHODS OF CHARGING

It might be of interest to mention two methods of charging used on 20-ton alloy furnaces (20 tons of alloy per day). One method was by straight drop, and the other by a movable spout.

In the straight drop method, stacks were built between electrodes, and the charge dropped directly from above. The drop was about 11 ft. It was necessary to drop at least 500 lb. of mixed charge at one time. The

charge descended on the top of the ore column with heavy impact and packed the material densely around two adjacent electrodes, as the stacks were between. Then by rabbling usually the light charcoal was pushed or pulled away, and only ore and limerock were left around the electrodes. This condition, it was seen, soon upset furnace conditions and a change was necessary. The charges were then dropped, charcoal or coke first, and then ore and limerock. This rectified matters to a certain extent, but it required more labor on the feed floor, and caused considerable loss of charcoal, by dusting, and disintegration of lump charcoal when the ore fell upon it. The whole method of charging was then changed, having proved unsatisfactory.

A movable hopper over the furnace running on a track had an outlet into a long spout, which, being on a swinging joint, moved forward and backward. When a charge was dumped from the scale car into the hopper the spout was moved forward and backward, thus distributing the charge from front to back of the furnace. From the lip of the spout the charge fell about 3 ft. to the top of the ore column. As the deflection plate in the hopper was set at about a 45 degree angle, the velocity of the charge was checked from a straight-down fall, and as a result the charge fell with considerable less impact from the lip of the spout. In this way charcoal or coke could be fed separately, properly distributed, and ore following could be also be distributed. A minimum of rabbling was necessary and operations soon showed the merits of the change. The results were uniform smelting, absence of heavy crusts in the furnace, less labor necessary on the working floor, cooler top, better cooling of side walls, less dust in charging, less disintegration of charcoal.

The continual addition of 200 to 400-lb. charges is to be preferred to 700 to 1000-lb. charges dumped at long intervals of time. In this instance again, dimensional features and methods of meeting increased capacities control recoveries and production of metal. Also, the 16-ton furnace may be made equally efficient as and more efficient than the 5-ton furnace, by proper mechanical appliances.

While the method of charging a furnace is largely dependent on the quantity of charge required per unit of time, the capacity of the furnace usually decides the method to be used. With the dimensional features of the furnace already decided upon, the method of proper charging should be given careful consideration.

RELATIVE METAL RECOVERIES FROM DIFFERENT ORES

I also wish to refer here on the relative recoveries of metal from smelting carbonate. ores, compared with smelting the oxide and silicate ores of manganese. Carbonate ores carrying not over 8 per cent silica should give lower losses from slag loss than silicate ores. As the manganese oxides are not chemically combined with silica to such large percentages in the carbonate ores, less slag losses would occur, the reason for this being freedom from rhodonite in the rhodochroisite ores.

In the silicate ores (rhodonite) the lime replaces some manganese oxide, forming lime silicates. But as lime replaces some manganous oxide, all replaced manganous oxide is not reduced by carbon to manganese. A large per cent of the manganous oxide passes again into solution with the silicates of lime and escapes reduction, forming double silicates of lime and manganese. Therefore on rhodonite ores and oxide ores carrying rhodonite, it is doubtful if slag content of manganous