A TREATISE OR ELECTRO-CHEMISTRY EDITED BY BERTRAM BLOUNT, F.I.C., ETC. THE ELECTRO-METALLURGY OF STEEL A TREATISE OF ELECTRO-^HEMISTP.y\ EDITED BY BERTRAM BLG4JNTYF.LCU ETC* . . THE ELECTRO-METALLURGY OF STEEL BY C. C. GOW Assoc.R.S.M., M.I.M.M., B.Sc. (ENG.) WITH A PREFACE BY DONALD F. CAMPBELL M.A., A.R.S.M., M.I. MM., M.I.E.E. NEW YORK D. VAN NOSTRAND COMPANY EIGHT WARREN STREET 1922 %?*,-,'.-.;,; . * -'"" A; **l:-': : ' ? ' : \ ^/'\J PRINTED IN GREAT BRITAIN BY THE ABERDEEN UNIVERSITY PRESS LIMITED PREFACE. ALTHOUGH electro-metallurgy is still in the early stages of development, no apology is needed for the devotion of a volume to the application of electricity to the melting and treatment of steel. The electric furnace gives us a new atmosphere, in which steel-making can be accomplished. Purification of miscellaneous qualities of steel containing noxious substances can be controlled with great exactitude. Scientific synthesis of complex steels by the addition of many valuable elements to pure metallic iron under accurate chemical control has replaced the old methods which are sometimes suggestive of the mysterious prac- tices of the alchemists. The melting of steel in a truly reducing atmosphere is only possible with electric heat, and, consequently, new phenomena are observed and often striking results obtained. Some of these we understand, or suppose we do, but others must be the subject of further research. The electric furnace has given us at reasonable prices new materials of special value to the steel-maker, such as low carbon ferro-chrome and high grade ferro-silicon, which, in turn, have been economically made into such products as stainless steel and transformer iron. The former product is being used by engineers for remark- able new applications, and the latter has increased the efficiency of electric transformers to an extent which represents, an annual saving of hundreds of thousands of tons of coal per annum. The electric furnace came as a corollary to the con- struction of the first dynamos, and although only twenty years old, it is already absorbing millions of electrical 5001-12 VI PEEFACE horse-power for various purposes, and has produced more than a million tons of steel. The widest application of this process in the imme- diate future will probably be for the treatment of the phosphoric ores of Alsace-Lorraine, where the process of Thomas and Gilchrist will be supplemented by electric refining. The stringent requirements of modern engi- neering will then be met by electrically refined steel of the highest quality, and the valuable phosphates will be rendered available for the great industry of agriculture. Other ores containing insufficient phosphorus for the basic bessemer process may soon be smelted with phos- phoric limestone, and similarly treated, so that the phosphorus now wasted may be recovered and returned to the soil. To the electrician, the problem of electric furnace design and operation is of absorbing interest. With currents of 20,000 amperes or more, phenomena of re- actance, eddy currents, and skin effect have involved new problems only recently recognised, and as yet not fully appreciated. The author has written a book which should attract both the student and the practical steel-maker, as his scientific attainments and wide experience of steelworks in several European countries enable him to appreciate the difficulties of those who have to deal with the electro-metallurgy of steel both in its theoretical and in its practical aspects. The volume gives the early history of a new branch of metallurgical science, which is surrounded by many fascinating problems and great promise of new achieve- ment, while a clear statement is made of our present knowledge of the use of electric furnaces for the metal- lurgy of steel. DONALD R CAMPBELL. INTRODUCTION. THE electro-metallurgy of steel as an industrial science owes its present status to the vast development of the past few years. Prior to the outbreak of the Great War, the electric furnace had only a very limited appli- cation in Great Britain and other countries. The shortage of high grade raw materials, the enormous demand for alloy steels, and the vast accumulation of heavy steel turnings presented exceptional opportunities for the electric furnace, and it was only then that its economic advantages in certain branches of steel-making began to be generally recognised. Since that time various modifications have been made in the manner of utilising electric energy in arc furnaces, and a far clearer and wider understanding of the chemical reactions peculiar to the processes of steel- making in these furnaces has also resulted. The induction furnace at one time certainly received greater prominence than the arc type, although it is doubtful whether it ever had a corresponding advantage in actual output of steel. Many furnace designers, in their belief that the principle of induction heating had superior advantages, concentrated their efforts on the production of a furnace which could operate on any standard electric supply system, and, at the same time, meet all the requirements of the steel-maker. As a result, numerous types were introduced, and consider- able publicity given to their construction and operation. vii Vlll INTEODUCTION In this book, the electro-thermic processes of steel- making have been studied more especially in their relation to arc furnaces, from which almost the entire output of electric steel is now being made. Little space has, therefore, been devoted to the construction and operation of induction furnaces, although sufficient has been written to indicate their electrical, constructional, and metallurgical features and the principles involved. In the earliest days of crucible steel-making, the im- portance of the character of the fuel used, the method of heat application and temperature control were fully recognised and studied by the steel-maker; precisely the same consideration is also given to these factors now by those responsible for the manufacture of steel in open-hearth gas furnaces. It is obvious that a full knowledge of both the chemical and physical con- ditions, which constitute a thermo-chemical process for the production of a metal, should be within the province of the metallurgical engineer, and there is no reason why any such process conducted by electro-thermic means should be an exception. Naturally, it may be thought that the application of electric energy to either induction or arc heating is hardly comparable to that of solid or gaseous fuel, and demands the study of a special branch of electrical engineering. This is certainly true from the point of view of electric furnace design, but does not properly apply to the metallurgist, who will only require a general knowledge of the electric fuel provided. The metallurgist best knows the thermal and chemical conditions which conduce to maximum economy and technical efficiency, and should be in a position to judge whether the electric energy is being utilised in the most suitable manner. The chemical and electrical conditions are always INTRODUCTION ix mutually dependent upon one another, and, without a knowledge of the general principles underlying the use of alternating currents for arc heating, it is impossible to know whether a slight adjustment of the electrical or the chemical conditions is required to remedy an unsatisfactory state of either. The problem is, there- fore, more complex than for those processes in which the thermal conditions are quite uninfluenced by the chemical operations. The general principles and application of alternating currents have been very briefly described in Chapters If. -IV., and every endeavour has been made to pre- sent this introductory study in a manner comprehensible to those without any mathematical or other special knowledge of the subject. This attempt probably falls far short of its object, and may receive criticism from those well acquainted with this special branch of electrical engineering. However, a little knowledge is not always dangerous, particularly if it should enable the cause of certain phenomena to be correctly inter- preted and discussed with those who are responsible for effecting the remedy. Factors bearing upon the cost and economic use of electric energy have been dealt with in Chapter VI., and it is hoped that the issues raised will indicate how the maximum economy of power cost may be attained. It has been impossible to give any definite figures bearing upon the cost of steel production as governed by factors other than the actual cost of power, raw materials, and labour. It has been emphasised through- out this book that such all- important factors as power- consumption, life of linings, and electrode consumption are not only all mutually interdependent for any given type of furnace, but are absolutely controlled by the X INTRODUCTION furnace load factor and the precise nature of the process used. For these reasons, even comparative costs are dangerous and misleading without stating the exact conditions of operation in each case, and it is, therefore, far better to ascertain such figures under the exact working conditions provided for from any of the numerous sources available. The author has to express his great indebtedness to Mr. R P. Abel for the unstinted assistance he has given in connection with the electrical chapters, while his thanks are also due to Messrs. Thwaites Bros, for affording every facility to investigate their plant, and for providing several of the illustrations ; to Messrs. B. Drinkwater and J. Holden for their assistance in compiling the Appendix ; to those constructors of electric furnaces who have kindly placed at his disposal information and drawings relating to their operation and design ; and to numerous friends who have given advice and information. The majority of the illustrations are original, and permission to reproduce many of the remainder was kindly granted by The Iron and Steel Institute, The Faraday Society, The American Electro-chemical Society, and the Canadian Government Commission. Lastly, the author desires to express his grateful acknowledgment to Mr. I). F. Campbell for contributing the Preface. C. C. GOW. CONTENTS. CHAPTER I. PAGE HISTORICAL DEVELOPMENT OF ELECTRIC STEEL FURNACES. . . .1 CHAPTER II. DEFINITIONS OF ALTERNATING CURRENT CHARACTERISTICS . . ^ . 42 CHAPTER III. APPLICATION OF SINGLE AND POLYPHASE CURRENTS . . . *. . 52 CHAPTER IV. GENERATION AND CONTROL OF SINGLE AND POLYPHASE CURRENTS . . 74 CHAPTER V. AUTOMATIC REGULATORS AND ACCESSORY INSTRUMENTS . . . .86 CHAPTER VI. POWER CONSUMPTION AND CONTRIBUTORY FACTORS 102 CHAPTER VII. ELECTRO-METALLURGICAL METHODS OF MELTING AND REFINING COLD CHARGES . . . . . . . . . 113 CHAPTER VIII. LIQUID STEEL REFINING .".-. 157 CHAPTER IX. INGOT CASTING . . . . 165 CHAPTER X. APPLICATION OF THE ELECTRIC FURNACE TO FOUNDRY PRACTICE . . 197 CHAPTER XI. CHARACTERISTIC PRINCIPLES AND FEATURES OF ELECTRIC FURNACE DESIGN 214 CHAPTER XII. MODERN TYPES OF ELECTRIC STEEL FURNACES .... . 234 Xll CONTENTS CHAPTER XIII. PAGE REFRACTORY MATERIALS AND THEIR APPLICATION TO ELECTRIC FURNACE CONSTRUCTION 281 CHAPTER XIV. FURNACE LINING AND LINING REPAIRS 292 CHAPTER XV. PROPERTIES AND MANUFACTURE OF CARBON ELECTRODES .... 306 APPENDIX I. RAPID METHODS OF ANALYSIS FOR BATH SAMPLES ..... 330 INDEX 343 LIST OF ILLUSTRATIONS. CHAPTEK I. FIG. NO. PAGE 1. Siemens' Direct Arc Furnace . ; ..... . , ' 2 2. Siemens' Indirect Arc Furnace . . . . . . . .2 3. Ferranti Induction Furnace . . . . . . . . ."8 4. Ferranti Induction Furnace . . . . . . . . .4 5. Colby Induction Furnace . . . . . . . . 5 6. Kjellin Fixed Induction Furnace (Horizontal Section) . . . . 6 7. Kjellin Fixed Induction Furnace (Vertical Section) . . . .7 8. KjeUin Tilting Furnace (Vertical Section) . ..'... . 8 9. Rflchling-Rodenhauser Single-phase Furnace (Horizontal Section) . 9 10. Rochling-Rodenhauser Single-phase Furnace (Vertical Section) . . 10 11. Rochling-Rodenhauser Three-phase Furnace (Horizontal Section) . 11 12. Rochling-Rodenhauser Three-phase Furnace (Vertical Section) . . 11 13. Frick Induction Furnace (Vertical Section) . ... . .12 14. Frick Induction Furnace (Front Elevation) . . . . . .14 15. Frick Induction Furnace (Plan) ........ 15 16. Hiorth Induction Furnace . . 16 17. Stassano Smelting Furnace '. . 17 18. Stassano Smelting Furnace (later type) ... . . . . 18 19. Stassano Smelting Furnace (later type) . . . . .19 20. Heroult Fixed Furnace (earlier form) . * . . . . . . 23 21. Heroult Tilting Furnace . , 24 22. Heroult Tilting Furnace at La Praz (France) . . Facing page 25 23. Keller Furnace at Unieux (France) . . . . . . .27 24. Keller Furnace (earlier type) . . . . ' , . ... .29 25. Keller's Conducting Hearth . . ..... . . . . 30 26. Keller's Radiating Conductors .32 27. Girod Bottom Conductors . . .".... . . .34 28. Girod Annular Conductor . . . . .- . . . . 34 29. Electro-metal Tilting Furnace . ... . . '. .38 30. Snyder Furnace . . . . . . . . . . . 40 CHAPTER II. 31. Wave Form and Clock Diagram .... . . . . . 42 32. Clock Diagram with Two Vectors . . , 43 33. Wave Forms for Two Vectors . * - . . . - . , . 43 34. Conductors and Magnetic Field . . ... . . .46 35. Closed Ring and Two Circuits . , ,. . ....... 46 36. (Two-phase Voltage) Wave Form Diagram . . -. ; , i ..V ^ xiii XIV LIST OF ILLUSTRATIONS CHAPTER III. FIG. NO. PAGE 37. (Three-phase Voltage) Wave Form Diagram 51 38. Series Arc Single-phase Connections 52 39. Single Direct Arc Connections 52 40. Single Indirect Arc Connections 53 41. Induction Furnace Connections 53 42. Three Wire Single-phase Connections . .53 43. Two- phase Circuits 54 44. Two-phase Wave Curves and Resultant 54 45. Heroult Two-phase Connections 55 46. Electro-Metals Two-phase Connection 56 47. Electro-Metals Two-phase Connection 56 48. Girod Two-phase Connections 56 49. Stobie Two-phase Connections 57 50. Rennerfelt Two-phase Connections 58 51. Dixon Two-phase Connections . . . . < . . .58 52. Three-phase Wave Curves .59 53. Three-phase Circuits 60 54. Star Connection Diagram . . 60 55. Three-phase Nomenclature Diagram .62 56. Three-phase Clock Diagram 63 57. Three-phase Circuits 63 58. Mesh-connected grouping ......... 63 59. 60 Three-phase Wave Curves 65 60. Heroult Three-phase Connections 67 61. Three-phase Inverted Star . . 67 62. Four-phase Wave Curves 69 63. Dixon Four-phase Star Connection 70 64. Dixon Four-phase Mesh Connection 71 65. Electro-Metal Four-phase . . 72 66. Resultant of Unequal Waves (120 apart) . . . . . .72 CHAPTER IV. 67. Miles-Walker Generator, Characteristic Curve 76 68. Scott Two-phase to Three-phase Transformer Connections ... 80 69. Scott Three-phase to Two-phase Transformer Connections ... 80 70. Change-voltage Switch (Mesh Connections) 81 71. Change-voltage Switch (Star Connections) 81 72. High Tension Reactance with Tappings 82 73. Circuit Reactance and Resistance Volts 84 74. Circuit Reactance and Resistance Volts 84 CHAPTER V. 75. Thury Regulator Facing page 90 76. Thury Regulator, Diagram of Connections 94 77. Curve of Load Factor and Cost per Unit 106 78. Curve of Operating Load Factor and Useful Energy .... 109 79. Curve of Monthly Load Factor and Power Consumption and Curve of Monthly Loa'd Factor and Weekly Output 110 LIST OF ILLUSTRATIONS XV CHAPTER VII. PIG. NO. PAGE 80. Pouring Spout to Hold Back Slag 148 81. Furnace Tools 149 CHAPTER IX. 82. Illustration of Solidification 169 83. Fracture Showing Chill and Equixed Crystals . . . Facing page 170 84. Solidification of Steel in Parallel Mould . . ' . . . .172 85. Tilting Ladle . . . . . . . . .. : . . . . 182 86. Bottom End of Built Up Stopper Rod . .. . , . . . . 183 87. Ladle Heating Furnace . . . ...... .185 88. Ladle Heating Blast Pipe- ......... 186 89. Ingot Section Cast Small End up ... . . . . 189 90. Ingot Section Cast Big End up . . . -. . . . .189 91. Closed Top Mould and Bottom Plate . ,'. . ' ~ : . '. . , . 191 92. Four- Way and Six- Way Runner Bricks . . . - . . . 192 93. Trumpet Pipes . . . . . . . . ' ... 192 94. Mould for Bottom-running (Big end up) . . . . . 192 95. Dozzles . . . . . . . . .... . .195 96. Cheek Bricks. . .... 'i . ... .195 97. Tun Dish '/ . : . . . . . . 196 CHAPTER X. TABLE 1. ADVANTAGES AND DISADVANTAGES OF CRUCIBLE, CONVERTER, AND ELECTRIC FURNACE. 98. Furnace Installation in Foundry . . . . . Facing page 204 69. Microphotographs Showing Effect of Annealing . . 208 100. Microphotographs Showing Effect of Annealing . . ,, 208 101. Slag Inclusions . . ..-'. . . . . . ,. ,. 212 CHAPTER XII. 102. Section of Stassano Furnace Melting Chambers . . . . .235 103. Vertical Section of Stassano Furnace ...--. . . . .235 104. Rennerfelt Furnace Wiring Diagram . . . . . . . 238 105. Rennerfelt Furnace Section of Trunnion Mounted Body Facing page 240 106. Rennerfelt Furnace (rectangular) . . . . . . 241 107. Rennerfelt Furnace. (Section Through a Basic Lining) . . . 242 108. Heroult Three-phase Furnace (6 Tons capacity) 248 109. Heroult One and a Half Ton Furnace . , . . . . 251 110. Girod Single-Phase Diagrams . . . . . . . . 252 111. Girod Single-Phase Furnace ... . . . .. % . .253 112. Electro-Metals Wiring Diagram Facing page 256 113. Electro-Metals Seven and a Half Ton Furnace . 257 114. Electro-Metals Four-phase Furnace , . . . ... 258 115. Stobie Five Ton Two-phase Furnace ....... 261 116. Snyder Power and Current Curves ... .... 264 117. Snyder Furnace Load Chart . . . . . . . . . 266 118. Snyder Five Ton Furnace . . . . * . . . . 268 119. Snyder Furnace Section . . . . ... . .269 XVI LIST OF ILLUSTRATIONS FIG. NO. PAGE 120. Greaves-Etchells Twelve Ton Furnace Section 272 121. Greaves-Etchells Twelve Ton Furnace Section (Back Elevation) . . 274 122. Booth-Hall Furnace Connections 275 123. Booth-Hall Furnace (Transverse Sections) 278 CHAPTER XIV. 124. Key Sketches for Lining Furnace 294 125. Lining of Seven Ton Furnace 295 12G. Lining of Conductive Hearth Furnace 298 CHAPTER XV. 127. Diagram of Apparatus for Electrode Conductivity Test . . . 316 128. Double Electrode Screw Plug 319 129. Stobie Economiser 327 130. Electro-Metals Economisers . . 328 131. Ball and Socket Sleeve Economiser 328 APPENDIX I. 132 Carbon Combustion Apparatus Facing page 340 CHAPTER I. HISTOKICAL DEVELOPMENT OF ELECTRIC FUENACES. THE Electro-metallurgy of steel is now an applied science of considerable industrial importance, and has only reached its present stage of development during comparatively recent years. The application of electric energy for melting steel had been demonstrated by Sir William Siemens almost twenty years before its commercial possibilities were recognised by later in- vestigators, to whose work the present status of the electric steel industry is primarily due. The generation of electric currents by means of the dynamo or similar device was only discovered in 1867, so that the slow advancement made subsequent to the work of Siemens was no doubt mainly due to the lack of thoroughly reliable electrical equipment, without which any electro-chemical process, though conducted with the best technical and organising skill, is doomed to commercial failure. The use of electrical heating for melting steel on a com- mercial scale was first demonstrated by the induction furnace, but no real impetus was given to the industry until the introduction of the arc furnace a few years later. For several succeeding years the electro-thermic processes were limited to localities peculiarly favourable to economy of power production and to a few large undertakings, which, at the instigation of the inventors, constructed larger units for the further study of the products and cost of production. Before investigating further the development of the electric steel furnace at this period, the work of Sir William Siemens in 1882 must be more closely considered. The melting of steel was accomplished in a small crucible furnace in which the principles of both direct and indirect arc heating were utilised 1 ^LECTRO-METALLURGY OF STEEL i gtst v est&bUslied; f .Figs. 1 and 2 illustrate the two arrange- ments used. In the case of the former (Fig. 1), the furnace charge serves as one electrode, and is connected to the source of electrical supply through a carbon or metallic pole penetrating the furnace bottom or hearth. The electric current enters the furnace through an upper hanging electrode, and strikes an arc at the point of the conducting charge where contact is made. The principle in- volved is similar to that embodied in several modern types of furnaces. The furnace represented in Fig. 2 is de- pendent on indirect arc heating ; and is furnished with horizontal electrodes which pass through the walls. The arc is struck between the ends of the electrodes, and the furnace charge receives heat by radiation from the arc and reflection from the roof and .walls. Here, again, the principle FIG. 1. Siemens' direct arc furnace. FIG. 2. Siemens' indirect arc furnace. is identical with that employed in another class of modern electric furnaces of which the Stassano furnace was the fore- runner. To Sir William Siemens, therefore, belongs the credit of having first demonstrated two systems of electric arc heating as HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 3 applied to steel melting, and in this respect he may be con- sidered the pioneer of the industry. His researches were certainly confined to the laboratory scale, but it is more than probable that, had the requisite electrical plant been available at that time, his efforts would have been further extended to the commercial application of his discovery. A period of inactivity in the development of arc heating followed until 1898, but during that time other discoveries were made which demonstrated the possibility of melting metals by the heating effect of low voltage currents induced FIG. 3. Ferranti induction furnace. in the metal itself. The electrical principles involved in the operation of induction furnaces are dealt with in Chapter II. Ferranti 1 in 1887 constructed a furnace on this principle in which he was able to melt metals, although no special claim was made for melting steel. The principle of induction heating was later adopted in various furnaces which at one time promised to bear an important part in electric steel melting, but which have now been almost entirely abandoned in favour of arc furnaces. The melting chamber of Ferranti's furnace (Figs. 3 and 4) consists of an annular channel A, provided with a suitable cover- 1 British Patent Specification, No. 700, 1887. 4 THE ELECTKOMETALLURGY OF STEEL ing, surrounding the centre limb of a double closed magnetic circuit C. Primary windings B are also wound round the same limb immediately below the melting chamber. The apparatus consists essentially of a simple transformer, whose secondary winding of one short-circuited coil is composed of a ring of the metal to be melted. The voltage of the induced current is exceedingly low, owing to the low resistance of the secondary cir- cuit, and the intensity of the current correspondingly high. The transformer core, consisting of laminated sheets of iron, is mounted in a framework which tilts on trunnions, so that the F IG . 4. Ferranti induction furnace. molten metal can be easily poured. Three years later Ferranti was followed by Edward Colby, 1 who took out several patents in America for melting metals in induction furnaces. In his original type the primary winding surrounded the secondary or annular melting chamber, but this arrangement was later given up and their respective positions reversed. Means were also provided for tilting the furnace and pouring the metal direct. At least eight years elapsed before the first steel was made (1898-1899) in an induction furnace by Colby and Dr. Waldo in America, and 111 El. Chem. Industry," Vol. V, p. 55; U.S.A. Patent Specification, No. 428,378, 428,379, and 428,552. HISTOKICAL DEVELOPMENT OF ELECTEIC FURNACES although several of these furnaces were later installed for the pro- duction of electric steel, the introduction of modified types soon gained prominence. One of the distinctive features of the later Colby furnace l (Fig. 5) is the double magnetic circuit com- prised of three vertical legs connected together by horizontal members. Both the primary and secondary circuits surrounded the centre leg and were themselves enclosed between the two outer legs. The primary windings, it will be seen, are as close as possible to the centre core, and, by means of suitable water cooling arrangements, the secondary circuit or melting chamber could be constructed in close proximity to it, thus reducing magnetic leakage to a minimum and increasing the power factor. This was undoubtedly an excellent feature, but the difficulty of FIG. 5. Colby induction furnace. embodying this construction in larger furnaces (i.e. above 3 cwts.) was considerable on account of the high-voltage current necessary for the supply of increased power to the primary windings, which would then require elaborate insulation. The power factor of the furnace is given by Colby as about '9, which compares favourably with modern arc furnaces. Colby's induction furnace, as actually applied to steel-making, was patented in America in 1900 2 or at least one year after his first production of steel. The same year was marked by further improvements in induction furnaces, and, what is far more im- portant, by the re-introduction of arc heating, both direct and indirect, for steel melting. Thus the year 1900 may be regarded 1 U.S.A. Patent Specification, No. 859,641 (1907). 2 "El. Chem. Industry," Vol. Ill, 1905, pp. 80, 134, 299, and 341, also Vol. V, 1907, p. 232. 6 THE ELECTRO-METALLURGY OF STEEL as the most important in the history of the electric steel industry. To avoid confusion the further development of the induction furnace will be first considered, returning later to the history of arc furnaces, which in their commercial form date from this year. The Kjellin induction furnace, first patented in 1900 and operating in 1902, embodied the same electrical principles as used FIG. 6. Section AB. by Ferranti and Colby, the salient difference lying in the use of high-voltage current in the primary windings for the purpose of increasing the furnace power capacity. One of the magnetic circuits was also eliminated, leaving of the three vertical legs the centre core and one outer leg, which were set in the plane of tilting. Figures 6 and 7 show a diagrammatic plan and section of the fixed furnace in operation at Gysinge in Sweden, HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 7 which was reported upon by the Canadian Government Com- mission in February, 1903. The melting chamber B, when charged with metal, constitutes a short-circuited secondary wind- ing of one turn, and surrounds the primary windings A and one limb of the iron core C. The primary windings being carefully insulated were supplied with a single-phase current of 90 amperes at 3000 volts and 15 cycles, representing about 168 kw. at a power factor of "62. They were further protected from the 4_ _i FIG. 7. heat radiated from the melting chamber by means of water cir- culation supplementing natural air cooling. The melting chamber was closed in by a number of small cover bricks KK, which could be easily removed for purposes of charging, repair- ing and inspection, and was provided with a lining composed of silica or magnesite bricks filling the portion marked D set on a foundation of firebricks D 1 . Sufficient metal was always left in the furnace after tapping to provide an unbroken secondary circuit for the induced current when starting a subsequent heat. 8 THE ELECTKO-METALLUKGY OE STEEL The radiation loss of the furnace mentioned was equivalent to 80 kw., leaving only 88 kw. for melting, and with such a ratio the power consumption could not possibly have been anything but high. The furnace was chiefly used commercially for melt- ing Swedish white iron and steel scrap requiring no refining treatment. In this respect any advantage of the induction furnace over crucible melting lay solely in economy of heat utilisation. The later type of Kjellin furnace was adapted for tilting and is illustrated in Fig. 8. The low frequency required necessitated FIG. 8. Kjellin tilting furnace. the use of special generating plant, and consequently increased the cost of production by the heavier capital outlay involved. The power factor of the furnace was unavoidably low, being about '6 to '65, and fell still further on increasing the furnace capacity. For this reason a lower frequency than 15 was necessary for the larger units. In 1909 there were in operation only ten Kjellin and one Colby furnace, so that ten years after the introduction of the induction furnace no considerable pro- gress had been made. The first Kjellin furnace to operate in England was erected by Vickers Sons & Maxim and began operation in 1908, but this did not lead to further development. HISTOEICAL DEVELOPMENT OF ELECTKIC FURNACES 9 Although at that time considerable . advances had been made in other countries with induction and more especially arc furnaces, this Kjellin furnace was the first electric furnace plant to be installed in the United Kingdom. A fixed furnace l of 1 ton capacity, operating at Gysinge, produced 950 tons of tool steel and special steel ingots during the year 1906. The bulk of the output was produced from charges composed of about 80 per cent. Swedish white iron and 20 per cent, of steel scrap; briquettes of ore were added to regulate the percentage of carbon in the steel cast. The energy consumption 2 was nor- mally 1128 kw. hours per ton, but when white iron and steel scrap were melted without briquettes the power consumption fell to 886 units per ton. A distinct departure from the design of Kjellin was made in the Kochling-Rodenhauser furnace, which was patented in May, 1906, and designed to operate with either single or three-phase current. In each case the iron cores and primary windings were enclosed by a secondary circuit. Auxiliary secondary windings, 1 "Iron and Steel Institute Jour .al," Vol. Ill, 1906, p. 397. 2 Ibid., Vol. I, 1909, p. 298. 10 THE ELECTRO-METALLURGY OF STEEL consisting of a few turns of heavy copper bar, were connected to terminal plates embedded in the furnace lining, and provided a further source of heat to the furnace charge, besides improving the electrical efficiency. Figs. 9 and 10 l show a sectional plan and elevation of the single-phase furnace, and Figs. 11 and 12 corresponding views of the three-phase design. Two furnaces, one of which was an 8- ton single-phase furnace and the other a 2-ton three-phase furnace, were being operated at the Rochling Iron and Steelworks at Volklingen in 1909. The 8-ton furnace of 600 kw. capacity was supplied with power from a 4000 to 5000 volt 5 cycle per second generator, while the smaller three-phase furnace of 250 kw. capacity operated on a supply of 400 volts at 50 cycles. The higher periodicity per- missible with the three-phase furnace results from (a) its smaller capacity, (b) the higher power factor obtainable by the closer and better arrangement of the transformer cores relative to the primary and secondary windings, (c) the auxiliary secondary windings which reduce magnetic leakage. In both furnaces the annular melting chambers meet in a common hearth of wider proportions, situated between the vertical legs of the iron core, and it is here that the heat is 1 " Iron and Steel Institute Journal," Vol. I, 1909, p. 270. HISTORICAL DEVELOPMENT OP ELECTKIC FURNACES 11 intended to be absorbed by the current generated in the auxiliary FIG. 11. FIG. 12. secondary windings. The increased dimensions of this hearth enabled certain metallurgical operations to be performed which 12 THE ELECTEO-METALLUEGY OF STEEL had hitherto been impossible in the older forms of induction furnaces. The central chamber is arched over (Figs. 10 and 12), and provided with a charging door and spout. Fluxing materials can thus be easily charged and the bulk of slag removed if desired. The greater part of the heat is generated in the narrowed section of the secondary channels, and trans- mitted to the metal in the centre portion by circulation. The FJG. 13. auxiliary heating already mentioned is of doubtful advantage, and it is not clear how the bulk of this heat can be generated anywhere but in the refractory lining separating the terminal plates from the metal. The circulation of metal along the narrow portion of the secondary circuit is promoted by electro- magnetic effects in a fixed direction, and maintains a sufficiently high temperature in the centre portion of the bath to promote HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 13 slag reaction and for casting. According to Kodenhauser the single-phase furnace, when under three tons capacity, is superior as regards efficiency to the polyphase, while for larger sizes the polyphase type is preferred. This is mainly due to the power factor of the single-phase furnace becoming increasingly lower as the distance of the secondary melting chamber from the iron core becomes greater. From a metallurgical standpoint, the " Combination " furnace, as it was called by Rodenhauser, was a material advance on the simple induction furnace ; not only could pure materials be melted and poured, but the construction of the hearth, as previously explained, permitted the removal of phos- phorus and sulphur from less pure classes of scrap. In the year 1913 l there were thirteen Rochling-Rodenhauser furnaces actually in operation, ten of which were employed for refining molten steel or pig-iron, and the remaining three for melting and refining cold charges. Two years after the introduction of the Rochling-Roden- hauser furnace another induction type furnace, designed by Frick, 1 was operating at Krupp's Essen works with a power input of 750 kw. The single-phase furnace, as illustrated in Figs. 13, 14, and 15, has certain features in which it resembles both the Colby and Kjellin furnaces. A three-legged, laminated iron core is adopted, and the primary windings are designed for high-voltage currents. The peculiar construction of the primary windings was intended to improve the power factor by reducing magnetic leakage, and to effect this object the coils were split up into three portions, the top and bottom coils being flattened out to enclose the secondary circuit. This was com- monly known as the "Umbrella" design. Side inspection doors were also provided in place of the removable cover bricks of the earliest Kjellin pattern. Frick, in his paper read before the Iron and Steel Institute in 1913, describes at con- siderable length the electrical and metallurgical operation of his furnace, but does not indicate any considerable improve- ment in power factor resulting from the special arrangement of the primary windings. The annular melting chamber, 1 ' : Iron and Steel Institute Journal," Vol. II, 1913, pp. 297 et seq. 14 THE ELECTRO-METALLUEGY OF STEEL necessarily steep-sided, was composed of a very pure magne- site treated in such a way as to resist contraction and ex- pansion at high temperatures. This lining enabled charges FIG. 14. Frick induction furnace. to be melted down under basic slags for purposes of phos- phorus removal. Further, a rotatable cover was provided for the melting chamber to facilitate uniform charging and repairing. HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 15 There now only remains to be mentioned the Hiorth 1 furnace which was patented in Norway in May, 1909, and put into operation early in 1910. J. W. Eichards, in a paper read before the American Electro-chemical Society in 1910, describes a 5-ton capacity' unit supplied with 400-500 kw. at 250 volts, 12- cycles. The furnace (Fig. 16) resembles the Eodenhauser arrangement of a double secondary circuit joining to form a o . i FIG. 15. Frick induction furnace. central hearth, and embodies also the " Umbrella " type of primary winding used by Frick. In this case, however, the primary voltage is low, and to enable the heavier current to be carried at greater current density the primary coils are con- structed of copper tubing for water circulation. During the heat made on the occasion of J. W. Kichard's visit to the Jossingford works, the power factor varied from '57 to *8. A 1 " Am. Electro-Chem. Soc.," Vol. XVIII, 1910, p. 191. 16 THE ELECTROMETALLURGY OF STEEL novel feature of the furnace construction enables the furnace body to be tilted for pouring without movement of the iron core and upper set of primary coils ; the lower set, like that in the Frick furnace, is attached to the furnace bottom and moves with it. The development of the induction furnace shows that considerable inventive skill has been concentrated on attempts to improve its electrical features, and to render it a useful ap- FIG. 16. pliance in the hands of the steel-maker. Unfortunately the fundamental elements of an induction furnace are adverse to the flexibility and economy of operation which are essential to the production of electric steel as a commercial commodity. It is therefore not surprising that arc furnaces by their simul- taneous development have practically supplanted the induction furnace for the melting and refining of steel. The induction furnaces mentioned in this survey are only the better-known types that have been used commercially. On returning to the history of arc furnace development a HISTOKICAL DEVELOPMENT OF ELECTRIC FURNACES 17 brief description of each type will be sufficient to indicate the material points of difference, a detailed description of those types now in extensive use being given later. Following the publication of Sir William Siemens' work with arc furnaces, sixteen years elapsed before the problem of melting steel by arc heating was again attacked. In February, 1899, Stassano made known the results of experiments on the reduction of iron ores by electro-thermic means conducted during the preceding year. His experiments were confined to the re- duction of ore to metal, which he proposed to further refine in the same furnace for removal of carbon when necessary. His first furnace, built in 1898 (Fig. 17), clearly resembles that type of Siemens' electric crucible furnace which embodied the principle of indirect arc heating. The furnace was built by Stassano in Kome and used solely for experiments on the re- duction of iron ore. A mixed charge consisting of briquettes was supported in the basic lined shaft A by an iron grating fixed 8 inches above the arc zone. The reduced metal trickled 18 THE ELECTRO-METALLURGY OF STEEL through the unreduced charge into the crucible C, from whence it was tapped at intervals. The slag, however, was not suffi- ciently fusible, and remained as a solid arch above the arc zone, preventing further reduction of the charge above. This serious difficulty led to a modified form (Fig. 18) in which the mixed charge was introduced below the arc zone instead of above it, and in this way the difficulties of maintaining a continuous reaction were over- come. It was found, however, that during the first stages of reduction the power required was much greater than during the final stages when the bulk of the metal and slag had been fused and melted. Owing to the impossibility of reducing the load without forming a long and unstable arc, Stassano was later forced to modify this furnace by using two or three pairs of elec- trodes in place of one pair (Fig. 19) ; the furnace load could then be varied within certain limits by extinguishing one or more arcs. The reduction from full to low load was at that time objectionable to the generating station, and it was then suggested that, by the use of two furnaces always operating at different FIG. 18. stages of the reduction process, the mean load taken might approach a more constant figure. The plurality of electrodes and the excessive difficulty of operat- ing the two furnaces so as to require at all times a constant generating station load soon showed the impracticability of this arrangement. Stassano then argued that, if the period when the minimum power is required could be shortened by rapid treatment and removal of the metallic mass during the reducing fusion of the HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 19 charge, the furnace might be continuously operated at a higher and more uniform load. He had also observed that increased economy was gained, both in time and fuel, by using a furnace with a movable hearth for refining pig-iron, and thereupon conceived the idea of embodying this principle in his furnace to overcome his difficulties. This was the origin of the Stassano furnace as later used for melting steel scrap, and which in its first form was patented FIG. 19. in 1902. In 1903 Dr. Goldschmidt, reporting on behalf of the German Patent Office, stated that mild steel could be success- fully made from pure ores, but that the process was too expen- sive for economic competition. Dr. Haanel's report of 1904 contains information communi- cated by Stassano relative to a furnace of 1000 horse-power rating. The output was given as 4 to 5 tons per day according to the quality of ore used. The furnace was provided with 20 THE ELECTRO-METALLURGY OF STEEL two pairs of electrodes in parallel, each pair supplying an arc with 2450 amperes at 150 volts. Apart from the usual fettling, the lining was estimated to last at least forty days without repair. Later the rotating furnace was further modified to meet the requirements of melting mixed steel and iron scrap, and thus the production of mild steel or iron by direct reduction of ore gave place to simple melting and refining. While Stassano was endeavouring to apply electric heating to the more primitive method of steel-making by direct reduc- tion from ore, Dr. Heroult, in the year 1899, began his extensive work on the production of steel from pig-iron, and later from common scrap-iron and steel. He saw more prospect of success by following the lines of a well-established process in general use, which consisted of refining pig-iron and scrap in the basic open hearth furnace, and succeeded in solving the problem of carbon contamination by interposing a layer of slag between the bath of metal and the electrodes, and further by the use of a non-carbonaceous lining. This slag covering, which might be varied in composition, had the added advantage of effecting removal of injurious impurities from the bath. The develop- ment of the series arc design was both interesting and logical, and entirely fulfilled the expectations of the inventor. In the year 1899 Heroult was using a single electrode furnace, resembling in principle the direct arc crucible furnace of Siemens, for the manufacture of ferro-alloys. The furnace, lined throughout with carbon, was supplied with current by two electrodes, one embedded in the bottom and the other hanging vertically so as to complete the circuit through the conductive charge. The carbon-lined bottom was satisfactory for the production of high carbon ferro-alloys, but with a grow- ing demand for a low carbon ferro-chrome it had to be abandoned. As a substitute Heroult first employed a bottom composed of chromite ore, surrounding a single carbon pole placed centrally. He expected that the carbon would eventually be consumed, and replaced by metal to a depth where the tempera- ture was sufficiently low to prevent further absorption. This bottom carbon electrode was, however, found unsatisfactory, and the logical conclusion of supplying power to the upper portion of the conductive charge followed, this being accom- HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 21 plished by substituting another vertical top electrode for the one embedded in the bottom. In this way the Heroult steel furnace provided with two hanging electrodes and two arcs in series originated. The principle, then, of direct arc heating as first used by Siemens is followed, but with the one all-important difference that the electrodes carrying current to the furnace both enter from above the charge and so eliminate all electric connections made with the furnace hearth. Finding the new type of fur- nace successful for melting low carbon alloys having a high melting-point, Heroult then turned to the problem of melting and refining steel. The furnace lining could be made similar to that commonly used in the Siemens-Martin furnace, so that there seemed little reason why, given sufficient temperature, the same metallurgical results should not be achieved ; not only was this the case but it was soon found that this form of furnace possessed important features which permitted refining operations to be carried far beyond anything hitherto possible in the open-hearth furnace. Up to the year 1904 nine distinct patents, covering novelties of furnace design and metallurgical operations peculiar to the electric furnace, were issued in different countries throughout the world. To show how thoroughly the investigations were carried out, and how far the value of the hitherto unknown chemical reactions was understood and appreciated, it is only necessary to quote the titles of the French patents enumerated below : I. No. 298,656. Nineteen patents issued. "Improvements in electric furnaces, with a view to obtaining soft metals and other materials, in which it is necessary to prevent contamina- tion with carbon from the electrodes.". II. No. 305,317. Four patents issued. " Process and ap- paratus for the electric manufacture of wrought-iron, steel and cast-iron by electric heating. " III. No. 305,373. Patent rights granted in eleven countries. " Process and apparatus to make use of the waste heat resulting from the manufacture of pig-iron." IV. No. 307,739. Thirteen patents issued. " Tilting elec- tric furnace." 22 THE ELECTROMETALLURGY OF STEEL V. No. 318,638. Thirteen patents issued. " Electric fur- nace with movable electrodes." VI. No. 320,682. Nine patents issued. "Process for deoxidising and carburising liquid steel." VII. No. 328,350. Six patents issued. " Improvements in the production of iron and steel by electro-metallurgical means." VIII. No. 336,705. Patent issued in France only. "Pro- cess for deoxidising and desulphurising steel." IX. No. 356,714. Patent of addition. " Improvements in the production of iron and steel by electro-metallurgical means." The first patent was applied for in France in March, 1900, and covered the use of a rectangular open-top furnace, provided with two or more vertical electrodes suspended above it and connected in series. Particular mention is made of two volt- meter circuits, connected between the electrodes and the metal in the crucible for the purpose of regulating the voltage of each arc, and so dividing the load equally between them. The elec- trodes were capable of being raised or lowered, and the use of single-phase and polyphase alternating currents provided for. The crucible could be lined with any known refractory material, and this, together with the use of a slag covering, enabled metals to be melted without the usual contamination with carbon from the bottom or from the electrodes. Heroult's first furnace was fixed, as is shown in Fig. 20, 1 and further comment is unneces- sary beyond drawing attention to the power and voltmeter circuits. In the second patent 2 Heroult describes the application of his first furnace for the production of wrought-iron and steel, and outlines the process to be used. The manufacture of steel was conducted in successive operations in the same furnace. Iron ore was first reduced to pig-iron, which, after removal of the bulk of the slag through the upper tap hole, could be further refined by the addition of ore. The " refining " action, or boil- ing down, could be stopped at any moment when the desired percentage of carbon had been arrived at, this being determined by constant sampling and fracture of bath tests. By making 1 British Patent Specification, No. 16,293, A.D. 1900. 2 Ibid., No. 14,486, A.T>. 1901. HISTOEICAL DEVELOPMENT OF ELECTRIC FURNACES 23 the slag thoroughly basic a " purifying " action, resulting in the removal of phosphorus, was made feasible. According to Heroult the "purifying" and "refining" actions using his own terms could proceed simultaneously or otherwise as re- quired, and the process could be conducted in the same manner whether the carburised metal was charged molten or in solid pieces. In the patent specification a disclaimer is made point- ing out the essential differences between the crucible furnace of Siemens and the furnace in question. The furnace was substantially the same as that shown in Fig. 20, but was pro- vided with a movable roof which confined the heat to the interior and could be readily changed. The next patent may be passed over as it does not bear directly upon the manufac- ture of steel. The tilting furnace l was the next development (Fig. 21). The construction shown embodied all the essential features of the modern furnace, and was designed to facilitate the refining and purifying processes as previously conducted in the fixed furnace. Other novelties were at the same time introduced, including the method of conveying current to the 1 British Patent Speciacation, No. 14,643, A.D. 1901. 24 THE ELECTRO-METALLURGY OF STEEL movable electrodes, and the use of an air blast to hasten the refining action. A method of deoxidising and carburising steel produced by the refining and purifying processes is described in a later patent. 1 The method consisted of throwing on to the bath briquettes composed of a carbonaceous material agglomerated with iron and steel filings or turnings, which, by their density, forced the briquettes through the slag into contact with the steel, so that the carbon might be more easily absorbed. The use of this material has not met with much favour, as the FIG. 21. carbon absorption was too low and erratic, any variation in the size of the pieces, slag condition, and temperature of the bath influencing the degree of absorption. It is doubtful also whether any deoxidisation was effected except by the subsequent indirect action of any carbon entrapped in the slag. It was later realised that the process of decarburising pig- iron by electrical energy was not economical, and it was then proposed to transfer liquid steel, which had been either blown in a Bessemer converter or boiled down in an open hearth furnace, to the electric furnace for the final conversion to finished 1 British Patent Specification, No. 6950, A.I>. 1903. Fm. 22. [To face p. 25. HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 25 steel. This method, which amounts to a Duplex process, is described in a further patent. 1 Lastly, the process for deoxidising and desulphurising steel as is now known was patented in France in November, 1903, so that, in rather more than three years, all those metallurgical treatments essential to the production of high quality electric steel in arc furnaces had been discovered and made use of com- mercially. These treatments may be briefly summarised as follows : I. 2nd Patent. Melting cold cast-iron and steel scrap by arc heating. II. 1st Patent. Kegulation of load. III. 2nd Patent. Oxidation of carbon and removal of phosphorus and other impurities by a basic slag. IV. 2nd Patent. Kemoval of the foul slag. V. 6th Patent. Subsequent carburising of the" bath if required. VI. 8th Patent. Deoxidation and desulphurising by means of a highly basic slag free from oxides. VII. 6th Patent. Addition of alloys to the bath without loss prior to casting. VIII. 7th and 9th Patents. Kenning liquid steel. In 1904 the Canadian Government Commission visited La Praz to investigate the Heroult process then being practised in the first tilting furnace built (Fig. 22). The furnace was connected to a single-phase alternator coupled direct to a water turbine. At normal speed the voltage was 110 and the frequency 33, the power at full load being about 360 kw. and generally much less during the melting down operation. The power consumption, in the case of those heats witnessed by the commission, averaged 1100 units per ton of ingots both for dead soft and high carbon steels. The miscel- laneous scrap used was very impure, containing 0*22 per cent, phosphorus, and the period of refining was consequently pro- longed. The radiation loss of the furnace was also very high in relation to the total available power, so that the high power con- sumption was to be expected. 1 British Patent Specification, No. 7027, A.D, X903, 26 THE ELECTRO-METALLUKGY OF STEEL The single-phase rectangular furnace was followed later by the three-phase furnace of circular form, and with this modifi- cation the ultimate development of the Heroult furnace was reached, as characterised by its essential features. At the same time that Heroult was working on the problem of making low carbon alloys, another investigator, Charles A. Keller, was similarly engaged. The early work of the latter, which led to the issue of a patent 1 in France in 1900, was briefly described by the inventor himself in a paper delivered before the American Electro-Chemical Society in 1909. 2 From the description, the furnace closely resembled that patented by Heroult in March of the same year embodying : I. The use of two separate heating zones at the upper sur- face of the conductive charge. II. Voltage regulation of each electrode. III. Arrangement of the electrodes in series. IV. Construction of the furnace to obtain products by the tapping method. Keller experimented with a furnace of 15 cwts. capacity up to the year 1902 at the Kerrouse Works (Morhiban, France), more especially for the manufacture of steel from cold scrap. The ingots were tested at different works near St. Etienne and, following the favourable results obtained, a larger furnace of 2^ tons capacity was installed at the works of the Societe des Etablissements Keller Leleux at Livet (France). This furnace was likewise provided with two electrodes and was experimented with for three years, during which time the metallurgical problems and products made were investigated by J. Holtzer & Co., of the Unieux Steel Wo rks^ (Loire). Prompted by further success, it was decided to install a furnace of 8 to 10 tons capacity at the latter works, several modifications of interest being incorporated in the design. The furnace body (Fig. 23), consisting of a steel shell lined with a suitable refractory material and mounted for tilting on rollers, was entirely independent of the electrode carrying gear. A novel feature was introduced by interleaving the bus bars and bringing them above and then over the furnace body, where they terminated in a special 1 French Patent, No. 300,630, 23/5/1900. 2 "Am. Electro-Chem. Society," Vol. XV, 1909. HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 27 distribution block. The bus bars of different polarity were there divided, and each group provided with two separate sets of lugs, so arranged that the flexible connection to the electrode holders might be rapidly and securely made when changing electrodes. The latter were suspended from jib cranes, which swung into a position over-hanging the furnace roof. The 28 THE ELECTRO-METALLURGY OF STEEL furnace operated on a single-phase supply, and the load circuit contained two pairs of arcs in series in place of two single arcs, as in the Heroult furnace, the line voltage being equally divided between each pair of arcs. The two electrodes of each group were capable of individual movement and electrically connected in parallel. There were, then, four distinct arcs, one pair of which was in series with the other pair, whilst each pair con- sisted of two arcs in parallel. The regulation was necessarily complicated, any inequality of current flowing through the two electrodes of either pair being adjusted by the simultaneous lowering of one and raising of the other ; balance of voltage between each pair and the bath was likewise effected by raising one pair and lowering the other. Other combinations of move- ments were also made possible by a system of valves operating hydraulic cylinders. The furnace at Unieux was used for refin- ing open hearth furnace steel already low in phosphorus but high in sulphur. The power consumption per ton of steel averaged 275 kw.-hours, a load of 750 kw. being taken for about 2| hours. The experimental work at Livet done prior to the erection of this large furnace at Unieux was concluded in 1905, and in the meantime led to the issue of several French patents. I. French Patent, No. 300,630. 23/5/1900. "Electric fur- nace improvements." II. French Patent, No. 322,700. 2/7/02. " Process for melt- ing and refining metals and other substances electrically." This invention was also patented in England l and described as follows : " Metals or other substances are electrically heated for re- fining or other purposes by current passing between electrodes E (Fig. 24) dipping into the materials. Preferably several electrodes are used, so that one may be changed without stopping the operation. The heating may be effected in a furnace G, arranged to receive the product of a cupola furnace F : or in an ordinary foundry ladle, which is carried on a movable truck so that it may be charged from one or more furnaces and then placed under vertically movable electrodes. Metal may thus be collected and kept hot for casting, and other materials may 1 British Patent Speciacation, No. 15,271, 1902. HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 29 be added to effect refining ; waste iron or steel may be added and steel produced." The specification clearly states that the electrodes dip into the "material" (slag in the case of steel-making), and are ar- ranged in two groups of opposite polarity, each group being built up of two or more electrodes separately movable and con- nected in parallel. FIG. 24. III. No. 329,013. 2/2/03. " Improvements in the process of electric melting and refining." The feature of this patent lies in the use of free arcs formed between the electrodes and the slag, as opposed to the electrodes dipping into the slag and heating purely by resistance. In this way alone can an oxidised slag be maintained. Oxidising and dephosphorising could be performed under a suitable slag and then followed by a refining period under a different slag. IV. No. 387,461. 6/5/07. "Process for carburising a liquid metal." 30 THE ELECTRO-METALLURGY OF STEEL Composite blocks are made by pouring a metal into a mould previously filled with small pieces of broken carbon. Such blocks when thrown on a bath of metal will sink by their weight, into close contact with the metal to be and bring the carbon FIG. 25. carburised. In the case of steel, cast-iron is the metal used for making the blocks. V. No. 393,740. 4/11/07. " System of conducting hearth for electric furnaces " (Fig. 25). x This invention relates to the use of a conducting hearth, composed of a refractory material suitable for steel manufacture 1 " Am. Electro-Chem. Society," Vol. XV, p. 98. HISTOKICAL DEVELOPMENT OF ELECTKIC FURNACES 81 and capable of carrying the full load current. It will be re- membered that a carbon bottom, serving as one electrode, had been used prior to 1900 for the manufacture of ferro-alloys, but was abandoned by Heroult for the reason of carbon contamina- tion in cases where low carbon alloys or steel were to be made. On returning to the use of a conducting hearth, it was necessary to employ a material free from carbon, which was at the same time sufficiently conductive to carry the full current when cold. For this purpose Keller used a composite bottom, consisting of a number of iron rods connected to a fixed bottom plate and embedded in magnesite or dolomite, mixed with a tar or pitch binder rammed in to form a solid plug. The metallic portion of the hearth was relied upon to complete the circuit when cold. The distribution of the current was equalised over the entire section of the hearth, and in this respect differs from the metallic bottom electrodes used by Girod about three years earlier. VI. No. 400,461. 6/6/08. " Process for the exact carburis- ing of steel." This method consists in the immersion of a carbon block into the bath of steel in such a manner that the exact loss in weight by absorption in the bath is accurately recorded as carburising proceeds. VII. No. 399,643. 27/4/08. " Eadiating distribution of con- ductors for multiple electrode furnaces " (Fig. 26). l This patent refers to a method of interleaving the bus bars of opposite polarity and bringing them to a central point of dis- tribution, whence the current flows through suitable radiating connecting lugs to the four electrodes, as previously described for the 8 to 10 ton furnace built at the Unieux works. VIII. No. 400,655. 15/6/08. " System of regulating the circuits supplying multiple electrode furnaces." This method was embodied in the four-electrode furnace as described, and consists of voltage and current regulation of the electrodes in series and parallel respectively. IX. No. 14,728/393,740. 21/11/10. Patent of Addition. " System of a conducting hearth for electric furnaces." 1 " Am. Electro-Chem. Society," Vol. XV, p. 114. 32 THE ELECTRO-METALLURGY OF STEEL In this case, the composite bottom previously described is replaced by ramming the entire hearth with a mixture of magnesite and iron filings, bound together by tar or pitch. For operation on a three-phase supply, Keller proposed to connect the three electrode circuits either in delta or star, when, in the latter case, the conducting hearth is made the neutral and con- nected to the star point of the transformer secondary circuits. FIG. 26. In a general survey of Keller's inventions it will be seen that for seven years his efforts were restricted to furnaces employing suspended electrodes operating in series, the bottom electrode or conducting hearth being only adopted towards the end of 1907. Paul Girod had already in 1905 utilised a, modification of the old carbon bottom for melting steel, and his contribution to the general study of electric steel-making must next be mentioned. HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 33 Girod had been engaged for many years in the use of electric furnaces for the reduction of ores before turning to steel melt- ing. His first invention is particularly interesting, as it was the first application of Siemens' original principle of direct arc heating to the commercial manufacture of steel. Heroult and Keller had overcome the difficulties due to a carbon electrode bottom by eliminating it, whereas Girod preferred to construct a conductive hearth which would prevent all possibility of carbon contamination. The successive steps and improve- ments, evolved from the time of his first studying the subject, will be again best followed by enumeration of the several patents issued in France. I. French Patent, No. 350,524- 4/1/05. "Electric furnaces." The nature of this invention is concisely described in the resume of the specification of which the following is a trans- lation : " An electric furnace in which one of the poles is formed by one or several graphite electrodes, moved mechanically or by hand, the other pole being constituted of several electrodes buried in the furnace lining, and situated at such distance from the hearth, or cooled artificially in such fashion, that at their extremities a certain quantity of fused material is chilled and forms upon them a solid protective layer which prevents con- tact of these electrodes with the metal being manufactured. Contact between the latter and the movable electrode is avoided by the fact that the said electrode dips only into the layer of slag which covers the metal on melting." In the subject-matter of the specification further details are given of the manner of constructing the bottom conductors (Fig. 27). The actual poles, which may be of metal or graphite, are usually water-cooled, and, if insulated, may be connected in parallel or in series to the bus bars. In the case of graphite, the poles are covered at their upper extremities with a metallic pole piece, which either partly melts or becomes enlarged during operation of the furnace. When the furnace is circular in shape the poles give place to an annular channel, which is filled with cast-iron and similarly 34 THE ELECTRO-METALLURGY OF STEEL water-cooled ; this particular construction was especially sug- gested for making steel from pig-iron (Fig. 28). The furnace Fm. 27. w//////////// FIG, 28. can be either fixed or constructed for tilting, and can operate on either continuous or alternating current supplies, "With proper HISTORICAL DEVELOPMENT OF ELECTEIC FURNACES 35 choice of voltage the electrodes will form arcs above the slag, when the known processes of decarburising, dephosphorising and desulphurising can be conducted. Preference is given to the use of a non-conducting material surrounding the pole pieces or conductors. In the single-phase design the current from the upper central electrode is forced to take a diagonal path to the two lower electrodes, causing circulation of the metal. In 1902 Heroult filed a patent application in Belgium which covered the use of carbon-bottom electrodes with metallic end pieces to prevent carbon contamination of the bath. Girod, therefore, was not the first to conceive the idea of such bottom electrodes, but was the first to put it to commercial use. II. French Patent, No. 350,802. 16/1/05. " Tilting furnace with conducting hearth." In his first patent, Girod suggested that his furnace might be constructed to tilt. In this patent a specific method of tilting the furnace on trunnions is described, and the provision of two tapholes at different levels is mentioned. III. Patent of Addition, No. 4829/350,524. 4/1/05. "Electric furnaces." In place of the annular conducting ring previously patented, the conducting poles may be formed by leaving channels in the furnace lining extending upwards from the shell plate to the level of the working bottom. These channels are filled up with solid lumps of metal, which, during a preliminary operation of the furnace, become fritted together to form a solid pole. IV. No. 388,614. 4/6/07. " Process for the manufacture of iron and steel in the electric furnace." This process requires the use of at least two electric furnaces, one being employed to melt and refine cold scrap, the other of much smaller capacity being used solely to finish the oxidised and dephosphorised steel from the former. The large melting furnace is never emptied during continuous operation, sufficient steel being retained to take up an additional charge of cold scrap equal to the quantity transferred to the smaller furnace for finishing. This process saves the time and labour of skim- ming off the oxidised slag prior to finishing, and, at the same 36 THE ELECTRO-METALLURGY OF STEEL time, obviates the loss of heat when charging cold scrap into an empty furnace. V. No. 402,758. 6/5/09. "Process of refining liquid steel from furnaces other than electric." Purposely under-oxidised steel from an open-hearth furnace or converter is transferred to the electric furnace, and there cooled down to promote evolution of dissolved gases (hydrogen and nitrogen). This is followed by again raising the temperature, when a further elimination of C, Si, Mn and P is effected in the ordinary manner under a basic slag. VI. No. 11924/402,758. 7/12/09. " Patent of addition." The process is similar to the above, but is carried further by successive raising and lowering of temperature, the evolution of dissolved gases being assisted by a small addition of carbon before reducing the temperature. This addition is also made after dephosphorising. VII. No. 12771/402,758. 15/6/10. "Patent of addition." This patent has for its object the removal of dissolved oxides together with the gases. A small quantity of C, Si, and Mn is left in the steel before transference to the electric furnace, and on reduction of temperature reacts with the dissolved oxides to form fusible insoluble silicates, which readily fuse and rise up- wards through the bath. C, Si, and Mn may be actually added to promote this reaction. VIII. No. 416,927. 9/6/10. " Method of arranging electric furnaces for three-phase working." This patent refers to a special method of connecting the three upper electrodes in star fashion, the bottom electrode being connected to the neutral point of the secondary circuits. One phase of the star connection is reversed, which compels the major portion of the current to flow through the bottom elec- trode, while the load is nearly equally balanced on the supply phases (Fig. 61). IX. No. 422,717. 26/8/10. " Method of arranging electric furnaces for three-phase working." Three methods are given of connecting a furnace, provided with two upper electrodes and a bottom electrode, to operate HISTORICAL DEVELOPMENT OF ELECTEIC FUENACES 37 on a three-phase supply system. In each case the low tension supply to the furnace is actually two-phase. Up to the year 1905 the only direct arc furnaces used com- mercially for steel-making belonged to that class in which the electrode circuits were independent of the furnace lining. The advent of the Girod furnace in that year, and the practical re- sults achieved, led others to modify the hearth construction and to adopt it for use with different systems of electrical connec- tions. Girod restricted the construction of his conducting hearth to metallic or graphite poles, embedded in a material which was preferably a poor conductor at high temperatures, and therefore relied entirely upon primary conductors for com- pletion of the circuit through the charge and furnace hearth. Keller in 1907 modified the above method by using a composite bottom as already described, and the value of a conducting material, consisting of magnesite with a carbonaceous binder, was fully realised by him although used in conjunction with metallic rods as primary conductors. There was, however, considerable prejudice against the use of metallic conductors embedded in a furnace bottom when directly in contact with the molten bath, and it was suggested that considerable difficulty might be occasioned in repairing a bottom between heats, owing to partial emptying of the holes which normally constituted the metallic electrodes. This pre- judice was not always borne out in practice. With a view, no doubt, to avoid such difficulties, a furnace was constructed at the Firminy Steel Works in France and a patent applied for in March, 1908. The hearth had no metallic conductors, and was built up of successive layers of a refractory material, such as magnesite or dolomite, mixed with a carbonaceous binder in decreasing quantities towards the upper surface. It was claimed that a highly conducting hearth could be built up in this manner without fear of carbon contamination of the bath, and with a possibility of restarting a furnace when cold. This type of hearth construction has now been generally adopted in preference to the metallic electrodes used by Girod and Keller, and of modern furnaces using conductive hearths the Electro- metals design was the first to gain a wide application. The early forms of single-phase arc furnace usually required 38 THE ELECTRO-METALLURGY OF STEEL a special generating plant, owing to the impossibility of trans- forming from either two- or three-phase supplies with static transformers, and it was to overcome this fundamental objec- tion that the Electro-metals design was introduced ; the fur- nace operated on a two-phase low tension supply, transformed from either a two-phase or three-phase high tension system. The original patent l was applied for in Sweden in August, 1908, and embodied the use of one upper electrode in each phase circuit, which also included a conductive hearth and a common neutral return conductor, the latter being connected from the hearth to the neutral point of the two phases (Fig. 46). A later 2 patent describes an improved form of rectangular tilting furnace ; the electrode holders are mounted on swivel brackets, and so can be raised and swung round clear of the roof. The electrode regulating motors impart movement to the electrode brackets through a system of screw feed and tele- scopic shafting, and so can be situated away from the furnace body. This construction (Fig. 29) is not, however, adopted in modern types. In 1911 a modification of the Electro-metals furnace was introduced by Stobie, 3 who substituted a return conductor for each phase in place of the common return. The two phases were thus separated and constituted a four-wire two-phase system in place of a three-wire neutral return system ; the bottom elec- trode of each phase was situated diagonally opposite its corre- sponding upper electrode in place of below it (Fig. 49). In a more recent furnace 4 of larger capacity the two-phase bottom electrode type gives place to a four electrode three-phase star- FIG. 29. Electro-metal tilting furnace. 1 Swedish Patent, No. 28,687, 1/8/08. 2 British Patent, No. 12,430, 21/5/09. 4 Ibid., No. 2081, 1912. Ubid., No. 6741, 1911. HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 39 connected type, in which one electrode is connected to the star-point of the three-phase system and serves as a return for any unbalanced current. An ingenious application of two-phase current to indirect arc furnaces of the Stassano type was introduced by Renner- felt 1 in 1912 with the object of reducing the excessive wear of the roof and lining. In this furnace two arcs are formed be- tween the horizontal phase electrodes and a vertical neutral, and are deflected downwards by mutual repulsion of the magnetic fields set up. The value of small electric furnaces for the manufacture of light castings had been well proved, and other new types were introduced to meet the special requirements of the foundry. The Snyder furnace, 2 together with the Rennerfelt furnace just alluded to, may be regarded as falling in this category. The furnace operates on a single-phase supply, and is provided with one top and one bottom electrode, the latter being em- bedded in the hearth and in direct communication with the metallic charge or bath. The chief feature lies in the use of a high arc voltage, usually about 110 volts at full load, which produces a long arc from which the roof and upper lining of the furnace is only slightly shaded by the electrode itself. By using a high voltage arc the current required for a given power input is correspondingly small, which entails a saving in the cost of electrical equipment. It is also claimed that a charge of steel scrap may be melted quicker, and this is now generally accepted as being true, provided the electrical equipment is at the same time designed to prevent heavy current fluctuations. The Snyder furnace, designed on this principle, enables a very steady load to be maintained even when melting irregular shaped scrap, but this can only be done at the sacrifice of power factor which is considerably lowered by increasing the reactance of the circuit. The construction of the original furnace is shown diagrammatically by Fig. 30. The original specification lays stress on the special construction of the fur- nace shell, which is composed of a number of laminated steel sheets to prevent the heating effect of eddy currents, and at the same time provides the necessary reactance of the circuit. 1 British Patent Specification, No. 7367, 1912. 2 76id., No. 25,171, 1913. 40 THE ELECTROMETALLUKGY OF STEEL In the case of the conductive hearth furnaces so far con- sidered, the current flowing through the hearth has been fixed as a definite proportion of the total current flowing through the upper electrode circuits. This has been considered an objection by some, and attempts have been made to arrange the electrical installation and the method of furnace connections so as to admit of varying at will, within certain limits, the amount of current flowing through the conducting hearth. The first move in this direction, was made by Dixon, whose earliest patent application l was filed in 1914. The furnace was provided with four or six upper electrodes and one bottom electrode. By means of switches in the high tension circuit the phase relations of the low tension transformer circuits could be altered to vary the proportion of the current flow- ing through the conducting hearth ; the same result could also be ac- complished by reversing one or more of the transformer connections, which had been suggested by Girod in 1911 in the case of three- phase supply. Two later patents 2 referred to special arrange- ments of a two-phase low tension supply suitably connected to four electrodes so as to render equal distribution of the load between the four electrodes possible by hand or automatic con- trol depending upon current variation only. A simple single-phase indirect-arc furnace was introduced in Italy in 1914 by F. Bassanese and has there acquired con- siderable popularity. Certain unusual features were presented by the method of electrode adjustment, but in general appear- ance the furnace resembles the revolving steel melting furnace of Stassano. Another furnace of more simple design, but presenting the same feature of a variable bottom electrode current as first conceived by Dixon, was introduced by T. H. Watson & Co. 1 British Specification, No. 4742, 1914. *Ibid., No. 111,103, October, 1916; No. 111,104, October, 1916. FIG. 30. Snyder furnace. HISTORICAL DEVELOPMENT OF ELECTRIC FURNACES 41 (Sheffield), and is better known as the Greaves-Etchells furnace. According to the first patent 1 the furnace could operate on either a three-phase or two-phase supply system, and further by the use of only two upper electrodes a good balance of load could be obtained on the primary supply phases, where static transformers are used. In one case, when the current through the two upper electrodes is equal, the voltage in the secondary windings connected to the lower electrode can be varied to produce the required balance. Further modifications 2 were later made in the use of three-phase low tension circuits for supplying power to the furnace. In conclusion, mention must be made of the Booth-Hall furnace, which combines the essential features of the Eennerfelt and original Stobie furnaces. In this historical review it has only been possible to mention those types of furnaces which are either in use at the present day, or which have in the past de- monstrated certain principles that are now the foundation of more widely known and later types. 1 British Specification, No. 106,626, March, 1916. 2 Ibid., No. 118,233, 1918; No. 121,563, 1918. CHAPTEE II. ELECTEICAL DEFINITIONS. I. Alternating 1 Current Supply. A source of electrical energy is " alternating " when the voltage passes through periodic variations of magnitude, together with reversal of sign. II. Alternating Current. If a source of alternating electrical energy is connected to a suitable circuit, a current will flow the magnitude and direction of which will, at all moments, vary directly as the voltage. III. Wave Form or Wave Shape. The wave form of an alternating current is found by plotting a curve with instan- FIG. 31. taneous values of current as ordinates against corresponding time values as abscissae. The wave form of the impressed voltage may similarly be plotted. Owing to the characteristic design of alternating current generators, the wave form approximates to that of a sine wave, which can be developed by simple geo- metrical means. Kef erring to Fig. 31, imagine a point P revolving on the circumference of a circle AYBXA at a uniform speed and in a clock-wise direction from X through AYB, and back again to X; then, by plotting the vertical distance of the point P from the horizontal diameter XY against time, or corresponding (42) ELECTRICAL DEFINITIONS 43 angular displacement from a time axis XY, a curve XAYBX will be obtained. IV. Cycle. Kef erring again to Fig. 31, the revolving point P, starting from any position with a uniform velocity, will have performed one complete cycle when it has returned to its start- ing point, having made one complete circuit. One cycle is therefore completed in 360 of angular displace- ment. Used in a purely electrical sense, a cycle is regular and periodic. A " cycle " is represented by the symbol <. V. Electrical Degree. One 360th part of a cycle. VI. Period. The time required to complete one cycle. VII. Frequency or Periodicity. The number of cycles per- formed in one second. FIG. 33. In electrical equations frequency is denoted also by the symbol -^. VIII. Phase. If a point P revolves about a centre (Fig. 32) then the phase of that point, relative to any time axis AO, is its angular displacement from that axis. Again, if two points P and P' revolve about the same centre O with the same frequency but not necessarily equal radii, and if the angle between OP and OP' equals <, then the points P and P' are said to be " out of phase ". By plotting two wave curves for these points P and P' (Fig. 33), representing a complete cycle and beginning at simul- taneous .moments when these points are at given vertical distances from XY and out of phase, it will be seen that the wave curve for point P is always out of phase with the curve for P'. In this particular instance the plotting of the wave curves has been commenced at a moment when the point P is 44 THE ELECTEO-METALLUEGY OF STEEL and the point P'(0 + ) past their maximum positive value at A. These wave curves might equally well have been obtained by plotting instantaneous values of voltage against time for two separate sources of alternating current of similar frequency, but out of phase with one another and having voltages of differ- ent maximum values. It will be shown later how two or more alternating currents of the same frequency, but out of phase, may be combined to produce a resultant effect in the same way that forces of equal or unequal magnitude, acting on a body at different angles, may be compounded by graphical methods to produce one resultant force. IX. Root = Mean -Square or Effective Values of Voltage and Current. It has been stated (see Def. II.) that, when a suitable closed circuit is connected to an alternating current supply, a current will flow whose magnitude and direction at any moment will depend upon the instantaneous value of the voltage as regards both magnitude and sign. Since the voltage across the terminals of such a closed cir- cuit undergoes periodic variation of magnitude and reversal of sign, it may be represented by a sine wave curve ; likewise the current flowing through the circuit may be similarly represented. Now the physical effects of any electric current are generally due to the power developed, which is as will be seen later proportional to the -product of current and impressed voltage. It is necessary, therefore, for purposes of measurement and cal- culation, to know the value of the average current or voltage which would, in any half cycle, produce exactly the same effect as the sum of all the instantaneous current or voltage values, which rise from zero to a maximum positive or negative value and then fall back again to zero. Again, since power is proportional to C x E, and either C or E can be expressed in terms of one another and circuit p resistance by the equation C = =~, it follows that the physical XV effect produced is proportional to the square of the current and likewise to the square of the voltage. ELECTEICAL DEFINITIONS 45 i.e. if the power W (watts) = C x E then W = C x CK (R = resistance of the circuit) = C 2 R and again, if power W = C x E then W = 1 x E XX = E 2 R' From this reasoning it follows that the mean effective value of an alternating current or voltage is proportional to the square root of the mean of the instantaneous values squared, and not to the mean of the instantaneous values themselves. For any alternating current supply whose voltage may be represented by a simple sine wave, the effective values of voltage or current flowing through a connected circuit are equal to the maximum or crest values -r- ^2. Ammeters and voltmeters always indicate the effective values. X. Lead and Lag. Suppose Fig. 33 represents the voltage or current curves for one complete cycle of two alternating currents of similar frequency, but differing in phase. The maximum values of the curve PT' occur exactly cf> before those of curve PP and are said to " lead " by <, whereas correspond- ing values of PP " lag " behind PT'. If the curves PT' and PP represent the voltage and current waves respectively of an alternating current, then the current is " l a *?g m g " $ behind the voltage, and the " angle of lag" is ; similarly the current may sometimes "lead" the voltage and then have a certain " angle of lead". XI. Induction and Induced Currents. Whenever a current flows through a conductor a magnetic field is produced, which, according to the usual convention, is said to contain a certain number of magnetic lines of force, whose number and direction will depend upon the magnitude and direction of the current and the permeability of the medium through which the magnetic lines of force pass. The conductor, as shown in Fig. 34, may be either straight or coiled, and will produce magnetic fields in the directions indicated when carrying a current likewise shown. Conversely, if a conductor lies in a magnetic field of varying intensity and in a position other than in the exact direction of 46 THE ELECTEO-METALLUEGY OF STEEL the lines of force, then a voltage will be induced within it, the magnitude and sign corresponding at any moment to the rate of change in the number and direction of the magnetic lines of force cutting the conductor; no voltage is induced so long as the magnetic field remains unchanged. Such a phenomenon is known as " induction," and, accord- ing to convention, 1 volt will be induced when a conductor cuts magnetic lines of force at the rate of 10 s lines per second. Supposing now that a magnetic field of varying intensity and direction is set up in an iron ring B by a coiled conductor A carrying an alternating current (Fig. 35), then a voltage will be induced in the coiled conductor C, which, if made into a closed circuit, will carry an "induced current" alternating FIG. 34. FIG. 35. at the same frequency as the current in the coil A. The con- ductor that sets up the magnetic field is generally known as the " primary winding," and that in which the voltage is induced as the " secondary winding ". The induced or secondary volts will bear the same ratio to the exciting or primary volts as the number of turns of the respective windings. Each turn may be regarded as a separate conductor, so that increasing the number of turns of the secondary is equivalent to adding further induced voltage in series, and consequently raising the total voltage for the same magnetic field. If a certain amount of power is utilised in a primary winding, it is obvious that only the same amount of power can be de- veloped by the secondary, even assuming 100 per cent, efficiency of conversion. Therefore, since the primary and secondary ELECTEICAL DEFINITIONS 47 voltages are proportional to the respective number of turns, the primary and secondary currents will be inversely proportional, so that the product of volts and amperes or volt-amperes is equal for both circuits. This constitutes the underlying prin- ciple of step-down transformers, to which type all induction furnaces belong. XII. Self-induction, Reactance, or Reactive Resistance. When an alternating current flows through a conductor, either straight or in the form of a coil or solenoid, it will produce a magnetic field constantly changing, as regards both the number of lines of force and reversal of their direction. As explained in Def. XL, such a conductor would be under the influence of the changing magnetic field, and an alternating voltage would therefore be induced within it. The simple wave form of an alternating current shows that the maximum rate of change of current, and, consequently, of the number of magnetic lines of force produced, takes place at the steepest point of the curve, which is just at the moment of reversal. Although there are no magnetic lines of force at the exact moment of reversal, the rate of diminution or increase in number is then at a maximum, and at this point, therefore, the " self-induced " or " reactive " voltage reaches its highest value. It follows, therefore, that the wave forms of the current flowing and the self-induced voltage, if plotted simultaneously, would be 90 out of phase and have the same frequency. The result of such self-induced voltage, which is out of phase with the applied voltage, is to reduce the power of the latter to force current through the conductor, and at the same time to cause the wave curve of the current flowing to lag behind that of the applied voltage. This latter effect is important in con- nection with " wattless current " and power factor. The self- induction, reactance, or reactive resistance of a conductor or circuit is, then, a property by virtue of which it offers resistance to the flow of alternating current. Such resistance is in no way due to the material of which the conductor is composed, but to characteristics which promote the generation of a self- induced or reactive voltage as above described. The self-induction or reactance of any given circuit is equal to 2-7T ^> L ohms. The co-efficient of self-induction or inductance 48 THE ELECTRO-METALLURGY OF STEEL (symbol L equals one Henry) is present when one ampere produces 10 8 interlinkings of lines of force and windings (i.e. number of turns x lines of force per second). The number of lines of force or the magnetic flux produced by a current of one ampere may be calculated from various formulae according to the nature of the electric and magnetic circuits. XIII. Reactance Voltage or E.M.F. of Self = Induction. The self-induced E.M.F. or voltage may be calculated as the product of current and reactance, and is equal to 2?r > LC. XIV. Percentage Reactance Drop. The term " percentage reactance drop" is used for expressing the reactance voltage in a circuit as a percentage ratio of the voltage applied to the circuit, usually at normal full load current. XV. Wattless Current. When the voltage and current curves of an alternating current are in phase, the entire " effec- tive " or root-mean-square current value is useful for doing work as electrical energy. This is not true for a current which "lags " or " leads" the voltage, when only that component of the cur- rent which is in phase with the voltage is capable of doing useful work, the other component being "wattless ". It should be here mentioned that a wave curve of current or voltage may easily be resolved graphically into two compon- ents, and in the same way two waves of similar frequency, but differing in phase or magnitude, may be compounded to produce one wave curve. XVI. Volt = ampere. The product of the effective or measured value of volts across a circuit and the effective current value in the circuit is expressed in terms of volt-amperes (V.A.) or, when divided by 1000, of kilo-volt-amperes (K.V.A.). XVII. Power Factor. The power factor of a circuit in which electrical energy is transformed or dissipated, is the cosine of that angle usually denoted by $ by which the current " lags " or " leads " the impressed voltage. Power factor is of great importance in all electric furnace installations, and, except in certain cases where considerable circuit reactance is purposely introduced, should approach unity as far as possible at normal loads. ELECTEICAL DEFINITIONS 49 The power factor of a circuit indicates the ratio of useful current flowing to the total current, and since the capacity of transformers and generators is limited by the permissible cur- rent, it is obvious that the power factor represents the ratio of actual power output to the maximum possible output. In other words, if the power factor is *6, then only '6 of the power plant capacity is available for doing useful work. XVIII. Watt and Kilowatt. The true power absorbed in a circuit is measured in terms of watts or kilowatts, and is the product of volt-amperes or kilo-volt-amperes and power factor, i.e. Watts = V x A x cos < (where < is the angle of "lag" or "lead"), Kilowatts = V x A x cos < -=- 1000. If the current and voltage wave forms are in phase, then cos = 1, and the volt-amperes are equivalent to watts. XIX. Surging. Strictly speaking, the term "surging" should be used only to express a more or less constant and periodic current or voltage fluctuation, but for furnace loads may be more loosely used to denote persistent current instability or fluctuation of considerable magnitude. In the case of direct arc furnaces it sometimes happens that the nature of the charge is such that a steady load is most difficult to maintain. This con- dition is most likely to occur with a loosely packed charge of heavy and irregular shaped scrap, and results from the con- stantly varying resistance of the charge. This is due to local fusion of the metal by the arcs formed at the various points of contact, which may either facilitate or interrupt the passage of current. When the resistance of a charge of scrap is very low, slight movement of the electrodes may be sufficient to cause sudden rushes of current, which may at times reach values 100 per cent, in excess of the normal full load current. Surging is sometimes experienced even when the entire charge is molten, and is then probably due to a phenomenon known as " pinch effect," which occurs in the slag covering ; this, however, only happens when the electrode is almost, if not actually, in contact with certain basic and acid slags. Surging is highly objectionable for many reasons, some simple and some very complex. So long as a current is violently 50 THE ELECTRO-METALLURGY OF STEEL fluctuating it is only possible to prevent it reaching an in- stantaneous value, large enough to automatically open the main switch, by operating the furnace at a considerably reduced aver- age load. The installation would not then be working at its full capacity, and reduction of output accompanied by an in- crease in power and other costs would result. Surging or cur- rent fluctuation can to a great extent be reduced, if not prevented, by increasing the self-induction of the load circuit, which can be done by the aid of reactance or choking coils. XX. Single=phase Alternating Current. The alternating current flowing through a single circuit by virtue of an alternat- ing voltage is known as a single-phase current. A source of single-phase current can only have one wave curve for its im- pressed voltage. (See also Def. I.) XXI. Two-phase Alternating Current. A so-called " two- phase current" denotes the availability or application of two sources of single-phase current of similar frequency and voltage magnitude, but generally differing in phase by 90. This par- ticular phase displacement permits of a useful combination of two such single-phase currents flowing through separate circuits for different purposes. The two wave curves of the individual' impressed voltages are represented in Fig. 36. Such alternating currents are also said to be in " Quadrature," because the phase displacement is one-quarter of a cycle. 90* FIG. 36. XXII. Three=phase Alternating Current. A so-called "three-phase current" denotes the availability or application of three sources of single-phase current of similar frequency and voltage magnitude but differing in phase by 120. Again, this phase displacement permits of a variety of combinations of three such single-phase currents flowing through separate circuits, which may be for lighting, heating, power generation or con- ELECTKICAL DEFINITIONS 51 version to two-phase and even four-phasfe current. The wave curves of the individual impressed voltages are shown by Fig. 37. FIG. 37. XXIII. Load Factor. This term must not be confused with "power factor". The percentage ratio of average power de- mand to the maximum power demand, during any given period of time, is the " load factor " of the installation during that period. If a furnace installation, which has a maximum power absorbing capacity of 1000 kw., is actually supplied with an average of only 800 kw. during a certain period, then the load factor is 80 per cent, over that period. The load factor is of the very greatest importance in furnace work, as it is only by working as nearly as possible at the full plant capacity that the maximum economy of production can be attained. The load factor of an installation is sometimes averaged over a period of a month or year, which will include all delays and stoppages. In any case, when referring to load factor, the period upon which it is calculated should be clearly stated, when not otherwise understood. CHAPTEE III. APPLICATION OF SINGLE AND POLYPHASE CUKRENTS TO FURNACE OPERATION. Application of Single-phase Current. It has been shown how a single-phase current may be produced in a circuit by a single impressed voltage that is passing through periodic changes of magnitude and sign. An outgoing current must have a re- turn to complete its circuit, otherwise no current flows, so that at least two main leads are required to supply energy to a furnace from any source of single-phase current. The nature of the circuit and the manner in which the power is absorbed depends upon the particular furnace design. The various FIG. 38. FIG. 39. methods of using single-phase current for arc furnaces from a two-wire supply are shown in Figs, 38, 39, and 40, which also illustrate the principles of direct and indirect arc heating used in certain modern furnaces. Fig. 41 illustrates in plan the heat developing and power supply circuits of a single-phase induction furnace, A being the laminated transformer core in section, B the primary windings from the main supply leads, and C the short-circuited secondary coil in which the low tension current is induced. A three-wire system was used in the Giffre furnace, the (52) SINGLE AND POLYPHASE CURRENTS TO FUENACE OPERATION 53 third wire being connected from the middle terminals of two single-phase alternators to a bottom electrode, as shown in Fig. 42. The single-phase alternators, mechanically coupled together, are connected in series and generate synchronously at the same frequency and voltage. If the terminal voltage of each alternator is 50 volts, then the arc voltage will be 50, FIG. 40. FIG. 41. assuming the two arcs are balanced. If the electrodes are at any time unbalanced, then a current will flow through the bottom electrode back to the middle terminal. Application of. Twophase Currents. It is more economical to generate and transmit two-phase current, owing to : (i) The reduced cost of the generating plant. (ii) The saving of copper in conductors. FIG. 42. It has been explained that a two-phase system is simply a suitable combination of two single-phase currents, 90 out of phase, which can be readily compounded to produce a resultant current in a common return conductor. A two-phase supply, then, might constitute two entirely separate single-phase circuits with impressed voltages of similar frequency and magnitude, but 90 out of phase. If the separate currents were utilised 54 THE ELECTRO-METALLUKGY OF STEEL in a machine or furnace and kept distinct throughout, it is obvious that four wires would be required to transmit the two currents. Two such circuits A and B, in which single-phase currents are being generated 90 out of phase, are represented in Fig. 43. If, for example, the voltage across A is at its maximum positive value of 100 volts at terminal D and just beginning to fall off, then the voltage across the terminals of B, if lagging by 90, will be zero and just reversing to a positive sign at one terminal, say D'. In practice it is customary to connect one terminal of each circuit together, and to connect the remaining " outer" terminals to two corresponding terminals of the power absorbing circuits. The latter are likewise connected together at a common terminal, which is connected to the corresponding V V V V V V D ' VV V VV V C* D' 1 +IOOV FIG. 43. FIG. 44. terminal of the generating circuits by a common return con- ductor (Fig. 46). In this way only three conductors are used in place of four, which constitutes a saving of copper,, although the "neutral" return conductor carries a heavier current than that flowing through the individual circuits. In Fig. 44 the heavy dotted line EFGH shows the resultant current wave of the two current waves CD and C'D' for the circuits A^nd B in Fig. 43. The effective value of this curve equals N / 1 2 its crest value, and is ~- times the sum of the effective current values of the wave curves CD and C'D'. Keferring again to Fig. 43, if the terminals C and D', or C and C' were connected together to form the terminal of the return conductor, the resultant effect for furnace work would be the same. It should be carefully borne in mind that such inter- SINGLE AND POLYPHASE CUEEENTS TO FUENACE OPEEATION 55 changeability of terminal connections does not apply in the case of three-phase circuit combinations. It is equally possible to compound two single-phase circuits which differ in phase by 60 in place of 90. Two such single- phase circuits may be obtained, for example, by taking current from two separate phases of a three-phase system, as was sug- gested by Girod in 1910. The various systems of two-phase connections adopted for furnace operation will now be mentioned in the order in which they have been successively introduced. Heroult Two- phase System. It was proposed in 1905 to use two-phase current for supplying power to the electric mixing furnace patented in that year, in which case the circuits would be arranged as shown by Fig. 45. As a three-phase system presented considerable advantages over the. two-phase, the latter was never applied in practice, but is in- teresting in view of recent furnace develop- ment. Electro -metals Two -phase Three -wire System. The two-phase low tension cir- cuits are represented in Fig. 46 by the windings A and B drawn at right angles, according to convention, showing them to be 90 out of phase. The two phases are connected at the common neutral point 0, from which a neutral conductor goes to an electrode in the furnace bottom. Each upper electrode is connected to the outer terminal of one phase- The hearth only becomes a conductor when hot, and will then carry practically the full return current equivalent to ,/2 of the current flowing through each electrode when balanced. When cold the bottom is non-conductive and the two electrodes then operate in series. The two phases A and B, which have simi- larly only a series connection, combine to produce a resultant effect equivalent to a simple single-phase power generating circuit, the voltage across the electrodes being 1'41 times the voltage across either phase winding A or B. The two-phase low tension supply may be generated for each furnace by generators, but is usually transformed down by static FIG. 45. 56 THE ELECTROMETALLURGY OF STEEL transformers to 'the required voltage from either a two or three- phase high tension system; in the latter case the transformer windings are connected in the well-known Scott method, which gives perfect balance on the three supply phases, provided the secondary side is properly loaded. Figs. 46 and 47 show the loading on the primary side of a three-phase system after and before the hearth becomes fully conductive (Hill and Fleming). 1 Qirod Two-phase Connection (1910). Instead of using two-phase current 90 out of phase, Girod proposed to transform from two phases only of a three-phase system. Normally the FIG. 46. FIG. 47. FIG. 48. two low tension phases will be 60 apart if connected together symmetrically, but by reversing one phase, the phase displace- ment will then be 120. Consequently, the low tension two-phase currents will be 120 out of phase in place of the usual 90 for two-phase systems. By this arrangement (Fig. 48), if the voltage between each top and bottom electrode is 55 volts, then the voltage between the upper electrodes will be 95 volts, and a current will flow not only from top to bottom, but between the upper electrodes also. This results in the load being nearly equally balanced in the three supply phases, when the load is equally divided between the two arc circuits. 1 Trans. Faraday Soc., Jan., 1919. SINGLE AND POLYPHASE CURRENTS TO FURNACE OPERATION 57 The furnace low tension connections, it will be seen, closely resemble both the Electro-metal and Stobie systems. Stobie Two -phase Furnace Connections. In this system the common return wire of the Electro-metals furnace is re- placed by separate leads from the two phases to two bottom electrodes, as shown in Fig. 49. A and B represent the two- phase circuits, whose outer terminals are connected to electrodes C and D ; the return wires of the phases A and B are separately connected to bottom electrodes E and F respectively, which are embedded in a lining that is conductive only when hot. This arrangement is specially designed to lengthen the path of current through the bath for the purpose of improving the metallurgical operations by better distribution of heat and circulation of the metal. The return wires are not connected outside the furnace, and there is no neutral point common to the two phases until the lining becomes conductive. Under these conditions the furnace cannot be operated without pre-heating by gas or other means to render the bottom conductive. As soon as the hearth becomes a conductor, the two bottom electrodes are electrically connected, and the conditions are then somewhat similar to those obtaining in the Electro-metals furnace. The path of current, however, from the top to the bottom electrodes must be to some extent different, but the actual direction is problematical, since the bottom electrodes are elec- trically connected through both the conductive hearth and the bath of metal, which together constitute a neutral point of the two circuits. Rennerfelt Two-phase Furnace Connections. Two-phase current is here employed to form two arcs which are entirely independent of the furnace charge or lining for any part of their circuit. The combination of phases is similar to that used in the electro-metals furnace, but the neutral conductor, instead of being connected to a conductive lining, is attached to a third, vertically suspended electrode on to the end of which the arcs FIG. 49. 58 THE ELECTBO-METALLUKGY OF STEEL strike from two other electrodes. The arrangement is shown diagrammatically by Fig. 50, the two-phase low tension current being transformed from three-phase by Scott-connected trans- formers. In certain cases, the neutral electrode C may rest on the furnace charge or actually dip into the bath, when the condi- tions would be such that two separate direct arcs are formed. Dixon Two-phase Furnace. Heroult had originally pro- posed to split a two-phase current supply into two entirely separate circuits, each having two upper electrodes arranged in series, and thus eliminate all bottom electrode connections. Such a furnace would resemble two single-phase Heroult fur- FIG. 50. FIG. 51. naces assembled in one body. A single-phase circuit with two arcs in series, supplied with power from a source at constant voltage, requires more complicated regulation than if connected to an alternator designed with a " drooping characteristic," since not only must the voltage be equally divided between each pair of arcs in series, but the current flowing through the circuit must likewise be controlled. The system of connections adopted by Dixon (Fig. 51) was designed to simplify such complicated load regulation of each electrode. The two phases A and B are connected together at their middle points by a conductor, the outer terminal ends of the two phases being connected to electrodes C, D, E, and F. It will be seen that the current leaving any electrode will have SINGLE AND POLYPHASE CURRENTS TO FURNACE OPERATION 59 more than one return path, so that any variation of the current flowing through one will be distributed between at least two others. By this arrangement the regulation of each electrode can be effected solely by the current flowing in its circuit. Several methods of compounding two-phase currents sup- plied to furnaces of the conducting hearth type were introduced at a prior date, and, for a detailed description of the numerous ways adopted, reference should be made to British Specification No. 4742, 1914. Booth- Hall Furnace Connections. The electrical features of this furnace (Fig. 122) l embody the combined principles of the Eennerfelt and Stobie furnaces. The impossibility of starting a Stobie two-phase conductive hearth furnace when cold has been mentioned, and to overcome this objection a third electrode is introduced, which serves as a neutral return for the current +U2 B A FIG. 52. Three-phase wave curves. flowing through the phase electrodes. When starting to melt, this auxiliary electrode rests upon the charge, and is later re- moved as soon as the hearth is sufficiently conductive to carry the full return current. Application of Three-phase Currents. It has been shown how two-phase currents may be compounded to produce a re- sultant flow in a neutral return wire of a three-wire system. The same case applies to three-phase currents, which, however, may be compounded in several different ways. Usually three-phase currents are primarily generated 120 out of phase, and the voltage curves may be represented by Fig. 52. Three such separate single-phase currents may have entirely independent circuits throughout, and in that case three distinct pairs of wires would be required to carry the separate Met. and Chem. Eng.," Vol. XXIII, p. 212. 60 THE ELECTRO-METALLURGY OF STEEL currents from the point of generation to a furnace, or other power-absorbing apparatus, and back again. The curves A, B, and C represent three voltage wave forms of equal magnitude and frequency, but 120 out of phase. At any given moment M on the time axis XY, there will be definite values of impressed voltage for each circuit, which are repre- sented in Fig. 53 as being + 6, + 6, and - 12 for the circuits A, B, and C in reference to the imaginary scale indicated in Fig. 52. The voltage across circuit A is falling, that across B rising, and that across C is at its maximum negative value, but be- ginning to fall towards zero prior to reversing in sign to positive. The algebraic sum of three such im- pressed voltages at any given moment is always zero. Now supposing the terminals D, F, and H are connected together, then the circuits A, B, and C may be represented in the conventional manner as shown in Fig. 54, demonstrating them to be 120 out of phase. The point whera the three terminals D, F, and H join is known as the " Star point," and the circuits are said to be " star con- nected ". If equal resistances r are put in circuit between the outer terminals E, G, and I and a common conductor Q which con- nects the three resistances to the " star point " of the circuits A, B, and C, then at a given moment a current will flow through each circuit whose magnitude and direction is exactly propor- tional to the magnitude and sign respectively of the impressed voltage. If the voltages at the terminals E, G, and I are 4- 6, + 6, and -- 12 respectively at a given moment, then currents FIG. 54. SINGLE AND POLYPHASE CURRENTS TO FURNACE OPERATION 61 will flow through the circuits ED, GF, and IH in the direction of the arrows, and with magnitudes of 6, 6, and 12, supposing the resistances r are each equal to unity ; in this case the current through HI will act as the return for the combined currents DE and FG. This is equally true at any other moment when the resistances r are equal, and the result is that no current flows through the neutral conductor Q. This means that the " effective " or measured currents, as opposed to in- stantaneous values, are also equal, and under such conditions the load is said to be " balanced ". Now supposing the resistance r in series with the circuit C is increased, then. the current flowing in that circuit will be reduced, and will be less than the sum of the currents through the circuits A and B at the particular moment previously con- sidered. The neutral conductor Q will then act as a return for part of the current flowing through A and B, and the system will then be " out of balance," and the three circuits A, B, and C will not be carrying the same " effective " or measured cur- rent. The main line wires of a star-connected power-generating system are connected to the outer terminal of each phase, and may be connected to a three-phase apparatus absorbing the electrical energy so that each power-absorbing circuit of that apparatus is in series with one line wire, the three outer ends of the circuits being themselves connected to a common star point. It will be seen that the three resistances r in Fig'. 54 are so connected to a common neutral, and are, therefore, star- connected. In all three-phase direct arc furnaces the electrodes or line conductors are similarly "star" connected through the arcs to a common star or neutral point, which is the metallic charge or bath of molten metal. In furnace work the charge or bath, which so constitutes the neutral point, is sometimes connected through either a conducting hearth or bottom electrode to a neutral return conductor, which is itself connected to the com- mon point of a star-connected generating system. As has been explained, this neutral return conductor only carries current when the load is not equally distributed or balanced between the three power-absorbing circuits. 62 THE ELECTKO-METALLTTRGY OF STEEL Star Point If, in a balanced three-phase star-connected system, the effective or measured voltage across the terminals of each power- generating phase or, in other words, between the star point and the outer terminals of each phase is equal to 100 volts, then the " line voltage," or the voltage between any pair of the outer terminals, will be ^3 times 100 volts or 173 volts. If the three power-absorbing resistances such as arcs are star-connected, as is the case in all direct arc furnaces, the voltage across each resistance or arc will be the " line voltage " -r- \/3, when the load is equally balanced. Knowing this relationship between line voltage and arc voltage, it is easy to calculate the total load on a three-phase star- connected direct arc furnace, if the current flowing through each of the three arcs is known and is the same for all, which indicates balance. Thus, if a furnace is supplied with power from transformers, the line voltage will be invariable for given loads, and the arc voltage will be this voltage divided by 1*73. The total load in K.V.A. will then be the current flowing through each arc multiplied by the arc voltage x 3. This rule only requires knowledge of the line voltage and the current flowing through each arc, and is quite independent of the trans- former grouping. Fig. 55 illustrates the nomenclature used, and shows the low tension generating phases star-connected. Delta or Mesh Grouping*. Referring again to the instan- taneous voltage values of the three wave forms A, B, and C (Fig. 52) at a moment M on the time axis XY, the three im- pressed voltages may be graphically represented by a clock dia- gram (Fig. 56), which shows their instantaneous magnitude and phase displacement. Since the amplitudes, which represent the maximum voltages of the three wave forms, are equal, the radii OD, OF, and OG are equal. At the moment chosen for setting out the clock diagram the voltage magnitudes of A and B are equal, and therefore each Electrode FIG. 55. Nomenclature diagram. SINGLE AND POLYPHASE CUEEENTS TO FUENACE OPEEATION 63 half the magnitude of C. The vertical distances of the points D, F, and G from the horizontal axis is a measure of the mag- nitude and sign of the voltages at a particular moment, and, if the radii are swinging in a clockwise direction, the magnitude of the voltage in the circuit A is falling, that in B rising, and that in C at a maximum negative value, but falling towards zero. The radii OD, OF, and OG, as set out in the diagram, B 0' 01 \0 2 F\ \0 3 G FIG. 56. graphically fulfil these conditions, and, if rotating, may be as- sumed to represent the constantly varying voltage conditions in the three circuits A, B, and C (Fig. 57). At the particular moment shown in the clock diagram it is assumed that the in- stantaneous voltage values at D, F, and G are + 6, + 6, and - 12 relatively to the terminals O', O 2 , and O 3 respectively. The impressed voltages in the circuits A and B both have a positive sign at the terminals D and F, when 0' and O 2 are con- nected together ; supposing now the terminal O' is connected to terminal F, it is obvious that it is equivalent to reversing the sign of the voltage impressed in the cir- cuit B, relative to the circuit A, or, in other words, to reversing the circuit B through 180. Therefore, if B were originally lagging 120 behind A, it will now lead A by 60, and the radii 2 F and O'D may be now shown graphi- cally in their correct relationship by the lines DO', FO 2 of Fig. 58, assuming always the radii O'D, FO 2 to be rotating in a clockwise direction. In the same way, when the terminals O 2 and O 3 are connected together, the voltage at terminal G is negative, and that at F positive ; supposing now that the terminal G is connected to FIG. 58. 64 THE ELECTROMETALLURGY OF STEEL O 2 , then it is again equivalent to reversing the sign of the volt- age impressed in the circuit C, relative to the circuit B, and instead of lagging 120 behind B it will now lead B by 60. By drawing a line GO 3 from the point O 2 , to represent the new re- lationship of the circuits C and B, a closed triangle will be formed which represents graphically Delta or Mesh grouping of three- phase circuits. The circuits A and B at the moment considered have instan- taneous voltage values of + 6 at the terminals D and F, and the circuit C a value of -- 12 at the terminal G ; then ; on con- sidering the magnitude and direction of current flow through each circuit resulting from their impressed voltages, it is evident that the circuits A and B together oppose and neutralise the effects of C. This applies at any other given moment, so that three circuits generating alternating voltages, similar in magni- tude and frequency but differing in phase by 120, may be connected to form a closed ring circuit without causing any cur- rent to flow. If three line wires LY, L 2 , and L 3 are taken from the common terminals DO 3 , FO', GO 2 , and a resistance r, be connected across any one pair L' and L 2 , then a current will flow due to the voltage impressed in the circuit A. This also applies to any other resistance circuit connecting other pairs of line wires. If the outside resistances between each pair are equal, then the effective current flowing through each phase A, B, or C will also be equal, when they are said to be balanced. The resistance or load circuits, as distinct from the generating circuits A, B, and C, when connected across the line wires are themselves " mesh " connected. The resistance or load circuits might also be connected in series with the line wires, instead of across them, and then joined together at a common star or neutral point. This case is analo- gous to a three-phase direct arc furnace, where three arcs strike on to a charge or bath which constitutes a star point, the line wires being taken from three mesh-connected generating circuits. With this combination of mesh and star connections, the effective current flowing through each line conductor will be 1'73 times the current flowing through each generating phase, provided the three phases are balanced. The line or, in this case, the generating phase voltage will be 1'73 times the voltage SINGLE AND POLYPHASE CURRENTS TO FURNACE OPERATION 65 across each arc or other similarly grouped power-absorbing circuit. Inverted Star Grouping. It has been shown how three- phase circuits may be connected together in a simple star group- ing, represented graphically by three equal radii 120 apart. Now if one of the circuits is disconnected from the neutral point, reversed and again connected to the neutral point, it will no longer be 120 out of phase with the other two, but will lead one and lag the other by 60. This, perhaps, may be more simply understood by reference to a wave form dia- gram. In Fig. 59 three equal alternating voltages 120 apart are represented by their wave forms A, B, and C ; now, supposing the wave form C is reversed, which would result by reversing the connection of the phase C relative to A and B, then the wave C' will obviously be introduced into the system in place JLJ. FIG. 59. of C. It is plain, also, that the phase displacement between A and C', B and C' is 60. The simple inverted star connection has been used for fur- nace operation, but in its more recent applications a special modification is adopted. Suppose a circuit represented by the wave form C, after dis- connection from the common star point, were split into two parts, then the impressed voltage in each portion would be in the same phase, and might be represented by two separate wave forms of different magnitude, whose sum exactly equalled the original wave form C. Either component might be again star- connected to the other phases A and B so as to be 120 apart, while the other component might be reversed before connecting again to the star point. The graphical figure would then be a combination of a simple star and inverted star connection, 5 66 THE ELECTRO-METALLURGY OF STEEL After this brief description of the commonest methods of compounding three-phase alternating current circuits, their application for furnace operation may now be considered. Stassano Three= Phase Furnace. This furnace, being de- pendent upon indirect arc heating, is provided with three nearly horizontal electrodes, which converge towards a common centre and are set 120 apart. Arcs are struck between the electrode ends, and form an equilateral triangle whose apices are the electrode points. By this formation of the arcs, the elec- trodes may be regarded as mesh connected at that end where the electrical energy is being converted into heat in the arc gaps. The current flowing through each arc, when all three circuits are balanced, is then equal to the current flowing through any electrode divided by T73, so that the load in K.V.A. ^ taken by each arc = ^=-~ x V -f- 1000 1* I O where A = current flowing through any electrode V = arc voltage, which is in this case the line voltage The total load in the furnace is thus represented by the equation K.V.A. = -4o * V x 3 - 1000 l/o provided the value of the current flowing through each elec- trode is the same. Heroult Three = Phase System. When static transformers are used for supplying power, the low tension windings are usually mesh connected, as shown in Fig. 60. They may, in certain cases, be connected to form a " star " grouping (Fig. 54), but without the use of a neutral return circuit from the furnace hearth to the " star " point. These two methods of grouping are sometimes made interchangeable by means of special switchgear for the purpose of effecting a con- siderable variation of line voltage. The high tension windings are equally well connected in either " star " or mesh grouping, which can likewise be made interchangeable. The outstanding feature of the Heroult system is the absence of any hearth connection, so that the supply of energy to the furnace charge is entirely independent of the linjng. The three SINGLE AND POLYPHASE CURRENTS TO FURNACE OPERATION 67 arcs strike on to the charge or bath, which serves as the common star point, and are thus star connected. When the electrode circuits carry the same current the total load on the furnace in V KV.A. 1-73 x A x 3 - 1000 where A = current flowing through the electrodes V = line voltage or voltage between any two electrodes. The load in K.V-A. taken by each arc is equal to the current FIG. 60. FIG. 61. flowing multiplied by the arc voltage and divided by 1000, which is Y represented by ^= x A -=- 1000 in the above equation. Three -Phase Four -Wire System. This system generally entails the use of a bottom electrode connected by a neutral return conductor to the star point of the supply circuits, as shown in Fig. 55, and has been used by Keller, Girod, and Giffre in various modified ways. Girod Three -Phase Inverted Star Connection. This method of connecting the three supply circuits is shown in Fig. 61, where it will be seen that one phase of a simple star-connected system 68 THE ELECTEO-METALLUEGY OF STEEL has been reversed. Girod adopted this method in order to increase the current flowing through the bottom electrodes. It is very doubtful whether this was a wise step, and it will be seen later how the modern tendency is rather to restrict the current flowing through the furnace hearth without unbalancing the supply phases; Stobie Three= Phase Four=Wire System. The simple case of a four- wire three-phase grouping has already been mentioned, where the neutral return was connected between a bottom electrode and the star point. According to Stobie's method the neutral return is taken from the star point and connected to an upper electrode similar in mechanical operation and dimension to the other three electrodes placed in the line conductor circuits. This fourth neutral electrode carries any out-of-balance current when the three main electrodes are unbalanced, but is other- wise inoperative. Greaves- Etchells Three= Phase Systems. In all the pre- viously mentioned methods of supplying direct arc furnaces with power from three-phase circuits, with equal line voltages, an electrode has been interposed in each line conductor so as to produce three arcs star connected to a common point, which is the metallic charge or bath of metal. It has also been shown that the power supply circuits are only balanced when such a star-connected load is equally divided between each line resistance or arc. With three arcs the load can be balanced by adjustment of the arc length, but if the resistance of one arc is fixed, then balance could only be obtained by making the resistance of each of the other two equal to it. In the Greaves- Etchells furnace only two electrodes are used for the purpose of forming arcs, and one fixed electrode is imbedded in a hearth of variable resistance. The hearth resistance may be regarded as being substituted for one arc resistance, but, instead of being adjustable, is dependent solely upon the conductivity of the material used for its construction at different temperatures. The essence of all the Greaves-Etchells patents lies in the special methods of transformer design and grouping, so that a balanced load can be obtained on the primary supply phases when the hearth resistance does not necessarily equal the arc resistances. Provision is also made so that these conditions of SINGLE AND POLYPHASE CURRENTS TO FURNACE OPERATION 69 balance may be possible over a wide variation of hearth resist- ance, provided that each arc carries the same current. A full description of the several methods of transformer grouping possible is beyond the scope of this book, as it can only be comprehensibly studied by the use of complex vector diagrams. Application of Four = Phase Currents. In Fig. 62 are repre- sented the wave curves A and B of two single-phase currents 90 out of phase. The curve A may be split up into two equal components C and C', which, owing to their equality, are super- imposed on the diagram. Supposing terminal connections are so made that C' is reversed relatively to C, then there will be two single-phase waves 180 out of phase, and if the same be applied to the wave B to form two components D and D', of which the latter is reversed, there will then be four curves 90 apart, FIG. 62. arranged in the order C, D, C', and D'. If these four circuits were star connected, they could be represented in a clock diagram by equal radii 90 apart, which would denote all the relations of one phase to another. It is quite obvious that the algebraic sum of the impressed E.M.Fs. at any moment is zero, so that, if the four individual phases were connected in ring fashion, just as in the case of three-phase circuits, no current would flow. If, however, any outside circuit were connected across any two points on the periphery of the ring, then a current would flow through that circuit. The earlier types of two-phase furnaces were designed for small capacities up to about five tons, and for this size two upper electrodes, acting in conjunction with one or more bottom electrodes, are sufficient to convey the required amount of electrical energy to the furnace. With the growing tendency 70 THE ELECTEO-METALLUEGY OF STEEL towards larger units, it has become imperative to increase the number of upper carbon conductors, with the logical introduc- tion of four-phase currents. It has already been seen how two- phase currents could be applied in the case of four upper electrodes without a neutral return, and it now remains to describe other methods by which two-phase currents, which have been compounded into four-phases by splitting each phase and reversing one-half of each, can be utilised in either mesh or star grouping (with or without neutral return conductors). All designers of bottom-electrode furnaces have appreciated the importance of varying the return current without unbalanc- ing the primary phases, and it will'be seen how this feature has FIG. 63. been introduced into some of the systems of four-phase group- ing. Dixon's Four-Phase Star Grouping. 1 The diagram of con- nections (Fig. 63) shows a simple star grouping of four-phase circuits, the " star " point being connected by a neutral return conductor to a bottom electrode embedded in a conductive hearth. The four-phase current is derived from the two-phase circuits which form the secondary windings of two groups of Scott-connected transformers. By means of switches in the three-phase primary line conductors the two-phase secondary windings of one transformer group can be made either 180 or 120 out of phase with the other. In the first case the four- phase circuits are 90 out of phase, and no current will flow British Patent Specification, No. 4742, A.D. 1914. SINGLE AND POLYPHASE CURRENTS TO FURNACE OPERATION 71 under balanced loading through the neutral return. Various other modifications have been introduced for further varying the phase relationship of one pair of circuits relatively to the other, this being done for the purpose of varying the current flowing through the furnace bottom, while still maintaining a balanced loading on the primary phases. The system of con- nections shown in Fig. 63 is used on a Gronwall-Dixon 5-ton furnace operating at Detroit, U.S.A. Dixon's Four-phase Mesh Grouping. This method is far less complicated than the above and consists of a simple ring connection shown graphically in the form of a square (Fig. 64) From each corner is branched off a conductor, which conveys current to an electrode. The load on the supply phases is JIG. 64. balanced when the current flowing through each electrode is equal. This system may be regarded as the four-phase counter- part of the simple Heroult three-phase mesh connection. Electro= metals Four= phase Five- wire Grouping. This system of connections embodies a special arrangement of the three-phase secondary windings of a transformer group. Five line conductors are required, four of which are connected to upper electrodes and one to a conductive hearth. It is also a feature of this method that the bottom electrode conductor takes only little more current than any one of the upper electrodes under balanced loading. A full explanation of the manner by which the above conditions are accomplished is only possible by resorting to complicated vectorial or mathematical solutions, so that it is only proposed here to convey the general principle, 72 THE ELECTBO-METALLTJKGY OF STEEL and to point out the fundamental difference between this and the other methods of grouping previously described. The three secondary windings are split up into parts, and are connected to four upper electrodes and a bottom electrode in the manner shown by Fig. 65. AB, CD, and EF represent the three-phase secondary circuits of a transformer group. The transformer windings AB, CD, and KF are re- spectively tapped at the points G, H, and I. The points G and H are con- nected to opposite ends of the wind- ings EF, and the point I is connected to the bottom electrode 5. A and B are connected to upper electrodes 3 and 4, and C and D to 2 and 1 respec- tively. It is clear that the voltage be- tween the bottom electrode or the point I and any one upper electrode is always the resultant of the voltages induced in one section of either winding AB and CD, and one-half of the winding FE. The wave curve representing the voltage between electrodes 1 and 5 is the resultant of the wave curves of the impressed voltages across the windings HD and El, which are different in magnitude and 120 out of phase. In Fig. 66 the curves A and B, 120 apart, represent FIG. 65. FIG. 66. the impressed voltages across these windings El and HD ; curve C is drawn by plotting points whose distances above or below the line XY are equal to the algebraic sum of the vertical dis- tances of the curves A and B from X Y at several given moments. The curve C represents the resultant of curves A and B, both as regards magnitude of impressed E.M.F. and phase. Since SINGLE AND POLYPHASE CURRENTS TO FURNACE OPERATION 73 there are four distinct electrode circuits built up of two unequal windings in series, and whose phase relationships are in every case different, there will be four resultant curves, and, therefore, a four-phase system, in which, however, the phases are not 90 apart, as is usually the case. The magnitude of each resultant voltage curve must be the same, otherwise the load cannot be properly balanced, and it is for this reason that the windings AB and CD are both unequally divided. In the diagram the points E and H are connected together, and the voltage curve of El will then be 120 in advance of HD, assuming El and HD to be moving about a centre H in a clockwise direction, this being the convention followed through- out. Now El and HC are similarly connected, but in this case HC is 60 in advance of El, which is evident since the voltage induced in HC is exactly opposite in sign to that induced in HD in their relationship to EL The relative phase displace- ments of the four resultant curves are unimportant, so long as their magnitude is equal. Since the phase displacement between the windings HD, El, and HC, El are different, then the only way in which the magnitude of the resultant can be made the same is by making the voltages across HD and HC different, and to effect this both windings AB and CD are unequally divided. By this method of grouping the secondary windings of a three-phase system can be so arranged that the voltages between four electrodes and a neutral bottom are similar, and the current through the neutral return is nearly equal to the current through each electrode when balanced. CHAPTEE IV. GENEEATION AND CONTROL OF SINGLE AND POLYPHASE CURRENTS. Single -phase Installations. The early types of electric fur- naces operated on single-phase current supply, and usually de- rived their power from alternators attached to individual units. Power could equally well be taken from single-phase supplies generated at extra high voltage, and transformed down by static transformers. In this latter case, if the normal furnace load is only a fraction of the total power generated in the system, a heavy load fluctuation will not be accompanied by any serious diminution of the line voltage, and, provided the static trans- former is suitably designed, heavy overloads may occur without endangering any part of the electrical installation. This, however, does not apply to alternators separately con- nected to furnace units, and special provision has then to be made to prevent heavy sudden overloads which might prove disastrous, not only to the alternator, but also to its prime mover. Single-phase current is seldom generated by Power Com- panies for local consumption in high power machinery, or for long distance transmission other than for traction purposes. For this reason it is generally necessary to install special generating plant for single-phase furnace operation, since poly- phase currents cannot be satisfactorily transformed to single- phase by static transformers. This adds considerably to the capital cost of the installation, and, where motor-generator sets are used, a total loss of about 15 percent, of the power consumed is incurred. The problem of supplying power to single -phase furnaces stands out as a predominant objection to their use, unless, as is rarely the case, the Power Supply Co. allows the (74) GENEEATION OF SINGLE AND POLYPHASE CURRENTS 75 furnace load to be taken from separate phases of a polyphase system. Single -phase Generators. Alternators and their prime movers, supplying power to separate furnaces, are usually rated at normal full load capacity, and must be designed to prevent, automatically, heavy overloads which might otherwise prove a constant source of breakdown. The power absorbed in an alternating current circuit is the product of volts and amperes (volt-ampere) multiplied by the power factor, and a variation of any one of these multiples will cause a change in the effective power. It has been explained how the effect of the self-induc- tion of a circuit, which depends partly on the strength of the current flowing, produces an opposing E.M.F. or voltage and at the same time causes a lowering of the power factor, and it is this very property that is utilised in the design of generators supplying current subject to sudden and heavy fluctuation. The alternator is constructed to have a " drooping character- istic," which denotes a rapid falling off of the terminal voltage so soon as the current in the circuit rises above the normal ; the power capable of being developed is thereby automatically restricted by a reduction of the voltage, although the actual current will rise, but not to the same extent as if the normal line voltage were maintained. An exciting dynamo is generally mounted on the generator's shaft and is itself excited from an independent source, or it may be self -excited. The excitation current for the field windings of the alternator is controlled by varying the strength of current flowing through the field of the small exciter, and by this means the alternator may be regulated to maintain a constant voltage over a range of different current outputs. In effect then, the alternator may be regulated to supply power at the same voltage for any desired current within its capacity, any increase above this current giving rise to a drop of voltage. Eegulation of the furnace load is effected by maintaining the correct terminal voltage, which is done by adjustment of the electrodes. It is also customary to provide the generating set, whether the prime mover be an electric motor, steam or gas engine, with a considerable flywheel effect, which, by its capacity for storing or giving out energy, greatly minimises the result of load fluctuations upon the prime mover. 76 THE ELECTEO-METALLUKGY OF STEEL Miles-Walker Converter. A special type of three-phase to single-phase converter has recently been introduced for furnace operation by Professor Miles- Walker, which takes the form of a rotating balancing transformer. The converter takes H.T. power from a three-phase supply at unity power factor, and de- livers L.T. single-phase current without unbalancing the supply phases. As in the case of single-phase alternators, the balancer is designed to have very considerable self-induction in the secondary or low tension circuit, so that at normal full load the current curve will lag 45 behind the voltage curve, giving a low power factor of '7. With this arrangement it is impossible on dead 0, O Oo C) O 1 t. i Hour* 15 Mn9 560 KW 4-SO 1 - . Delay , \ I 7 . _ . Moxirnur 1C - ~L lOnnC Uo.r* fo 4t" 1200 z 4P 1 y ^ 70 " - ::::?::::: 3 5 70 --- \ 1 21 1 iiuu ^ 6 ::: j* 2' / z so : : : bl 2 ^ O |::: --- 30 -t % 20' 1 1 ' 4 L- 600 PERCENTAGE LOAD FACTOR.(OERAT.NcO FIG. 79. sundry delays, the average power input was 426 kw. The radia- tion loss was, on the other hand, continuous and constantly in- creasing during the full period of six hours, and has been estimated for the purpose of this example as averaging 170 kw. throughout. On this assumption the units used usefully for melting and refining purposes amount to 450 per ton of steel, which agrees fairly well with the generally accepted figure. The maximum permissible load was 580 kw., which gives in this case an , 426 x 100 average load factor of ^r -- = 73'5 per cent, during the five and three-quarter hours of actual operation. POWER CONSUMPTION COST AND CONTRIBUTORY FACTORS 111 It must be borne in mind that as the load factor decreases and so lengthens the period of each heat the average radiation loss will be somewhat greater than 170 kw., and so give power con- sumption figures slightly higher than those indicated by the curve. Besides this, there are other minor factors which should really be considered, but their introduction would only slightly modify and tend to obscure the result by the many complica- tions involved. The curve B shows still more directly the in- fluence of the operating load factor on output, allowing thirty-six minutes for charging and fifteen minutes delay during each heat, which is, of course, good practice but perfectly possible for this size of furnace. Such curves can be plotted for any furnace, provided the average radiation loss is based upon an observed figure at a medium bath temperature, and the maximum permissible load fixed at a given figure for the purpose of determining the load factors corresponding to different average loads. Load Fluctuation. The effect of load fluctuation on power costs resolves itself purely into a question of load factor and output, as apart from the objectionable results of a badly fluctuating load upon the electrical equipment, there is always the difficulty of maintaining a desired average load. Load adjustment is almost universally effected in accordance with the indications of ammeters, which, owing to the rapid oscillations of their pointers, fail to indicate the average load on the furnace ; this inevitably leads the furnace operators to underload the furnace, so as to avoid the heavy apparent overloads inaccurately indicated by undamped instruments. Everything, therefore, should be done in the way of furnace manipulation to secure a steady load, which can then be more easily maintained at the highest desirable figure. The character of cold scrap used may have a considerable influence on the steadiness of the load and operating load factor, so that those physical qualities conducive to the maintenance of good loads during the melting period should on no account be sacrificed for the sake of the apparent economy to be gained by the purchase of cheap unsuitable scrap. Equal attention should be paid to the manner of charging in order to secure the best possible electrical conductivity within the charge ; observations on this point have been previously made. 112 THE ELECTRO-METALLURGY OF STEEL The introduction of considerable reactance into the load circuit is undoubtedly the most effective way of damping out fluctuations, and so improving the load factor. Unfortunately^ this can only be done at the expense of power factor, which, if any really marked benefit is to be obtained from the use of reactance coils, falls below the usually guaranteed average figure of '8 to '85 at normal full load. Power Consumption. It will be apparent, after considera- tion of the above-mentioned factors, that the units consumed in melting and refining cold charges are subject to considerable variation, which in a large measure accounts for the very diverse figures obtained with electric furnaces of either similar or different types. It has been found by careful calorimetric determinations that the heat contained in molten steel of average casting temperature is equivalent to about 370 kw.-hours per ton. After making due allowance for the normal extent of chemical refining on a basic hearth, fusion of fluxes, and other minor power-absorbing functions, it is not unlikely that the above figure rises to 400 or 450 kw.-hours per ton of finished steel melted and refined. This figure, then, represents the useful energy input required per ton of steel produced, irrespective of the type of furnace, operating load factor, and other conditions. The curve A in Fig. 79 illustrates how this theoretical figure can easily rise to 1100 units under adverse conditions of operation which lead to a poor load factor, even though the radiation loss is not beyond a reasonable figure. For these reasons, little value can be attached to any power consumption figures other than those that indicate the best performance possible under conditions that should be clearly stated. A good average power consumption resulting from con- tinuous operation may be taken as 750 kw.-hours per ton of steel produced, but it must be understood that this figure is subject to considerable reduction or increase according to the degree of manipulative skill, the process employed, and the ratio of useful energy input to radiation loss. When using the acid process the power consumption will usually be somewhat lower than for the basic, other conditions being similar. CHAPTER VII. ELECTRO-METALLURGICAL METHODS OF MELTING AND REFINING COLD CHARGES. Introduction. The electric furnace is used in steel works to perform a variety of functions. Generally speaking, the smaller units serve to perform all the metallurgical operations for the production of finished steel from a crude charge of miscellaneous steel or iron scrap. Furnaces of the largest capacity are usually employed in conjunction with other steel-making plant, and are then used to perform one or sometimes two distinct operations of a process conducted in several stages. In either case the furnaces, which may be basic or acid lined, are suitable for carrying out operations which are common to the open hearth furnace, but in a modified manner. Owing to the nature of electric heating, which enables chemical reactions to proceed in an atmosphere uncontaminated with oxidising gases and under slags of special character, operations may also be per- formed which have no counterpart in a gas or any other type of furnace. The several processes used in the manufacture of steel, made either wholly or in part in electric furnaces, must be broadly classified according to the acid or basic character of the slag employed. Each process must also be studied separately in its application to the treatment of cold scrap and liquid steel. The basic process has been most generally used for the pro- duction of finished steel from cold charges, while the acid process affords special advantages for foundry practice and for finishing semi-refined liquid steel. The basic process, as applied to the open hearth or Bessemer converter, is far more limited in its application than when practised in the electric furnace. Since electric heating can be applied to a furnace charge in such a manner as to exclude all oxidising gases, it is not only possible (113) 8 114 THE ELECTRO-METALLURGY OF STEEL to maintain a reducing atmosphere, but also a powerfully re- ducing slag, exceedingly basic or limey in character, which has power both to deoxidise and desulphurise a bath of semi-finished steel. This property of deoxidation is entirely absent in open hearth and converter slags, which contain at least 16 per cent, of FeO, and whose range of chemical action is limited to one of oxidation alone. It is true that a basic open hearth slag, to which has been added a quantity of calcium chloride according to the Saniter process, has a certain power of desulphurising, but the degree of sulphur removal does not equal the extraordinarily low figures obtainable with electric furnace basic slags. With regard to the acid process, there is a far closer identity of behaviour of the acid slags used in converters and in electric and open hearth furnaces, as in each case the slag will only remove carbon, silicon, and manganese, without any reduction of sulphur and phosphorus. In certain instances, the acid slag in an electric furnace may be partially reduced by carbon, when it then helps to accelerate the deoxidation of the bath ; this applies more especially to liquid refining. In this latter respect it differs from open hearth and converter slags, which are not subjected to the same intense local heat. The general principles governing the selection of one or other process for use in the electric furnace are different from those which might be applied in the case of the open hearth or converter. In the latter cases, apart from the consideration of available pig-iron, the quality of steel required usually defines the choice of process. Acid steel made from high class raw materials is admittedly superior in quality to basic steel of similar composition, but this does not apply to electric furnace steel and, if anything, is the reverse. It is now generally accepted that basic electric steel, if properly made, meets all the requirements of crucible steel, whether it be plain carbon, simple alloy or high speed, but this is only possible if the phos- phorus and sulphur are exceedingly low and at a figure which acid electric steel, made even from a good class of steel scrap, could not approach. The acid electric process, however, has been used largely for finishing semi-refined steel. In this respect, the furnace merely serves as a convenient internally heated receptacle, in which METHODS OF MELTING AND REFINING COLD CHARGES 115 oxidised steel may be brought up to a casting temperature and deoxidised under slightly reducing or neutral conditions. Neces- sary additions of alloys may be also conveniently made prior to casting. Of recent years considerable attention has been directed to the beneficial effects derived by holding steel in a perfectly tranquil state for a short period before teeming. This procedure enables minute slag particles in suspension, and possibly gases, to rise through the steel and escape. The con- ditions required to promote this simple physical action are perfectly fulfilled by the electric furnace, and it is probably some such secondary effect, proceeding simultaneously with the process of deoxidation, that helps to impart the high qualities to carefully refined acid steel. This, of course, equally applies to basic electric steel. Owing to the extravagant cost of electric energy as a source of heat, metallurgical operations, which may be satis- factorily and economically performed in either the open hearth furnace or converter, are at once precluded from electric furnaces. Hence the electric furnace is used only where the others fail, and is therefore usually limited to the further refining and finishing of comparatively pure raw materials. The charge may equally well be cold or liquid, but in either case will not require more than a small degree of purification. Although the actual reduction of impurities may be very slight, it is just that final degree of refining which justifies the use of the electric process. As a general rule, therefore, it may be assumed that the charge will consist of liquid steel or steel scrap, and average only a small percentage of carbon, phosphorus and sulphur. Excess of the two former elements may prolong the period of chemical action considerably beyond that of melting, in which case the economic advantage of the electric process will be seriously impaired. BASIC PROCESS. This process, as usually conducted, may be briefly divided into three distinct stages : (i) Melting down under oxidising conditions. (ii) Skimming and carburising. (iii) Kefining under powerfully reducing conditions and finishing with alloy additions, etc. 116 THE ELECTRO-METALLURGY OF STEEL General Outline. The process, as conducted in the various types of basic lined furnaces, either with or without conductive hearths or bottom electrodes, is the same in general principle. The details of operation, more especially charging, may be slightly different, but the method of working will not deviate far from the following description, which applies more strictly to direct arc furnaces without conductive hearths. For the operation of indirect arc furnaces, slight modifications in methods of charging will suggest themselves, and are here purposely omitted so as not to confuse or destroy the continuity of the following description of the process. The various steps in the mechanical operation, together with the chemical changes that occur, may be briefly summarised in consecutive order : 1. Hand, or mechanical charging of fluxes and scrap into the furnace, which has been previously heated. 2. Load put on to the furnace and melting begins. 3. Scrap melts under the electrodes, which bore downwards until pools of metal are formed on the bottom with a slag covering. 4. Melting and further slag formation proceeds until all fluxing materials in the original charge are fused. 5. A considerable portion of the scrap melted, forming a bath ; appearance of slag observed, and additions of ore or lime made if necessary. Feed of scrap given, filling furnace as far as possible. Eemoval of carbon, manganese, silicon, and phosphorus proceeds, provided the slag is sufficiently oxidising and basic. 6. Second feed given. Chemical reactions still proceed. 7. All the charge melted. Bath hot and showing very slight boil. Slag appearance correct. 8. Load off, skimming started. 9. Bath skimmed. Carbon additions made to naked bath if necessary, followed sometimes by a small addition of ferro- silicon, and then a mixture of fluor spar and lime. Bath be- comes partly deoxidised by the ferro- silicon addition. 10. Load on. Fluxes begin to fuse and correct slag forma- tion is promoted by an addition of carbon dust. Bath fairly hot. 11. Slag fused ; bath well rabbled. Slag pale in colour. Bath sample taken for analysis, if required. Slag thickened by METHODS OF MELTING AND EEFINING COLD CHAEGES 117 further lime additions. Carbon dust added in small quantities, as required, to complete the removal of metallic oxides from the slag, and to promote the formation of calcium carbide. 12. Sample taken and a small addition of ferro-silicon made, if necessary. Further deoxidation of the bath proceeds, and if the slag is nearly or quite white, desulphurisation begins. 13. Heat tests and bath samples taken from time to time until the steel lies quiet in a sample mould. Deoxidation and desulphurising action finished. 14. Bath at casting temperature ; alloy additions made if required. Bath rabbled after five minutes, and final sample taken. 15. Load off. Steel poured. The above summary must not be regarded as a strictly ac- curate history of a basic heat, it being obvious that the chemical changes are progressive and gradual, and overlap the manipu- lative operations. It is only intended to assist in piecing to- gether the detailed description of the different phases of the process, and to serve as a guide for examining the relationship between the various operations and the chemical reactions which they produce. Choice of Scrap. The selection of scrap has an important bearing on the economic operation of any electric steel furnace. In many cases it may be impossible to use a type of scrap that is suitable in every way, but careful judgment in the selection and mixing of inferior grades may lead to equally satisfactory results, not only as regards the quality, but also the quantity of steel made. It should be remembered that chemical analysis is not the only standpoint from which the value of scrap can be judged, since the shape and size have considerable influence on the electrical conditions during melting. A suitable class of scrap should be such as will give no trouble with the metallurgi- cal operations, or in the maintenance of a steady electrical load. The employment of light scrap of irregular shape or very light turnings is sure to cause considerable difficulties in maintaining a full and steady load, unless judiciously mixed with some class of heavy scrap, which may help to eliminate the disadvantages otherwise experienced. An unsteady and low load, due to the irregular resistance of such scrap, results in a poor load factor, 118 THE ELECTKO-METALLUKGY OF STEEL which in itself is a fundamental cause of poor output, thermal inefficiency, and increased cost of production. It must be re- membered that electrical energy equivalent to the radiation loss of the furnace constitutes a large percentage of the maximum power input of the furnace (more especially in the smaller sizes), and, therefore, inability to maintain the power available for useful heating at its maximum results in considerable loss of time and failure to operate the plant at its full capacity. The nature of the scrap from a chemical standpoint is also important. The phosphorus and sulphur contents should be reasonably low, so as not to unduly prolong the period of the refining action of the basic oxidising and reducing slags. Gener- ally speaking, almost any class of steel scrap is sufficiently low in these elements and will be satisfactory, provided the percent- age of carbon is also low. Carbon, either chemically combined or mechanically mixed in the form of coke, cinders, or oil, affects the suitability of ordinary steel scrap more than any of its other chemical constituents, Coke and cinders are very commonly found in turnings which have been dumped on to freshly made ground, but of all forms of carbonaceous foreign matter, the worst is oil. It is generally inadvisable to melt oily turnings as the sole constituent of a charge ; the action of heat decom- poses the oil, which leaves a fine deposit of carbon on the turn- ings, the greater part of which is absorbed and passes into the bath of steel. For this reason, the bath, when otherwise ready to skim, will contain too much carbon, which must then be boiled out to ensure complete removal of phosphorus. When a charge consists of low carbon scrap, the process of decarburising and dephosphorising should proceed as fast as the melting operation, so that when all the charge is melted and hot enough for skimming, the carbon, phosphorus, manganese, and silicon contents will be as ]ow as required. This is not always possible when the scrap is over-charged with carbon, owing to the diffi- culty of maintaining a strongly oxidising basic slag during melting. The metallic oxides are rapidly reduced and the con- ductivity of the slag is thereby lowered ; in order, then, to maintain a full load current the electrodes may be forced to dip into the slag, and this under certain conditions gives rise to heavy current fluctuations. The electrodes, again, should never METHODS OF MELTING AND REFINING COLD CHARGES 119 under any circumstance dip into an oxidising slag, as any addi- tion of ore is rapidly reduced by the action of the carbon elec- trodes rather than by the action of the carbon in the bath. These facts demonstrate the excessive waste of time involved in boiling down carbon and ensuring the removal of phosphorus, which can only be accomplished in extreme cases by frequent skimmings and constant additions of ore. The use of light, stringy turnings should also be avoided, as the loss of time and heat occasioned by the frequent charging of such light, bulky scrap usually represents considerably more money than the extra cost of a more suitable scrap. It may also be pointed out that, when charging stringy turnings into a small furnace using graphite electrodes, there is considerable risk of breaking the latter ; also, the melting loss by oxidation is excessive. The following observations will serve as a guide in the selec- tion of suitable scrap or mixtures of scrap for melting : I. Scrap should be of such size and form as to pack closely in the furnace and render charging operations capable of being conducted with minimum expenditure of time and labour, and without risk of damaging the door jambs or electrodes. II. Scrap should be sufficiently heavy (a) to enable the en- tire charge to be introduced into the furnace without more than two feeds after the initial charging ; (b) to form pools of metal under each electrode of sufficient size (i) to prevent entire rup- ture of the electric circuit, which might otherwise occur more particularly with small graphite electrodes should the elec- trodes bore down to the bottom and then lose contact with the unmelted charge, (ii) to prevent the electrodes arcing on to the bottom in the case of conductive hearth furnaces. III. It should be moderately free from rust to prevent over- oxidation of the bath ; in reasonable amounts rust is useful, as it will assist the removal of carbon and maintain a slag charged with oxide during melting -operations without the addition of iron ore. IV. The use of high carbon scrap alone is generally avoided, as, when melted, the resulting bath will probably contain carbon, and require further boiling down with ore additions ; at the same time there is danger of imperfect removal of phosphorus. Dead soft steel or wrought-iron scrap is also objectionable if 120 THE ELECTEO-METALLUBGY OF STEEL melted alone. The bath is invariably over-oxidised and cold melted, and the erosion of the dolomite banks and wear on the roof and walls are considerably increased. The best average carbon content of a charge of scrap is from '3 per cent, to '4 per cent. V. The presence of carbon in scrap as foreign matter, such as unconsumed cinder and especially oil, has already been dealt with. VI. Wet scrap is objectionable as it is liable to cause ex- plosions when fed into a bath with a slag covering ; this, besides being a source of danger to the furnacemen, impedes the rate of charging, and consequently increases heat losses and reduces output. Wet turnings fresh from the machine shop should be allowed to drain for a week or more before being melted. VII. A high manganese content is sometimes imperfectly removed particularly if the carbon is also high. It must not be supposed that scrap, which does not fulfil all the conditions enumerated above, is unsatisfactory ; as previ- ously stated, a judicious selection and mixing of grades, un- suitable by themselves, is quite capable of producing the very best results both as regards quality and output. Method of Charging Scrap. Scrap should invariably be charged in a manner that will offer the least resistance to the passage of current through the charge. Kail and bar ends and other such shapes should be charged side by side as closely as possible, and not in criss-cross fashion. A little extra time and trouble spent in careful charging will often save considerable loss of time caused by a fluctuating load. Heavy solid scrap, such as ingot heads, when mixed with very light scrap, should be charged under the electrodes on the bottom ; this will pre- vent any possibility of the electrodes boring down through the light scrap to the bottom and losing contact with the charge. The initial charge should be of such quantity that one or, at the most, two subsequent feeds will suffice to complete the charging operation. Constant opening of the charging doors for this purpose is a practice to be avoided, owing to the con- siderable loss of heat from the interior of the furnace, more especially when the bulk of the charge is melted. METHODS OF MELTING AND EEFINING COLD CHAKGES 121 Care is always taken to charge scrap in a manner that will afford maximum protection to the furnace banks ; in cases where turnings constitute a large portion of the charge, it is an easy matter to effect this by reserving the turnings for covering the banks from the slag line upwards. In the immediate vicinity of the arc the turnings will rapidly melt, and during subsequent feeds fresh quantities should be charged, as far as possible, be- tween the electrodes and the nearest portion of the banks. Un- melted scrap remaining on the banks is left undisturbed until the charge is almost ready for skimming, when it may be gently pushed into the bath with the aid of a hooked bar. Irregular shaped heavy scrap, if used in small quantities, is preferably kept out of the initial charge and added to the bath during subsequent feeds to avoid disturbance of the furnace load. Scrap should not be fed into a furnace until the initial charge has been melted to form a fairly hot bath of steel be- tween the electrodes. The practice of pushing in fresh scrap as soon as there is any available space leads to all the troubles associated with cold melting. The slag, being thus constantly chilled, does not promote the rapid elimination of carbon and the metalloids from the cold bath of metal, as evidenced by a dead- looking appearance and absence of boil. The bath of metal is also likely to chill on the bottom, and will then require a small addition of pig for its removal, which could otherwise only be effected by prolonged heating. There is also the possibility of the bath becoming unduly charged with dissolved oxides, and this adds to the difficulty of the subsequent carburising (if any), deoxidising, and desulphurising treatments. Provided the com- position of the slag is correct, a slight boil will indicate that the temperature of the bath formed from the initial charge is suitable for fresh additions of scrap to be made. If a furnace is properly charged and left undisturbed while the load is maintained, the portion of the charge lying between the electrodes will frequently, and especially in the case of turnings, frit or weld together and ultimately collapse into the bath beneath it, causing a violent boil and heavy current over- load. This usually occurs rather suddenly, and, unless the melter is on the alert, may cause the automatic trip to operate 122 THE ELECTRO-METALLURGY OF STEEL and cut off the supply of power. A skilful melter will usually anticipate such an occurrence by gently pushing that portion of the charge into the bath just before it is likely to collapse. This point marks a favourable time for feeding, which may be done a few minutes after the slight resulting boil has subsided. A furnace charged and fed in this manner will not have acquired a high roof temperature by the time the bath is ready for skimming. Some of these remarks may not apply to, or may be modified to suit various types of direct arc furnace, but the general principles are the same for all. In the case of indirect arc furnaces, such as the Stassano, care has to be taken not to break the long high voltage arcs formed between the electrode tips by throwing cold scrap directly across their path. If only one of the three arcs formed in a Stassano furnace is broken, the other two remain nearly unaffected and the load is not entirely ruptured ; but the re- duction of power and the resultant out-of-balance load should be avoided. Basic Oxidising Slag. Function. In this case the basic slag will contain a large percentage of iron oxide for the purpose of eliminating carbon, manganese, silicon, and phosphorus from the bath of metal by oxidation. The presence of a high per- centage of lime in the slag is also essential for the satisfactory removal of phosphorus, which, after oxidation to phosphoric oxide by the oxides present in the slag and dissolved in the metal, is fixed by the lime to form a stable phosphate. The duty then of the oxidising slag, which is somewhat similar in character to basic open hearth slag, is to reduce the quantity of all impurities in the scrap other than S to a required degree, and by so doing, to produce a bath of steel as low as possible in C, Mn, Si, and P, without excessive oxidation of the steel. Fluxes. The fluxes used for the formation of such a slag are in most cases iron ore and lime. Limestone is sometimes used, but, although initially cheaper, proves less economical after the carbon dioxide has been driven off at the expense of electric energy. It is always advisable to use the best raw materials, which in their purest forms require the least amount of heat energy for fusion, and owing to their degree of con- centration have the maximum power of chemical action. METHODS OF MELTING AND EEFINING COLD CHAKGES 123 Lump ore is preferable, as in this form its action is more rapid when thrown on to a bath of steel for boiling down purposes. Character and Appearance. A satisfactory slag will usually contain about 25-30 per cent, combined FeO and Fe 2 O 3 , and 35-40 per cent. CaO, the remainder being MgO, Si0 2 , MnO, etc. A slag of such composition may be recognised by its appearance when molten, as also by the colour and fracture of a sample removed from the furnace. The molten slag should never appear glassy and reflect the light rays of the arc, but should have a dull matt appearance, indicating a sufficiency of lime. A more positive examination of the slag can be made by observing the colour and fracture of a cooled sample. Such a sample may be conveniently taken by dipping the end of an iron bar into the slag several times and momentarily with- drawing it, so that successive layers may chill and accumulate to form a thick covering to the bar. On removal of the bar the slag will rapidly set and cool, exhibiting a smooth shiny grey skin, and cracking off readily. The fracture should be close and stony in appearance and have a grey-black colour. Insufficient oxide is always evidenced by a brownish shade, which is not always easy to detect in artificial light. Infusibility, or a frothy pastiness, is also a positive sign of lack of oxide in the slag, and is generally accompanied by a high percentage of carbon in the bath and a copious production of pale coloured smoke and luminous flame. A sample of such slag will be brown or yellow, and generally full of small cavities. An over- oxidised slag is very fluid, black in colour and has a distinctly crystalline fracture. Excess of silica Will give a glassy appear- ance to the slag, either when molten or chilled, although the colour may be black. Experience soon enables the different characteristics to be recognised and slag formation to be regularly controlled in each heat from the time that the slag can first be observed. Inexperienced melters may not observe the unsuitable character of a slag until the bath is almost ready for skimming, and have then to consider whether it is better to risk a high phosphorus content in the finished steel, or to adopt the safer measure of correcting the slag composition at the expense of time and labour. 124 THE ELECTEO-METALLUEGY OF STEEL Formation and Control. The fluxes, consisting of burnt- lime and iron ore, are usually charged on to the furnace bottom before the scrap, but this practice is sometimes modified so that only a portion is at first charged, the remainder being added while the charge is melting down. The proportion of lime and ore will depend upon the chemical composition of the scrap, besides its physical condition and freedom from rust, dirt, and carbonaceous matter. Scrap containing much carbon, either chemically combined or mechanically mixed in the form of oil or cinders, will require an increased addition of ore before or after the melting. The quantity of ore required also de- pends upon the extent to which the scrap is oxidised in the furnace either before or during melting operations. The quantity of ore and lime used in a charge should be so judged as to form a sufficient slag covering to the bath ; if the slag blanket is too thin the arc will tend to strike on to the metal and cause a fluctuating load, and if too thick, the electrodes will touch the slag without forming a free arc. Between two and a half and three and a half hundredweights of lime and ore combined will usually be sufficient to form a good slag covering for a three-ton furnace. The quantities for smaller and larger furnaces may be based on the ratio of the re- spective bath areas. If, however, the scrap itself contains slag- forming foreign matter, the weight of ore and lime must be accordingly reduced. The correct quantity of ore is best de- termined by trial for a given class of scrap and method of working, and it is better, when in doubt, to begin a heat with an insufficiency of ore and to make additions from time to time until the character of the slag is. correct, as judged by its appear- ance. The total weight of ore used will serve as a guide for more accurately proportioning the ore and lime constituents of subsequent charges. The proportion of ore and lime should be such as to maintain a good black basic slag from the time of its first formation ; in this way the elimination of C, P, Mn, and Si will proceed simultaneously with the melting, and be carried sufficiently far when the bath is hot enough for skimming. Such conditions are easily maintained if the class of scrap and other conditions of working are not frequently altered, but un- controllable circumstances, such as the fracture of an electrode, METHODS OF MELTING AND EEFINING COLD CHARGES 125 are always liable to vitiate results in the case of individual heats. Satisfactory removal of P, Mn, and Si can only be effected provided the slag conditions have been correct for a certain length of time, so that it is not sufficient merely to finish with a slag of proper composition before its removal from the furnace. This question should always engage the careful attention of the operator, whose skill and experience will enable him to judge the amount of ore or lime to add for correct regulation of slag composition during the process of melting, and to estimate the extent of P elimination at a time when the bath of steel is ready for skimming. Low bath and slag temperatures sometimes produce a mis- leading appearance of the slag, which may be thin and dead- looking with apparent indication of over-oxidation. An increase in the temperature will usually suffice to improve its character, and to produce a slight boil, provided there is still sufficient carbon in the bath. An excessive quantity of slag, due either to foreign matter in the scrap or to the addition of large quantities of ore for boiling out carbon, should be avoided by pouring off the excess at inter- vals. A thick blanket of slag increases the resistance to the passage of current between the electrodes and the metal bath, and may at times prove so considerable that the electrodes either touch the slag or dip into it. When this happens, the reduc- tion of the metallic oxides by the carbon electrodes lowers the conductivity of the slag, and further intensifies the difficulty of carbon and phosphorus removal. In extreme cases, ,the phe- nomenon of " pinch effect" may also develop and give rise to a strongly fluctuating load. Slag Reactions : I. Removal of Carbon. Carbon in a bath of steel is oxidised both by the action of iron or other metallic oxides dissolved in it, and by the influence of a covering slag rich in either FeO or Fe 2 3 , or these two combined as Fe 3 O 4 ; in either case the speed of reaction is dependent upon temperature. A cold bath of steel will not boil even under a properly composed slag unless the temperature is high enough to promote the re- action, and it is, therefore, necessary to melt a charge of scrap in such a manner that the increasing bath of steel is always hot 126 THE ELECTRO-METALLUEGY OF STEEL enough (except during feeding) to allow the elimination of carbon to proceed. The best method of charging and feeding cold scrap to ensure this has already been given. The iron oxide constituent of the slag is mainly present as FeO, and results from the partial reduction of the higher oxides derived from either rust in the scrap or added in the form of iron ore or mill scale. This partial reduction of the higher oxides accompanies the oxidation of carbon in the bath of steel by the available oxygen, the reaction being in this case exothermic. The oxidation of carbon by FeO is endothermic, so that heat must be supplied to the furnace both to promote the reaction and to maintain the necessary fluidity of the steel as its melting- point rises. The carbon is oxidised to carbon monoxide, which either produces a frothy condition of the slag or bursts through it as numerous isolated bubbles. The escape of carbon monoxide in this latter way produces the appearance of boiling and the bath of steel is commonly said to be " on the boil ". The elimination of carbon can be regularly carried further in the electric furnace than is either usually safe or desirable in open hearth practice. The limitation of bath temperature in the latter case is a question of heat application, and there is thus grave risk of the bath becoming pasty should the carbon be removed too far. The electric furnace, on the other hand, provides a source of heat, by which the temperature of a bath can be raised to a point only limited by the fusing temperature of the refractory lining. II. Eemoval of Phosphorus. The removal of phosphorus from a bath of steel takes place in two stages. In the presence of oxide of iron, either dissolved in the metal or present in the slag, the phosphorus is oxidised to P 2 5 , which may, in the absence of another suitable base, combine with excess oxide to form an unstable ferrous phosphate. This compound is again split up by metallic iron to form a stable iron phosphide Fe a P, which remains in solution in the metal. Lime serves as a most suitable base for fixing the P 2 5 originally formed, pro- vided it is present in the slag in a sufficient degree of concentra- tion. The combination of CaO as a base with the P 2 5 is accompanied by evolution of heat, and therefore results in the formation of a stable compound which has the chemical formula 4CaO . P- METHODS OF MELTING AND REFINING COLD CHARGES 127 The extent to which phosphorus elimination can be carried in arc furnaces is in great measure due to the localisation of heat, which by raising the slag temperature enables the reac- tions to proceed to a far greater extent than is possible in other furnaces. A slag that may appear pasty or thick will, in the vicinity of the arc zones, be perfectly fluid ; this enables the composition of a slag to be varied in order to increase its power of furthering chemical reactions. If a charge of scrap, lime, and ore has been properly propor- tioned, the carbon, as has been pointed out, should at no time be high enough in the bath, during the melting down period, to prevent elimination of phosphorus and other metalloids. The removal of phosphorus proceeds gradually from the time a bath is first formed, and should therefore reach a high degree of elimination by the time the entire charge is melted and hot enough for skimming ; this is, of course, subject to the main- tenance of correct slag conditions, which have been dealt with. III. Removal of Manganese and Silicon. Manganese and silicon, which may be present in the bath in solution or as definite compounds, are removed by the oxidising action of iron oxide dissolved in the bath or present in the slag. The oxida- tion proceeds from the initial stages of the melting down period, and is influenced by the amount of carbon present in the bath and the bath temperature. As a general rule there is no difficulty in removing the small percentages of Mn and Si present in ordinary carbon steel scrap under the influence of a basic oxidising slag. IV. Removal of Sulphur. It has been shown (Dr. A. Mueller) that there is a removal of sulphur during the refining operation under a basic oxidising slag. The reaction which may be expressed by the equation FeS + 2FeO = 3Fe + SO 2 , can, however, only proceed when the FeS + FeO are present in a sufficient degree of mutual concentration. The figures in the table below (Mueller) show the degree of sulphur removal from three charges of liquid steel during the oxidising and dephosphorising operation : 128 THE ELECTRO-METALLURGY OF STEEL C. Mn. P. S. Si Per Cent. S Removal. Liquid steel charged . 19 76 02 045 trace After ore addition 13 48 01 044 Before slagging . 09 30 trace 033 26-7 per cent. Liquid steel charged . 2 64 03 056 After ore addition 2 52 02 056 Before slagging . 12 28 trace 039 30-4 per cent. Liquid steel charged . 23 62 03 059 After ore addition 22 44 01 048 21 42 01 045 Before slagging . 21 38 005 , 045 23 -7 per cent. Conditions of Bath before Skimming. From a chemical standpoint sufficient has been said, both in regard to slag forma- tion and bath reactions, to show that the process up to the time of skimming is conducive to the maximum removal of P, Mn, and Si with reduction of the carbon to a low figure. Carbon. The carbon content of the bath should be about '06 per cent, to '08 per cent, and if the slag composition and temperature are correct, may be easily judged by the very slight boil which will still be exhibited. Further removal of carbon, accompanied by excessive over-oxidation, is likely to cause trouble in the subsequent deoxidation and desulphurising treat- ments, and will also cause irregularity of carbon absorption, if carburising with anthracite or other form of carbon is to follow the skimming operation. A spoon sample is usually taken and poured on to an iron plate to form a thin narrow strip, which should bend double without fracture after being quenched in water if the carbon is sufficiently low. A small piece of aluminium thrown into the spoon before pouring will produce a strip of sound steel free from blow holes ; powdered ferro- silicon is not satisfactory for this purpose as an excess is likely to cause brittleness, and so vitiate the test. Phosphorus. Phosphorus should not exceed about '012 per cent, in the bath for tool and special alloy steels ; this will be usually equivalent to a final percentage of :018 per cent, to '02 per cent, in the finished steel. It is difficult to explain the dis- crepancy, since the increase cannot possibly be due to a METHODS OF MELTING AND KEFINING COLD CHAEGES 129 reduction of phosphorus from the inappreciable quantity of the oxidising slag left behind after a careful skimming. For castings there is seldom any necessity for reducing the phos- phorus below "03 per cent, in the finished steel. After working through a few trial heats with a given quality of scrap and correct slag conditions, experience will show whether the phosphorus is sufficiently low when the bath is otherwise ready for skimming. If there is any doubt, a bath sample may be taken at least half-an-hour before it is expected to skim the bath, so that no delay will be occasioned pending the result. A charge composed of ordinary steel scrap will usually be low enough in phosphorus when entirely melted and ready for skimming. Should, however, the phosphorus be too high after melting down a charge, it is often best to pour off the bulk of the slag and add fresh lime and ore to effect the final elimination. Manganese. Manganese in the bath may vary from a trace up to 0' 4 per cent, according to the carbon and manganese content of the steel scrap. It is preferable that the manganese be 0'15 per cent, or under, although a higher figure is quite permissible, provided it is not the result of improper slag conditions. It is often an advantage to be obliged to add 0*3 per cent. Mn before casting in the form of either spiegel or ferro -manganese, as this generally enables a final adjustment of the carbon content of the steel to be made before pouring without recourse to white iron or haematite pig. Scrap of medium carbon and O8 per cent. to 1 per cent. Mn will usually melt down to a bath carrying 0'3 per cent, to 0'4 per cent. Mn, unless the slag is carefully watched and kept up to a proper degree of oxidation. It does not always follow that a percentage of manganese in the bath above 0'2 per cent, signifies also a high phosphorus content due to insufficient oxidation, but, generally speaking, the possibility is greater, and caution should be exercised. When using scrap that leaves over 0'15 per cent. Mn in the bath, it is advisabls to determine the manganese in a bath sample taken at a later period, especially where the manganese has to be within close limits in the final specification. Silicon. If carbon and phosphorus are properly eliminated as indicated above, the silicon will invariably be reduced to a trace. An unexpectedly high percentage of silicon in the 9 130 THE ELECTRO-METALLURGY OF STEEL finished steel is sometimes accompanied by high manganese and phosphorus contents, which indicate improper slag conditions before skimming. Temperature. Temperature is an important factor in skim- ming operations, and should always be carefully observed immediately before skimming. Lack of heat often renders the slag thin and difficult to gather, and' will also cause the steel to set on the skimming tools and slagging spout. A bath that has not a sufficient reserve of heat to resist scumming over after removal of slag will, if carburising is to follow, result in a low and variable absorption of carbon. This makes the subsequent operation of finishing more difficult and uneconomical, especially when further carbon has to be added in the form of high grade white iron or haematite pig. The chemical and physical con- ditions of the bath should be as nearly as possible the same for successive heats, this being one of the chief factors which lead to economy of production and uniformity of the chemical and physical properties of the product. Removal of Oxidising Slag. If the correct slag and tempera- ture conditions are fulfilled, the slag can be removed without any difficulty. The furnace is tilted, and the bulk of slag poured off until the arcs become broken and snappy. The main switch is then opened, and the electrodes raised to allow free movement of the skimming tools over all parts of the bath ; neglect to do this often leads to a broken electrode particularly in the case of graphite electrodes of small diameter with the probable result of unknown absorption of carbon by the bath. The remainder of the slag is then gently skimmed off, the last remnant being more easily removed after thickening with a few shovels of fine dolomite or lime. Care is taken throughout the operation to prevent an accumulation of slag on the lip of the spout, which makes further removal very difficult. Skimming appears to be a simple operation, but if properly done, requires considerable skill. Experience alone will enable a man to skim in the shortest time without loss of metal and to the necessary degree of removal, which is a matter of judgment according to the operations that follow. After skimming has been completed, the furnace is again tiltecl to th normal position and is ready for the next operation. METHODS OF MKLTING AND EEFINING COLD CHARGES 131 Carburising. Carburising is best done by adding to the skimmed bath a calculated quantity of anthracite, or other form of carbon low in ash and volatile matter. The material chosen should contain a maximum percentage of carbon, and be in a physical condition conducive to rapid absorption by the steel. A high percentage of ash will form a slag covering, which mechanically prevents absorption of the last additions made ; this applies particularly to high carbon steels, and to furnaces in which the bath area bears a small ratio to its volume. A high percentage of volatile matter (10 per cent, and over) in anthracite is equally objectionable, as it causes the small particles to cake together as soon as they touch the bath, which prevents proper contact and mixing. If the conditions of the bath are correct before skimming and this latter operation is properly carried out, the addition of carbon will be accompanied by a boil, which is sometimes con- siderable at first but gradually subsides as the last additions are made. This reaction, at the beginning, must be carefully watched and kept under control by adding only small quantities until the period of violent ebullition is passed ; the carbon may then be added more rapidly without fear of a boil-out. This ebullition is particularly useful as it serves . to keep the carbon and steel continually in movement. When the practice of adding ferro- silicon is adopted, it is preferred by many to destroy this effect by making the addition before, instead of after, the carbon is added ; this method may be followed quite satisfactorily, pro- vided only a small addition of carbon is to be made, but in the case of high carbon steels the slag formed by the ferro-silicon addition, together with the lack of mechanical mixing, make it far more difficult to effect a regular and maximum degree of carbon absorption. The violent boil after the first carbon addition does not take place immediately, but only proceeds after the carbon has reached a certain degree of concentra- tion in the bath, when it then reacts with the dissolved oxides present. The violence of the reaction falls off as the reduction of oxides becomes more complete. With increasing weights of carbon added to the bath, the proportion actually absorbed will similarly increase ; this is obvious, since the first addition made may be regarded as lost during the oxide reaction. The ratio 132 THE ELECTRO-METALLURGY OF STEEL of the actual amount of carbon absorbed to the weight of carbon added is conveniently, but perhaps incorrectly, expressed in terms of percentage efficiency, so that for small additions the efficiency will be lower than for large. The " efficiency " will vary between 40 per cent, to 55 per cent., and will be lower still for very small additions. The conditions that favour maximum absorption are : (a) good bath temperature, (6) freedom of bath from slag, - (c) steel not too over-oxidised, (d) low ash and volatile content of the carburiser employed. (e) physical state of the carburiser, such as to prevent loss by dusting and to offer a maximum surface, of contact. Anthracite of good quality is usually employed, and should be carefully sized by rejecting all particles passing through a 30 mesh sieve, together with the oversize from a quarter- inch or, better still, a T 3 ^ inch mesh sieve. Certain varieties of Welsh anthracite are very suitable, an excellent sample of which has the following analysis : Fixed carbon 91 '6 per cent. Volatile matter . . . . 5'3 ,, Sulphur and phosphorus . . 0'8 ,, Ash about 1*75 ,, The ash and volatile matter are usually rather higher, rising up to 5 per cent, and 7 per cent, respectively. Sulphur and phosphorus should be as low as possible. Anthracite should be stored in a dry place as, if exposed to weather, it will carry sufficient moisture to considerably vitiate the calculated weight. After the last addition of carbon has been made, it is advisable to wait until all action has ceased, as shown by the entire drop of flame from the furnace, before giving a final stir to free any carbon held together by slag. The bath is then ready for the next and final stage of the process, conducted under a power- fully reducing slag. Basic Reducing Slag. Function. Up to the time of car- burising the process is usually conducted under conditions that exert an oxidising action on the bath of steel. The act of carburising undoubtedly removes a portion of the iron oxide METHODS OF MELTING AND REFINING COLD CHARGES 133 dissolved, but is incapable of carrying the reduction past that point at which there is a chemical equilibrium between the carbon and oxide still remaining dissolved in the steel. It re- quires, then, a far more powerful reducing agent to deoxidise the steel to such an extent that it will produce a perfectly sound ingot. For this purpose a strongly reducing slag composed of lime, fluor spar and carbon is used, and this is frequently supple- mented by ferro-silicon added to the bath after skimming and in very limited quantities during the process of slag deoxidation. Deoxidation is, however, possible without the use of ferro-silicon, but the period of refining is more prolonged. Another important function performed by this slag is the removal of sulphur, which can be carried to a considerable degree. The slag also serves as a perfectly neutral covering to the bath, so that the latter can be held at any desired tem- perature without changing its chemical composition ; this is exceedingly useful, as it enables the steel to be held in a tranquil, inert condition pending analysis of bath samples, or in the event of shop delays. These conditions are also conducive to the elimination of slag suspended in the bath as minute particles. The slag, when of correct composition, will not contain more than 0'5 per cent, of metallic oxides, so that ferro-alloy additions may be made according to calculation without any allowance for oxidation loss. Fluxes. (a) Ferro-silicon. Although ferro-silicon cannot be strictly regarded as a flux it is convenient to consider it under this category, as, apart from the part it plays in the deoxidation of the bath, it supplies at the same time a large part of the silica content of the reducing slag. The greater part of the total ferro-silicon addition is made to the bath of steel im- mediately after carburising. A rich alloy, containing at least 45 per cent. Si, should be used for this purpose, otherwise the chilling effect of the larger quantities required for a poorer grade will prevent proper absorption by the already cooled bath. The alloy should be crushed small to expose as large a surface as possible to the steel and so hasten absorption. For use after the reducing slag is formed, the ferro-silicon should not contain more than 45 per cent, to 50 per cent. Si, otherwise its specific gravity will be so low that the alloy will 134 THE ELECTEO-METALLUEGY OF STEEL only with difficulty pass through the slag and enter the steel. These latter additions of the alloy should also be made in lump form, avoiding small pieces which might be held up in the slag. The most convenient grade is undoubtedly the 45 per cent, to 50 per cent. Si alloy, as it serves equally well for both purposes ; the dust and smaller pieces are added to the naked bath, while the lumps are reserved for later use. (b) Lime. Hard-burnt lime is most commonly used, al- though crushed limestone is occasionally employed. In the latter case the same objections arise as when it is used for the oxidising slag formation ; also, the formation of the reducing slag is delayed until all the carbon dioxide has been driven off. The lime should be used in small pieces, or broken up just before use by moistening with water; this is important, as fusion of the fluxes and formation of the slag will otherwise be delayed, more especially if the bath has been well-skimmed before carburising. It is advisable to select a lime that is not too high in sulphur or phosphorus, but apart from this almost any burnt lime is suitable. (c) Fluor Spar. Fluor spar is a most useful flux both for promoting slag formation and for adjusting the fluidity and general character of the slag once formed. Its action is entirely mechanical, and only serves to diminish the stiffness of a limey and almost mono-basic slag by its extreme fluidity when molten. Its action is powerful, and so it is only used sparingly when rectifying the character of the slag after initial formation. Fluor spar is often associated with galena (PbS) and should be examined for this latter mineral, which, if present in any quantity, will render the spar unfit for use. Iron has a greater affinity for sulphur than lead has, so that simple replacement will follow direct contact with the bath. (d) Silica. Sand is sometimes employed as a substitute for fluor spar, either wholly or in part, but is liable to produce a slag too rich in silica, which is usually in the neighbourhood of 25 per cent, when ferro-silicon is used for deoxidising the bath. (e) Carbon Dust. This is added to the slag for reducing the metallic oxides which are present, and for the subsequent formation of calcium carbide. Anthracite, electrode carbon, or petroleum coke is used for this purpose in the form of fine dust, METHODS OF MELTING AND REFINING COLD CHARGES 135 which reduces the risk of carbon passing into the steel and effects reduction in the minimum of time. Character and Appearance. The physical characteristics of a proper finishing slag are very well defined, and undergo marked and rapid changes with slight variation in its con- dition ; it is, therefore, an easy matter to adjust the com- position and degree of fluidity from time to time, so as to satisfy the conditions most favourable to deoxidation and de- sulphurising. The slag first formed by fusion of the lime and spar, and on to which a quantity of carbon dust has been thrown, will at first be brown or brown-yellow in colour and rather stiff; on rise of temperature it will become much thinner, and, if there is sufficient carbon present for reduction of oxides, the colour will change to a pale yellow, and subsequently to white. During this change, a pronounced reaction takes place in the slag, giving it a frothy appearance due to the evolution of carbon monoxide. The surface should at all times appear quite dull if not frothing, a glassy appearance indicating presence of metallic oxides in the slag, or lack of basicity. Ex- perience alone can indicate to which cause the faulty character is due. As a general rule the slag should have a creamy con- sistency, be white or greyish in colour, fall to a fine powder on cooling from redness, and, when moistened with water, exhibit the presence of free calcium carbide by the smell of evolved acetylene. The peculiar property which the slag possesses of falling to a fine powder is not necessarily an indication of the presence of calcium carbide, since the slag will often fall at almost a red heat before any decomposition of CaC 2 is possible. It is rather an indication of a high lime content and freedom from metallic oxides. Analyses of Reducing Slags : CaC 2 2-1 9-77 5-74 4-59 CaF 2 20-7 20-0 22-6 20-5 CaS -74 -51 -73 -47 CaO 53-0 46-7 50-1 50-8 SiO 2 14-17 16-21 14-26 15-95 A1. 2 O 3 3-18 4-69 3-95 3-96 MgO 2-95 1-99 3-65 2-95 Free coke -80 '66 1-24 1-32 136 THE ELECTRO-METALLURGY OF STEEL The above are analyses of slags taken from a 15-ton furnace used for refining liquid steel, but usually the SiO 2 is somewhat higher, being about 20 per cent, or more when ferro- silicon is used to effect preliminary deoxidation of the bath. With such strongly reducing slags almost the entire deoxidation can be effected by the calcium carbide in the slag without the use of ferro-silicon, provided the bath is not in a highly oxidised con- dition. Slag samples rich in calcium carbide usually have a pale grey colour and smooth skin, and fall to a greyish-white powder. Slag Formation and Control of Bath Deoxidation. The addition of ferro-silicon immediately following carburisation of the bath has already been mentioned. When this practice is followed, it is usually quite safe to add a quantity equivalent to 0'2 per cent. Si in the bath without any risk of unoxidised Si remaining in the steel. This applies to charges that have been melted and worked down normally from a fairly low carbon scrap, yielding an oxidised bath at the time of skimming. After the ferro-silicon has all worked through, which may be aided by rabbling, the fluor spar is added together with about half the lime, this being done to promote more rapid fusion and formation of a slag covering. After partial fusion of the fluxes added, a liberal quantity of carbon dust is thrown on and allowed to work through for a few minutes. A slag sample is then taken, and if found to be pale brown or yellow, a bath sample can be taken for the analysis of carbon, manganese, and other constituents, after well stirring the bath. However per- fect the slag may be, this sample should on no account be taken unless the bath is fairly hot, but well below the temperature of casting, otherwise the sample cannot be accepted as truly representative owing to the non-homogeneity of the bath. By retaining one-half of the lime from the first addition of fluxes, time is saved by the more rapid formation of a slag covering, which, when well fused, enables a spoon sample to be taken without risk of spoiling the spoon. The remainder of the lime is added as soon as the bath sample has been taken, the in- creasing temperature of the slag being sufficient to maintain the necessary degree of fluidity. The slag should be carefully and frequently observed by taking a sample on an iron bar. METHODS OF MELTING A^B BEFINING COLD CHAKGES 137 Any tendency to revert to a darker shade of yellow or brown demands an addition of carbon dust ; if too thick, a small quantity of fluor spar should be given, and if too thin, lime is required. The change in the appearance of the slag is accom- panied by a simultaneous change in the appearance of the smoke issuing from the furnace doors or roof. After the first addition of fluxes, a copious quantity of a pale grey-yellow smoke will be evolved together with much luminous flame ; as slag formation proceeds and the oxides become reduced, the smoke will diminish in volume and become whiter and less dense, finally assuming the appearance of a thick, white haze. The flame also will subside and become less luminous. Carbon dust may be used liberally without fear of carbon entering the steel, and is, moreover, essential to the formation of free calcium carbide. The total quantity of fluxes used should be sufficient to form a good covering to the bath, which will prevent carbon absorption by the steel from the carbon dust added. The process of deoxidation of the bath by slag reaction is a function of time dependent upon the degree of oxidation, slag composition, and temperature, and is more prolonged than when aided by other deoxidisers. Whether ferro-silicon is used or not for effecting a preliminary and partial deoxidation after skim- ming, it is a common practice to make small additions of this alloy, equivalent to '05 per cent, to 1 per cent. Si, at a later stage to assist deoxidation by the carbide bearing slag. These small additions may be made from time to time according to the con- dition of the steel, as judged by the degree of " wildness " shown on solidification of a spoon sample poured into a small mould. When ferro-silicon is used purely as a deoxidising agent, the Si is oxidised and unites with the unreduced metallic oxides still present to form silicates. The greater part of these silicates certainly passes from the steel into the slag, but the remaining portion will exist either in a state of fine suspension or true solution. In either case the silicates which remain in the steel at the time of pouring will segregate on solidification and be distinguishable under microscopic examination. For this reason, when making certain special classes of steel, it is pre- ferable to use ferro-silicon and other slag forming deoxidisers as 138 THE ELECTRO-METALLURGY OF STEEL sparingly as possible and to utilise rather the deoxidising powers of the calcium carbide bearing slag to its utmost extent. Slag Reactions. (a) Removal of Oxygen. The removal of dissolved iron oxide from the bath results from the action of ferro-silicon, when used, and of the calcium carbide formed in the slag. In the case of the former the reaction takes place between the Si and in the bath itself, whereas in the latter case it is purely a contact reaction. The reducing action of ferro-silicon is simply due to the fact that Si has a greater affinity for O than Fe, the reaction taking place with evolution of heat. With re- gard to the true slag reaction, the power of deoxidation is due to formation of calcium carbide, which, when in contact with a bath containing dissolved oxides, is immediately decomposed with liberation of CO and CaO, as may be expressed by the equations : (1) 4CaC 2 + 3Fe 3 O 4 = 9Fe + SCO + 4CaO, (2) CaC 2 + 3FeO = 3Fe + 2CO + CaO. The reaction is accompanied by flame resulting from the combustion of the CO liberated. Owing to the liberation of carbon monoxide the amount of flame issuing from the furnace becomes insignificant when the slag is white and the steel freed from oxide, and this is also a sure indication of the satis- factory completion of the reaction. The elimination of oxygen solely by means of the carbide reaction is necessarily slower than when aided by ferro-silicon. (b) Removal of Sulphur. Sulphur can be eliminated to a remarkable degree by the action of a finishing slag having the characteristics previously described. It has been shown that the removal of sulphur is comparatively slow until the slag is free from oxides, as indicated by its white appearance or the presence of calcium carbide ; at this point the sulphur reaction proceeds with great rapidity, and its elimination is soon com- plete. This is clearly demonstrated by the analyses of slag samples taken at different periods during the reducing stage (Dr. A. Mueller) : METHODS OF MELTING AND EEFINING COLD CHARGES 139 Colour. FeO and Fe 2 3 MuO. S. After carburising ,, final flux addition Before tapping . Greyish- brown White granular powder 1-25 per cent. 27 13 1-44 52 trace 39 9 1-22 There is no difficulty in reducing the sulphur in the steel to below "02 per cent., the sulphur being fixed in the slag as CaS, which can only exist as such in the absence of manganese and iron oxides. This accounts for the negligible reduction of S before the slag becomes white. The reactions which take place may be represented by the following equations, into each of which carbon, either free or combined, enters (Amberg) : (1) FeS + CaO + C = Fe + CaS + CO. (2) 3FeS + 2CaO + CaC 2 = 3Fe + 3CaS + 2CO. The removal of sulphur, according to the first equation, is probably what also takes place in a blast furnace, where the production of low sulphur pig-iron is favoured by a somewhat basic slag and a high furnace temperature. There seems little doubt, therefore, that the first equation represents the correct nature of the chemical reaction. At the higher temperatures of the arc zone, where CaC 2 is formed, the reaction may take the form of the second equation. In the same way, silicon dust or other reducing agent present in the slag will suffice to prevent the reversible reaction according to the equation : FeS + CaO ^ CaS + FeO. Alloy Additions. The final addition of ferro-manganese and ferro-silicon for specification purposes should not be made until after the steel has been " killed " by slag reaction alone, or with the aid of small ferro-silicon additions, and then not less than five minutes before actually casting. When ferro-silicon has been used to assist deoxidation, it is usual to base the calcu- lation of the final addition of this alloy, for specification pur- poses, upon a bath content of 01 per cent. Si. Nickel, ferro- chrome, and other alloys should be added before the manganese and silicon, and the bath rabbled to ensure thorough mixing and homogeneity. When large additions are to be made the alloys should be added in small quantities at a time, and it should be ascertained that the bottom is perfectly clear by feeling it with 140 THE ELECTEO-METALLURGY OF STEEL an iron bar before each addition is made. By taking this simple precaution there is never the slightest risk of the bath chilling and setting on the bottom of furnaces in which no hearth heating is developed. Aluminium, if used at all, should not exceed 4 oz. to the ton when making ingots, but for foundry work it is often advisable to increase this proportion consider- ably as a precautionary measure against subsequent oxidation, especially in the case of green sand casting. Temperature Control. Temperature control is an impor- tant factor in refining and finishing electric steel. After skim- ming and carburising, the bath temperature will be very low, and it is advisable, on again heating, to operate at a load that will'fuse the fluxes added and raise the bath temperature to a degree suitable for spoon sampling at the end of about 15 to 20 minutes. The ratio of this load to full load will be found by experience, and will vary according to the furnace capacity and the temperature of the steel after carburising. The temperature of the bath when sampling should be moderate, but not hot enough for casting, and should then be held with little further rise until the steel is nearly ready for casting, or for receiving large additions of ferro-alloys. In the case of furnaces of small capacity 30 cwts. or less it is by no means easy to so regulate the temperature, as a small margin of power over and above that equivalent to the constant radiation loss will suffice to raise the temperature rapidly. When small additions of ferro-silicon are made during the process of deoxidation, it is of the greatest importance that the bath temperature should not exceed, but preferably remain below, normal casting temperature, otherwise difficulty may be encountered in " killing" the steel, and high silicon contents result. Various methods of judging temperature are employed. Some merely take a sample in a spoon about 3 inches in diameter, allow it to remain a few seconds, and then upset the steel on to an iron plate ; if the steel runs freely, and leaves the spoon per- fectly clear, it is deemed hot enough for casting. Another method commonly adopted is to cast a small flat rectangular ingot about inch to | inch thick, which is quenched in water and then broken in half ; if the steel is sufficiently hot, the fracture will exhibit a needle-shaped structure radiating from METHODS OF MELTING AND EEFINING COLD CHARGES 141 the bottom and sides, and showing the usual plane of junction running at an angle of 45 from the bottom corners of the frac- tured surface. Both these methods have the disadvantage that their accuracy is entirely dependent upon the manner in which the spoon sample is taken, and are therefore too subject to pos- sible error. The best method, which, however, is less economi- cal, consists of plunging a clean |-inch diameter iron rod into the bath to touch the bottom, and holding it immersed for a period of five to seven seconds according to the character of the steel, at the same time gently moving it in an axial direction. The rod, when rapidly withdrawn, will indicate the temperature of the bath from top to bottom by the extent to which the steel has either adhered to it or cut into it, as the case may be. Generally speaking, for low medium to high carbon steels the bar should be just left clean after immersion for five seconds. The first methods of taking temperature may well serve as a guide during the refining operation, but the actual casting tem- perature is more accurately judged by the rod test. Calculation of Additions. As a starting point for calculations of carbon and metallic additions, the approximate weight of the bath of steel must be known. There is always a certain melting loss which will vary, according to the nature of the scrap used, from 3 per cent, up to 13 per cent, or even more, the average for heavy turnings being roughly 7 per cent. This melting loss can only be determined by observation of losses in previous heats when similar scrap was melted. The gross weight of steel cast, less the weight of all ferro-alloy additions made, when deducted from the weight of scrap charged, will give the loss incurred during the entire operation with sufficient accuracy. The following examples show the method of calculat- ing the additions for a high carbon and a chrome steel heat, the same system being adopted for any variety of alloy steel. Example I. It is required to make a high carbon steel to the following specification : C . . . I'O per cent. Mn . . . . '25 Si . . . : r '15 P below . , . - '02 S -02 142 THE ELECTEO-METALLUBGY OF STEEL Weight of scrap charged ..... 6000 Ib. ] Assume a 7 per cent, melting loss. Deduct weight lost in melting (calculated) . . 420 ,, Actual weight of steel in bath at skimming . 5580 ,, Estimated carbon content of bath . '06 per cent. Eequired to carburise up to , . '95 ,, ,, (i.e. *05 per cent, less than specifi- cation). /. increased per cent. C to add to bath "89 per cent. 89 x 5580 ., i.e. weight of carbon to be added to bath = If anthracite containing 90 per cent, fixed carbon is to be used for carburising the bath, and the carbon absorption is assumed to be 50 per cent., then weight of anthracite to be used _ ' 89 x 5580 x IQQ x 100 100 x 50 x 90 = 110 Ib. Supposing the bath sample taken for carbon and manganese analysis after proper formation of the finishing slag is found to contain, C . . '92 per cent. Mn . . '08 then the carbon will have to be raised a further *08 per cent. and the manganese '17 per cent, before tapping. For the addition of manganese there is the choice of either spiegel or ferro-manganese, in which the relative percentages of manganese and carbon contained are as 4 to 1 and 11 to 1 respectively. Since only '17 per cent. Mn is required to be added, it is obvious that spiegel is preferable to use, as roughly '04 per cent. carbon can at the same time be added. Increased per cent. Mn to add to bath = . 17 per cent. Per cent. Mn in spiegel . . . = 21*0 ,, Per cent. C in spiegel . . . = 5*0 ,, mu , 17 x 5580 x 100 Then weight of spiegel required = 45 Ib. METHODS OF MELTING AND REFINING COLD CHARGES 143 The bath will now contain '92 per cent. + -04 per cent. C or '96 per cent. C, and is therefore still too low for the specifica- tion. The most accurate means of increasing the carbon content is by the addition of white iron or haematite pig-iron, which can be obtained with not more than 1*5 per cent. Si, although 2'5 per cent, is a more general figure. Assume, however, that white iron is to be used containing 4*5 per cent. C, then since the finished steel is to contain 1 per cent. G the white iron only contains 3*5 per cent. C available for addition to the bath. Now the calculated carbon content of the bath after the addition of spiegel = -96 per cent. C. /. per cent. C to be added by the white iron . . . . . = 04 per cent. C. , ,.. . -04 x 5580 x 100 . \ the wt. of white iron required . = - - Ib. 3'5 x 100 = 64jb. The weights, therefore, of carburisers and ferro-manganese required are : Anthracite (added after skimming) . 110 Ib. Spiegel (added 5 minutes before tapping) 45 ,, White iron (added before the spiegel) . 64 ,, In the above calculations it will be seen that no allowance has been made for the increase in the weight of the charge due to the anthracite and ferro-manganese additions ; the increase is so small that the error is negligible. In the case of the white iron addition, the carbon content is only just over four times the specification figure, so that the calculations have to be based on the carbon available for carburising. The same remarks might equally well apply to the carbon added by spiegel, but the carbon content is, in this case, slightly higher and the actual weight of alloy less, so that the discrepancy is negligible when compared to the crude estimation of melting loss which always varies from heat to heat. It has been previously stated that no loss of alloys, added after the formation of a white finishing slag, is allowed for in the calculations. 144 THE ELECTKO-METALLUKGY OF STEEL Example II. It is required to make a chrome steel to the following specification : C . .1*2 per cent. Cr . .3-0 Mn . . -4 Assume the same initial charge and melting loss as in the previous example : The actual weight of steel in the bath at skimming = 5580 Ib. For the addition of chromium suppose there are two grades of ferro-chrome available, both containing 65 per cent. Cr, but one with 5 per cent. C and the other 9 '2 per cent. C. These are the two commonest brands of ferro-chrome made, and it will be seen to what use the varying carbon contents can be put for the final adjustment of carbon, while adding the correct amount of chromium. Suppose equal chromium equivalents of the two alloys are used, and the per cent, carbon added by them is calculated, then, if the bath sample after carburising has a carbon content which, together with the carbon added by one grade of the ferro-chrome, would be either above or below the specification figure, the correct adjustment may possibly be made by varying the proportions of the high and low carbon grades used. In this example, however, an addition of manga- nese is necessary, so that there is a still further means of adjust- ing the carbon by the addition of either spiegel, ferro-manganese, or a mixture of the two. It will be supposed that it is preferable to use only the 4 per cent, to 6 per cent. C ferro-chrome, which for this specification will have 05 per cent, minus 3 per cent. Cr and 5 per cent, minus 1*2 per cent, carbon available for addition to the bath. Before calculating the quantity of anthracite for carburising, it is necessary to know about how much carbon will be added by the ferro-chrome and spiegel. Spiegel has, in this case, been chosen in preference to ferro-manganese for the purpose of calculation, so that in the event of the carbon found in the bath sample being lower or higher than was intended, there is still a possibility of adjustment by using the high carbon ferro- chrome in place of the 4 per cent, to 6 per cent. C grade, or ferro-manganese in place of spiegel respectively. METHODS OF MELTING AND REFINING COLD CHARGES 145 If both the lower carbon alloys were used, and the carbon was found to be too high in the bath after carburising, then there would be no means of reducing it to within the specifica- tion limits. Therefore, since carbon can easily be added by white iron or haematite pig additions, it is always a safer policy to purposely under-carburise slightly with anthracite, and rely upon alloys and, if then necessary, white iron or pig for the subse- quent and final carburisation. Calculating the ferro-chrome first : Wt. of steel in bath . . . = 5580 Ib. Per cent, of Cr required in bath = 3 per cent. Then weight of Cr to add (using the 4 per 3 x 5580 x 100 cent. - 6 per cent. C grade) . = 1QQ x (65 _ ^ = 270 Ib. Since the available percentages of Cr and C in the ferro- chrome are as 62 to 3*8, it follows that an addition of 3 per 62 cent. Cr is accompanied by an addition of 3 -r ^ or '184 per cent. C. Since spiegel is being used for the addition of say "3 per cent. Mn (assuming '1 per cent in the bath), Then the carbon added will equal "07 per cent, (about). Assume per cent. C in the bath before skimming = *06 per cent, and per cent. C added in ferro-chrome . . = 18 and per cent C added in spiegel . . .= '07 Then the total carbon in the bath, if the above additions are made without any addition of anthracite, would be "31 per cent. Now per cent. C required by specification . = 1'2 percent. Total per cent. C in bath with alloy additions = '31 ,, .". per cent. C to add as anthracite . . = '89 ,, Actually it is safer to aim rather lower than the calculated figure. Then per cent. C to be added . . = '8 (say) Now weight of steel in bath at skimming = 5580 Ib. Per cent. C required to add as anthracite = '8 per cent. Then wt. of anthracite required, assuming 90 per cent. C in the anthracite and 50 per cent. C absorption 10 146 . THE ELECTRO-METALLURGY OF STEEL 8 x 5580 x 100 x 100 100 x 50 x 90 = 99 Ib. Supposing the bath sample, taken after formation of the finishing slag, was found to contain C . . '79 per cent. Mn . .12 The per cent. C in the bath is lower than was expected from calculation, and it will be necessary to use the high carbon ferro-chrome in place of the lower carbon grade either wholly or in part. The spiegel addition must therefore be calculated first. Per cent. Mn found in bath . . = '12 per cent. ,, Mn required by specification = '4 ,, Mn to be added as spiegel . = '28 ,, It is known that about 120 Ib. of ferro-chrome will be added later, so that this weight may be added to the weight of the bath. Original wt. of bath at skimming . = 5580 Ib. wt. of ferro-chrome added . = 120 ,, wt. of bath for Mn calculation = 5700 ,, The above allowance is really unnecessary in practice, and is done in this example to convey the principle which must be applied in the case of large additions of alloys. ^ T . . . '28 x 5700 x 100 ., Wt. of spiegel required = - - Ib. lUU X AL = 76 Ib. Since the Mn content of spiegel is about four times the carbon, the per cent. C added by spiegel . . . . . = -28 -f- 4 = '07 per cent. Now per cent. C found in the bath sample = '79 per cent, and per cent. C added as spiegel . . = '07 ,, .'. Total carbon without addition from ferro- chrome . . . . . . . = '86 ,, Per cent. C required by specification . . = 1*2 ,, .'. C to be added by the ferro-chrome . . = "34 per cent. METHODS OF MELTING AND REFINING COLD CHARGES 147 From this figure it is obvious that the bulk of the ferro- chrome must be added as the high carbon alloy in which the available Cr and C are (65 - 3) per cent, and (9'2 - 1/2) per cent, respectively, the available carbon being therefore practic- ally one-eighth the chromium. If this alloy only is used, and per cent. Cr added as ferro-chrome = 3 per cent. (i.e. specification figure), then per cent. C added by this alloy =3^-8 = '37 per cent. But only '34 per cent. C is actually required, so that a portion of the high carbon grade must be replaced by the lower grade. Since the per cent. Cr is the same for both, the weight of the alloys, if mixed, will still be the amount previously found, i.e. 127 Ib. Trying a proportion of 100 Ib. high carbon alloy, and 27 Ib. of the lower carbon grade, and knowing that 127 Ib. adds 3 per cent. Cr to the bath, then per cent. Cr added by the higher C grade 100 x 3 = = 2'36 per cent. Cr, and per cent. C added by the higher C grade = '29 (i.e. one-eighth the chromium). Again the per cent. Cr added by the lower C grade X O nr*r '63o per cent. Cr, and per cent C added by the lower C grade = '04 per cent. . *. Total carbon added by the mixed alloys = "33 per cent. C. The above proportion of the two alloys is near enough. The additions, then, for the charge will be : Wt. of anthracite for carburising . = 99 Ib. Wt. of 8-10 per cent. C ferro-chrome = 100 Wt. of 4-6 per cent. C ferro-chrome . = 27 ,, Wt. of Spiegel . . . . . = 76 Tapping. There are a few points, which might be mentioned in connection with pouring, that may materially assist in pre- venting any deterioration in the quality and composition of the steel during transfer from the furnace to the ladle. A basic reducing slag, if in proper condition, is creamy in consistency, 148 THE ELECTKO-METALLUEGY OF STEEL and has no power of cohesion, as in the case of a vitreous or siliceous slag. The particles are easily broken up, and, owing to their high degree of inf usibility, do not readily escape from a mass of molten steel, when once entrapped. It is, therefore, preferable to hold back the slag when pouring into a ladle, and only allow it to pass over the spout with the last portion of steel. To effect this without running any risk of the slag ad- hering to the banks and remaining in the furnace, a small addition of fluor spar may be given just before pouring, in order to mechanically increase its fluidity at a reduced temperature. In many cases a special brick or a tapping spout (Fig. 80) is used, which holds back the slag until the steel has passed over into the ladle. It is also a good practice to hold the ladle for five minutes before teeming, as this will offer an opportunity for entrapped slag and gases to rise. The spout should always be per- fectly dry, so that there may be no risk of the steel boil- ing on it, and becoming oxi- dised. Furnace Tools and Manipulation. The tools shown in Fig. 81 are used to conduct the operations of fettling, skimming, slag sampling and charging heavy pieces of scrap. The dimen- sions shown are suitable for a 7-ton furnace, and will be rather less for smaller furnaces. The skimming rakes should be used with care, as under the best conditions they require frequent renewal of blades. If the bath is cold, the rakes will become covered with steel, and, if too hot, will be badly cut away unless withdrawn in time. The hooked bar is useful for clearing scrap off the banks, and for taking slag samples. The fettling shovel is best made as shown, so that- it can be drawn back on to the furnace door sills without lifting. Fettling. Fettling is a most important operation, upon which not only the life of a lining depends, but also the ease with which the proper metallurgical conditions can be main- Fia. 80. METHODS OF MELTING AND REFINING COLD CHARGES 149 tained. Fresh dolomite crushed to pass a f inch or inch mesh, and free from slaked dust, should be used. Care must be taken that the dolomite thrown on to the slag line does not roll down the banks and build up the bottom, which is a common error with unskilled furnacemen. The slag line is always fettled as far as possible, and then the furnace is charged with scrap up to that line. Turnings or other small scrap, when used, are best charged round the foot of the fettled slag line, and will form a seating on which more dolomite can be banked \Z/ S<~-~G RAK. iB'-o'tONO SAKWUMO SOOOM D FIG. 81. up. Dolomite can then be used liberally, and, with judicious fettling, may be built up as a facing to the lower part of a badly-cut silica wall and so greatly prolong its life. The door jambs and spout also require very careful attention, the former being kept in shape with crucible ganister or magnesite powder mixed with just enough clay to make it bind ; the spout is best lined with any kind of moulding sand, which is thrown on after removal of the slag, and beaten down to the correct shape. Ganister, although largely used, is not so good, as it requires far more drying and cracks when dry, while moulding sand, with a small percentage of water, only requires 150 THE ELECTEO-METALLUEGY OF STEEL a good skin drying and will not cause the steel to boil up on passing over. ACID PEOCESS. The chemistry of steel-making by the acid process is very similar, whether it is conducted in electric or gas furnaces, or even in the converter. In each case the removal of carbon, silicon, and manganese from the raw material, whether it be pig-iron, steel scrap, or a mixture of the two, follows from simple oxidation, which, except in the converter, proceeds under the direct influence of iron oxides and to a more limited degree oxygen in the furnace atmosphere. The acid process, as in the case of the basic process, may be employed for the conversion of steel scrap into ingots or, more generally, castings, as well as for refining molten oxidised steel. The process, as applied to the working of cold charges, may be briefly divided into three stages : (i) Melting down under oxidising conditions. (ii) Boiling out carbon under an oxidising slag. (iii) Addition of alloys and finishings. General Outline. In the acid process there is no reducing period during which dissolved oxides and sulphur are removed, and it is therefore far more important than in basic working to melt under conditions that do not conduce to over-oxidation of the bath. One great advantage is the economy resulting from the use of only one slag, which makes it possible to pro- duce a rather larger output than is possible with basic furnaces of similar power capacity. If, on the other hand, it is necessary for purposes of carburising to remove the first slag and then form another, this economy will naturally disappear. The acid process is more especially suitable for the manu- facture of castings from raw materials (usually steel scrap) sufficiently free from both phosphorus and sulphur to meet the required specification. Steel made for foundry purposes will not contain as a rule more than '35 per cent. C, so that its manu- facture is possible without any carburising addition other than ferro-alloys ; this makes it possible to operate with one slag, so that the full benefit of the process is derived in this particular METHODS OF MELTING AND KEFINING COLD CHARGES 151 application. To prevent over-oxidation of the bath during melting, it is advisable that the charge should contain a sufficient quantity of carbon, so that when entirely melted there may still be at least '3 per cent. C in the bath. Mild steel scrap, which would melt to form an over-oxidised bath with a carbon content below this figure, is usually mixed with a small quantity of pig- iron or preferably carbon dust, which, being absorbed on melt- ing, helps to limit the extent of bath oxidation. Manganese and silicon are removed by the oxide of iron, which is either added to the slag in the form of ore, or results from oxidation of the scrap in the furnace. Carbon will also be reduced until the slag becomes so impoverished in metallic oxides that further reaction becomes impossible. The carbon remaining in the bath after fusion of the charge may be further reduced by additions of ore to the slag, until it is sufficiently low for the final addition of spiegel or ferro-manganese to bring the bath within the specification figures for carbon and manganese. Before describing in detail the different operations during the melting down and finishing of a charge of scrap, it may be ad- visable to indicate briefly the consecutive steps in the furnace manipulation, and the order and nature of the chemical reactions that take place. The following description will apply more particularly to furnaces having a capacity of not more than three tons, but, with slight modification of the details of manipulation, will be equally applicable to larger furnaces : 1. Hand charging of scrap into the previously heated furnace. 2. Doors closed, load on and melting begins. 3. Scrap melts under the electrodes, forming pools of metal covered by a slag formed from the siliceous fettling material, dirt, etc. Oxidation of carbon, manganese and silicon begins, if sufficient oxide present. 4. Small quantity of ore and other fluxes added to the slag, if necessary; chemical reactions proceed slowly. 5. Melting proceeds until bulk of scrap has melted to form a large bath. 6. Further addition of scrap made and melting proceeds. 152 THE ELECTRO-METALLURGY OF STEEL 7. Kepetition of (6), if necessary. 8. Entire charge melted; carbon in bath too high for finishing, and Mn and Si low. Addition of ore made. 9. Boil begins and carbon is removed, until the slag becomes impoverished in iron oxide and the boil subsides. 10. Further ore added, if carbon is still too high. 11. Carbon content of bath sufficiently low ; iron oxide in slag reduced to a figure incapable of producing further oxidation of carbon. 12. Temperature of the bath adjusted as required, and finishing additions made before casting. 13. Load off; steel poured. Choice of Scrap. Scrap suitable for the acid process must not contain, when melted, more phosphorus and sulphur than the specification of the steel demands. Carbon contamination is not such a serious matter in the case of acid scrap for the reasons already indicated, but should not exceed a degree that would unduly lengthen the boiling down operation after com- plete fusion of the charge. Foreign matter and dirt present in carelessly collected and stored scrap will usually be siliceous in character, and is therefore less harmful to an acid lining. On the other hand, the scrap should preferably be clean and free from rust, so that it may be melted to a bath containing not less than *3 per cent, to '4 per cent. C and a minimum amount of dissolved oxide of iron. With regard to the shape, size, and other physical conditions which influence the ease of the furnace manipulation and mechanical control, the remarks which have already been made in reference to scrap suitable for the basic process will equally apply. Method of Charging. The method of charging miscellaneous scrap in a manner most favourable to the maintenance of a steady load and other desirable conditions is substantially the same for both acid and basic processes, but it should be noted that pig-iron or carbon dust, when used, is generally mixed with the scrap near the bottom in the initial charge. This is done so that the carburising action may proceed as soon as the metal forms a pool, and thereby prevent an undue absorption of oxide as melting continues, METHODS OF MELTING AND EEFINING COLD CHARGES 153 Formation and Function of Acid Slags. The functions of an acid slag are twofold : (i) To reduce the carbon, manganese, and silicon in the bath to a desired percentage. (ii) To serve as an inert covering of the bath of steel while ferro-alloy additions are being made, or whenever it is desired to prevent further chemical reaction taking place before pouring. The essential constituents of an active acid slag are SiO 2 , FeO and, generally, small quantities of CaO, A1 2 O 3 , MnO, and Fe 2 3 . The Si0 2 is derived from the acid hearth and the loose sand or ganister used as a fettling material. A small quantity of lime is sometimes added after the preliminary formation of the slag, generally for the purpose of thinning it when the FeO content has fallen to a low figure. There is usually sufficient iron oxide, derived from the rust on the scrap or from oxidation while melting, to combine with the silica for the formation of a slag covering. Other bases, such as A1 2 O 3 and CaO, will be readily absorbed by such a ferrous silicate slag which, owing to its high silica content, is powerfully acid in character. If the scrap used is exceptionally clean and the oxidation loss during melting very small, it may be necessary to add iron ore after a small bath of steel has been formed. This is done to provide the iron oxides necessary for the removal of carbon, silicon, and manganese, and at the same time to open out or thin the highly siliceous, pasty slag; the free oxygen of the Fe 2 3 plays an important part in the oxidation. The removal of carbon from the bath, which is always evidenced by a distinct boil due to evolution of carbon monoxide gas, only proceeds where the temperature of the slag and steel is above the reaction temperature. Accordingly, the oxidation of carbon only occurs in the neighbourhood of the arcs, until the entire bath with its slag covering is hot enough for the reaction to take place over the whole surface. An absence of " boil" during the melting down period does not, then, necessarily indicate lack of iron oxides in the slag. This latter condition can best be judged by the colour of a broken sample. It has been previously stated that the carbon content of the charge and the amount of FeO and 154 THE ELECTKO-METALLUKGY. OF STEEL Fe 2 O 3 passing into the slag or added to it as ore, is usually so balanced that the carbon content remaining in the bath after complete fusion of the charge will be about '3 per cent., this being done to prevent over-oxidation. As the removal of carbon from the bath proceeds, the slag becomes impoverished in metallic oxides, until finally it may be no longer capable of pro- moting oxidation, and the boil ceases. This reduction of the oxides is accompanied by very marked changes in the appear- ance of the slag when cold, and enables any subsequent addition of ore, for the further removal of carbon to a desired figure, to be correctly estimated. An acid slag capable of boiling out carbon will usually con- tain over 25 per cent, of combined FeO and MnO, which is rather less than in open hearth practice. The percentage of silica in the final slag will also be much higher, so that rather more lime is sometimes needed to obtain the required fluidity. The pasti- ness of the slag at tapping is a quality that is very useful in foundry practice where ladle lip-pouring is adopted, as the slag can be so readily held back in the ladle or skimmed off. The varying percentage of carbon in a bath during the " boil " can be judged by observing the fracture of bath samples ; such tests are generally made until the carbon is thought to be nearly low enough, when a test sample is sent to the laboratory for a carbon determination. Under proper conditions of working, the slag should cease to be active and become almost incapable of further carbon oxidation, just when the percentage of carbon in the bath has fallen to the required figure. This, of course, re- quires considerable skill and judgment when making the addi- tions of iron ore during the boiling down period. If the carbon is reduced below the desired figure it can at once be raised by the addition of pig-iron, a sufficient quantity being added to increase the carbon in the bath by the desired amount, and at the same time to compensate for any subsequent loss by slag reaction ; by this means the correct bath and slag conditions may be simultaneously obtained. The elimination of carbon by boiling gradually proceeds in successive stages, following each of the several small additions of iron ore which are made from time to time to promote the METHODS OF MELTING AND BEFINING COLD CHARGES 155 reaction. These additions are made more cautiously as the carbon approaches the desired percentage. Summarising the foregoing description of slag formation and reaction, it will be seen that the necessity of carefully control- ling the metallic oxides present in the slag is even greater than in the basic process, where over-oxidation of the bath can be almost completely corrected under a reducing slag without the aid of the final ferro-alloy additions. Physical and Chemical Characteristics of Acid Slags. The colour and fluidity of electric furnace acid slags change very markedly during the period of boiling. Assuming there is *3 per cent, to *4 per cent, carbon in the bath of steel after complete fusion of the charge, there should then be almost sufficient avail- able metallic oxides in the slag to reduce the carbon to the desired percentage. The oxides present in such a quantity will colour the slag light or dark brown, but, as the slag becomes impoverished in these oxides, the colour of the slag fracture will progressively alter. In the regular operation of a furnace carrying the same quantities of steel and slag in successive heats, the colour of the slag may be used to indicate the percentage of carbon which it is still capable of boiling out without further ore addition. This being the case, it is important to examine the slag at frequent intervals and, by knowing the approximate carbon content at a particular moment by a fracture test, it is possible to estimate fairly well just how much ore to add for the further removal of carbon required. In colour the slag succes- sively passes through various shades of yellow, until it finally becomes very faintly blue-green, and is then no longer capable of producing a vigorous boil. The viscosity will become greater by the partial removal of the iron oxide base, and may then be adjusted by small additions of some other base, such as lime. The considerable viscosity of electric furnace acid slags is mainly due to the high percentage of silica, which frequently reaches 65 per cent. A typical analysis of an electric furnace acid slag just before tapping is as follows : 156 THE ELECTROMETALLURGY OF STEEL Si0 2 .... 64'0 per cent. FeO .... 12-36 ALA .... 7-27 MnO .... 10-62 CaO . . ' . .5-9 MgO .... Trace 10015 per cent. The combined FeO and MnO are considerably lower than in acid open hearth slags, which indicates that the oxidation of carbon may be effected with a less oxidised slag and, probably, with less tendency towards over-oxidation of the steel. Ferro= Alloy Additions. Alloy additions are calculated and made in the same way as in the basic process, with the excep- tion that 10 per cent, to 20 per cent, loss of Mn and Si must be allowed for, without, however, counting on any loss of the carbon added. The addition of spiegel, ferro-manganese, and ferro- silicon is made only after the slag is pale green, and low in active oxides. The bath will then exhibit only a very slight boil, and the carbon content, either determined or calculated, is no longer liable to appreciable variation. The carbon added to the bath as a constituent of the ferro-alloys will not suffer any loss, but, owing to the slight oxidation of the bath at the end of the boil, there will be a small loss of manganese and silicon, which is allowed for as above. The proper allowance to be made for loss on this account and by a slight slag reaction can easily be determined by experience. CHAPTEK VIII. LIQUID STEEL REFINING. IN the preceding chapters, the methods by which high grade steel is made from miscellaneous steel scrap have been dealt with and classified according to the acid or basic character of the slag used. The chemical reactions upon which the basic and acid processes depend are mainly due to the interaction of molten slag and metal, and therefore proceed independently of the action of fusion. The conditions under which melting takes place certainly influence the chemical character both of the bath of metal and of the slag, and in this way alone does the process of melting affect the chemical reactions which follow, when the temperature and chemical conditions of the bath and slag are satisfied. For these reasons, the process of making highly refined steel from cold charges may be regarded simply as a process of melt- ing, followed by a period during which conditions suitable for chemical reactions are maintained, although in practice the two phases proceed simultaneously. In liquid refining the melting phase is eliminated and the chemical phase, when conducted under oxidising or non-oxidising slags, may be considered as identical in its functions, irrespective of how the molten steel has been produced. The advantage of liquid refining is economic rather than technical, as cheaper methods of producing liquid steel can be employed than is possible by melting scrap with electric energy. The electric furnace, if basic lined, may be used for purposes of dephosphorising followed by carburising, desulphurising and deoxidising treatments, and when acid lined, is used for the sole object of deoxidising liquid steel already low enough in phosphorus and sulphur. Liquid Refining in Basic Furnaces. The refining operations usually include removal of phosphorus as a preliminary to (157) 158 THE ELECTEO-METALLURGY OF STEEL carburising and the subsequent treatment for the removal of sulphur and dissolved oxides. Sometimes the liquid steel may not require further elimination of phosphorus, in which case it is carburised in the transfer ladle, or in the electric furnace itself, prior to the addition of fluxes for the formation of a reducing slag. When cold blown acid bessemer steel is elec- trically refined for further removal of phosphorus, it is some- times necessary to add about '3 per cent, carbon to the steel in the transfer ladle, so as to minimise the risk of skulling by causing a slight lowering of the melting-point ; this practice is generally followed when the time taken in transfer is consider- able. Both bottom teeming and lip pouring ladles are used for transferring the liquid steel, the latter type being used only when the slag covering is viscous and easily skimmed. In either case, a special launder is used to convey the steel from the ladle to the electric furnace, so that the steel may flow clear of the door sills on to the hearth. When a dephosphorising treatment is necessary, both iron oxide, in the form of iron ore or mill scale, and lime are shovelled into the furnace whilst the steel is still being poured. These fluxes quickly fuse and form a suit- able basic oxidising slag under which phosphorus removal is rapidly promoted. When the phosphorus removal has been carried far enough the bath is skimmed and carburised, if neces- sary ; fresh fluxes are then added, and the process of deoxidation and desulphurising followed in the same manner as described in Chapter VII. The table on the opposite page gives particulars of the materials used, time occupied, and power consumption per ton of metals charged for three typical heats, in each of which about 11 tons of liquid bessemer steel were dephosphorised and finished in a basic lined electric furnace. The power was supplied up to the time of skimming at about 2000 K.V.A., and reduced to a much lower figure during the last part of each heat. The following typical analysis of the steel, before and after refining, shows the extent to which the quality is improved. LIQUID STEEL EEFINING 159 Analysis of Bessemer Steel. C '05 --10 percent. Mn -05 10 Si -005 -015 S -035 -07 P -095 Analysis of Finished Steel. '09 per cent. 025 - O35 per cent, 015 - -04 I. II. III. Weight bessemer steel in Ib. . 26,860 25,060 25,900 Scale 700 800 600 i Ore 100 900 400 ,, Lime ,, ,, 800 800 600 Current on . 8 hrs. 15 mins. 4 hrs. 30 mins. 3 hrs. 10 mins. Began skimming . 9 hrs. 5 mins. 5 hrs. 30 mins. 3 hrs. 50 mins. Finished skimming 9 hrs. 15 mins. 5 hrs. 40 mins. 4 hrs. 10 mins. Time for skimming 10 mins. 10 mins. 20 mins. Weight crushed electrode (for carburising) . 150 ,, Lime 700 800 600 Fluor spar 325 290 275 Coke dust (added to slag) . . . 200 270 150 ,, Ferro -manganese . 270 200 150 Ferro-silicon, 50 per cent. 86 90 20 Nickel 820 Other ferro-alloys . 580 Time tapped 10 hrs. 10 mins. 7 hrs. mins. 5 hrs. mins. Total time . - . 1 hr. 55 mins. 2 hrs. 30 mins. 1 hr. 50 mins. Units consumed 3100 4200 2600 Unit per ton charged . 255 360 218 The removal of sulphur and phosphorus can always be carried to the lower limit when specified, and the silicon can also be kept low if necessary. The physical properties of electrically refined mild steel show an increase of 15 per cent, in the ultimate strength as compared with basic open hearth steel, but at the same time the elongation is decreased by 11 per cent. ; this comparison is shown in a table of tests compiled by C. G. Osborne for a paper read before the American Electro- chemical Society in 1911. Liquid Refining in Acid Furnaces. The theory and practice of deoxidising liquid steel in an acid-lined electric furnace has been carefully studied both in America and Germany. Various theories have been advanced to explain the exact manner by which deoxidation proceeds, but the several suggested chemical 160 THE ELECTRO-METALLURGY OF STEEL reactions responsible for the removal of oxygen by this process are not substantiated by any conclusive proof. Thallner, in 1913, advanced an ingenious theory to explain the physical characteristics of cast steel, based upon physical suppositions rather than upon chemical composition. According to his theory the quality of steel was greatly influenced by the size of the molecules when in the liquid condition, the physical pro- perties being improved the smaller their size. This particular condition of a bath of steel was, he considered, not only a func- tion of temperature, but also dependent upon the combined effect of several chemical reactions, in which carbon plays an essential part. The reactions which promote the deoxidation of steel in an acid-lined furnace, according to this theory, are briefly as follows : (a) Keduction of silicon from the acid hearth, with the formation of silicon carbide. (6) Decomposition of the silicon carbide by metallic oxides dissolved in the steel and present in the slag, with formation of iron carbide. (c) Partial decomposition of the iron carbide by the oxides of the slag. The silicon reduced from the lining does not then remain in the steel, but indirectly causes the formation of iron carbide to which is ascribed the specially fine grain obtained. This is only possible so long as oxide of iron is present in the slag or bath in sufficient quantity to split up the silicon-carbide first formed, otherwise silicon will be reduced from the lining in increasing quantity and remain in the finished steel. Accord- ing to the process evolved on the above assumption and practised at the works of the Lindenburg Steel Company at Eemscheid- Hasten, Germany, the carbon necessary for the above reaction is introduced into the bath in the form of briquettes, consisting of carbon and iron filings or borings. The reduction of silica to silicon is mostly from the hearth lining and only to a lesser degree from the slag, the reduction in this latter case being in- fluenced by the temperature, the amount of carbon present in the bath, and the percentage of silica in the slag. From the results obtained by melting and refining cold charges in an acid-lined three ton Girod furnace at Gutehoffnung- LIQUID STEEL REFINING 161 shiitte, the following conclusions have been drawn regarding the conditions which influence the reduction of silicon during the deoxidising period : (a) Provided that carbon is present in the bath the silicon reduction will be almost entirely derived from the acid hearth, and only to a lesser degree from the slag. (b) The reduction of silica to silicon from the slag may be increased by raising its silica content. (c) The reduction of silica to silicon from the slag may be considerably influenced by any excessive rise of temperature. (d) The reduction of silica to silicon is considerably in- fluenced by the amount of carbon in the bath, subject also to temperature. The process practised at these works consisted of cold melt- ing, followed by a careful refining. The first or oxidising slag is similar to acid open hearth slag containing a high percentage of mixed iron and manganese oxides. The charge after melting down is skimmed and fresh slag added consisting of about 75 per cent, crushed silica brick and 25 per cent, lime, to which is later added sufficient manganese oxide to give about 10-15 per cent, in the final slag ; this practice of using a finishing slag low in metallic oxide is comparable to true liquid refining of steel transferred from some other steel furnace, and is quite distinct from the cruder method of melting and finishing cold scrap charges under one slag as described in Chapter VII. The bath of steel should contain rather less than the final required per- centage of carbon before addition of the deoxidising slag. The following analyses are given as typical of the deoxidis- ing or finishing slags used at Gutehoffnungshiitte : Si0 2 54-4 52-6 54-2 66-27 CaO 11-5 13-7 23-2 15-26 MgO 5-2 3-2 3-05 2-03 ALjOj 1-72 1-52 1-86 -15 FeO 3-35 3-85 4-05 2-81 MnO 23-55 23-7 11-15 11-68 S -29 -13 -4 -51 P,O 5 03 -12 The advantages of acid refining lie in the rapidity of de- oxidation, reduction of lining repair costs due to the cheaper price of acid materials used, and the prolonged life of the silica roof. The power consumption is also less than for basic refining, 11 162 THE ELECTRO-METALLURGY OF STEEL being about 100 kw. -hours per ton for a 15-ton furnace, the period of refining being about 1-j- hours. It has been generally admitted that deoxidation, promoted under an acid slag by actual silicon reduction, is far more com- plete and produces better results than when done in the more rapid and cruder manner by additions of ferro-silicon. Deoxida- tion in the acid furnace is comparable to the " killing" action or " dead melting " associated with crucible steel manufacture. Thalmer's theories, which are based upon the initial production of silicon carbide, might reasonably be applied to the crucible, but he points out that the " killing reaction cannot be so com- plete in the latter case owing to the much lower silica content of the clay material ". The silicon-carbide theory is not generally accepted, and it is difficult to reconcile it with the fact that the minimum tempera- ture of silicon carbide formation is not less than 1800 C., a temperature that can only be reached in the arc zones. The theory generally favoured depends rather upon the alternate formation and dissociation of ferro-silicon due to the combined interaction between silica, carbon, and iron oxide. It is probable also that metallic iron itself, when in contact with a highly siliceous slag and in the presence of carbon, may pro- mote the reduction and absorption of silicon at temperatures below the reduction temperature of silicon from pure silica. In either case the silicon would not be reduced in its elemental form, but as a compound of iron and silicon. The silicon thus entering a bath of steel will immediately reduce any metallic oxides in solution, and in this manner cleanse the bath of these impurities. Acid refining based upon the above theory has been largely practised for the deoxidation of basic open hearth steel. Liquid steel, carburised to within a few points of the specifica- tion figure, is transferred to the electric furnace, and fluxes con- sisting of iron ore, lime, and sand are charged ; the greater proportion of the silica, however, is derived from the fettling material, which becomes detached from the banks and hearth. The slag becomes bluish when the steel is hot, and the silicon in the bath should not be above *05 to '08 per cent. After the bath becomes deoxidised the silicon content will rapidly rise, LIQUID STEEL REFINING 163 especially if the steel should be very hot, and for this reason the charge should be tapped without delay. This method of refining enables any class of carbon or alloy steel to be made, which is perfectly deoxidised and free from segregation in the ingot Apart from the chemical effect of liquid refining, the improvement due to purely physical reasons must not be dis- regarded. It has been stated elsewhere that considerable im- portance is now attached to allowing finished liquid steel to remain in a perfectly quiescent condition for some time previous to casting, this being done for the sole purpose of allowing finely suspended slag, or other foreign matter of low specific gravity and gases, to rise and pass out of the steel. The de- oxidation of steel in electric furnaces is not accompanied by any commotion due to evolution of carbon monoxide, so that during the process of refining the bath is also in a condition physically suitable for the free separation of minute slag particles. Scope and Application of Liquid Refining. Liquid refining in electric furnaces frequently constitutes the final stage of what are commonly known as Duplex and Triplex processes. These processes, as their names imply, embrace two or three distinct operations for the manufacture of finished steel, each operation being conducted in separate furnaces. When basic lined electric furnaces are used, the liquid steel may be simply crude blown metal produced by the acid or basic bessemer pro- cess ; in either case the blown steel may possibly require further dephosphorising, which can be done either in the basic electric furnace prior to deoxidation and desulphurisation, or in some other furnace before transfer to the electric. In the former case the process is Duplex, and in the latter case Triplex. The Duplex or Triplex process is used for the production of high class steel from unrefined liquid steel, such as is made by the basic bessemer process in Europe or the acid bessemer in America. In both cases the liquid product may be electrically refined to produce a steel equal in quality to that made by the more general method of melting and refining scrap. The basic open hearth furnace has also been used for producing cheap liquid dephosphorised steel, requiring subsequent deoxidation and desulphurising only in the basic electric furnace. 164 THE ELECTRO-MET ALLUKGY OF STEEL The application of electric furnaces to liquid refining is essentially suited to large outputs and rapid operation. The electric furnace should always be operated at a high load factor, and to render this possible a regular and frequent supply of liquid steel must be provided for. The bessemer process fulfils these conditions, and it is in conjunction with this method of steel-making that the electric furnace, as used for liquid refining, has been most generally applied. Tilting basic open- hearth furnaces working a continuous process, such as the Talbot, are equally, if not more suitable than bessemer con- verters, since the phosphorus can be sufficiently reduced to meet any acid open-hearth carbon or alloy steel specification, which thus shortens the period of subsequent refining in the electric furnace. From technical standpoints a Duplex or Triplex process, which embodies a final refining treatment of semi-finished steel by the electric process, is perfectly feasible, as has been con- clusively demonstrated both in America and Germany. The possibilities of liquid refining must be studied rather from an economic standpoint, and in this direction they will be depen- dent upon the following factors : I. The production of cheap liquid steel which can be en- hanced in value by further refining. II. A frequent and regular supply of liquid steel to the electric furnace, so that the load factor and output may be raised to a maximum, and all overhead charges correspondingly reduced. III. A suitable market for the electrically refined steel, which in the case of large outputs must be able to compete favourably with the higher grades of bessemer and open-hearth steels. CHAPTEE IX. INGOT CASTING. Theory of Ingot Formation. The art of steel-making, as applied to the manufacture of ingots, has for its ultimate object the production of steel in a crude form that will submit to subsequent physical or mechanical treatments, such as forging, rolling, machining, and heat treatment, without exhibiting or developing any structural defect. Liquid steel of excellent quality may be rendered quite unsuitable for the purpose in- tended by improper methods of casting and handling, and it is, therefore, of the utmost importance to adopt a method of casting that is satisfactory for each particular class of steel and shape of ingot. For instance, the existence of a long pipe may be harmless for one variety of steel, but might entirely ruin an in- got of another variety having imperfect welding properties. The solidification of steel in a cast-iron mould has been the subject of considerable investigation and discussion of recent years. More and more attention is being paid every day to this par- ticular branch of the art of steel-making, and even now the theories advanced by those who have long and carefully studied the matter are by no means unanimous in all respects. Iron is an element that possesses a definite crystalline form, so that solidification of steel first proceeds by crystallisation of the elementary iron from the molten metal, which contains other elements in solution. It is first necessary to study the process of solidification of a body of liquid steel under different thermal conditions, as this has a very important bearing upon the crystalline structure and chemical constitution of the solidified mass. Solidification naturally takes place wherever the temperature has fallen to the freezing-point of the steel, and may consequently be marked by an isothermal zone, which must progressively travel in an inward direction from all boundary (165) 166 THE ELECTROMETALLUKGY OF STEEL surfaces exposed to the influence of cooling. The rate at which such a zone travels at any moment is, of course, a measure of the speed of actual solidification, and depends upon the tem- perature of the liquid steel and the rate at which heat is being abstracted. Should the liquid steel be at a temperature well above its freezing-point and subject to rapid cooling, then the isothermal zone of solidification will, at any moment, be sharply de- fined and localised at the plane of junction between either the liquid steel and the walls of the mould, or the liquid steel and an already solidified envelope, owing to the abrupt temperature difference between the two in either case. Solidification under these thermal conditions will proceed by the constant deposition of thin films, or, in other words, by the slow and steady inward growth of the solidified envelope when once formed. This mode of solidification is therefore favoured by (a) a high initial casting temperature, (b) a rapid abstraction of heat. Now, even supposing solidification to have been proceeding in the above manner, the mean temperature of the still liquid steel will have been steadily falling by conduction of heat out- wardly to the surrounding walls, and might, in fact, reach a temperature near to its freezing-point before the process of solidification had proceeded very far. In that event there will be little or no temperature difference between the surface of the truly solid and liquid portions, and the isothermal zone, instead of being sharply defined as before, will become obliterated and merge into the liquid steel. Solidification will not then proceed in a distinct and well-defined manner, but will take place more or less irregularly in a zone of much greater depth. Again, if steel is cast into a mould having a very low conductivity and small thermal capacity, it will not be subject to sudden chilling at its boundary surfaces, and will fall in tem- perature slowly as a whole, until such a point is reached when a rapid irregular solidification, as above described, will result from any further lowering of temperature ; this case applies whether the steel be cast hot or cold. Solidification may also proceed in this manner when steel, at a temperature only slightly above its freezing-point, is cast into a mould of good conductivity and high thermal capacity ; this case is analogous INGOT CASTING 167 to that previously described, where rapid solidification in irregular zones followed well-defined progressive solidification. Those thermal conditions, then, which favour rapid irregular solidifica- tion are (a) a low initial casting temperature, (b) slow abstrac- tion of heat. Either of the above conditions, however, if sufficiently pronounced, will promote this mode of solidification, irrespective of the other. A careful distinction must be drawn between the total time of solidification and the actual rate at any moment, otherwise those thermal conditions, which have been mentioned as con- ducive to the two distinct modes of solidification, would appear to be quite erroneous and contrary to fact. For example, sup- pose equal weights of steel at the same temperature to be poured into an iron mould of large heat capacity and high conductivity, and a sand mould of similar dimensions having a very small heat capacity and low thermal conductivity. Obviously the actual rate of solidification in the iron mould will be rapid at first and gradually slow down until the mean temperature of the still liquid steel has fallen by conduction almost to its freezing-point, after which it becomes more rapid and general, as previously explained. In the case of the sand mould, solidification will be delayed owing to the low thermal conductivity of the material, and will hardly begin to proceed until the temperature of the liquid steel as a whole has more nearly approached its freezing-point ; solidification will then proceed rapidly throughout an ill-defined inwardly progressing zone. The total time elapsed between the moments of pouring and complete solidification will, of course, be considerably less in the case of the iron mould, although the bulk of the steel will have changed from the liquid to solid state at a slower rate. It has been already stated that solidification proceeds by the crystallisation of pure iron from the liquid steel, and that the process of crystallisation is itself influenced according to the manner of solidification at any moment. Having dealt with the various thermal conditions which influence the actual rate of change of state from liquid to solid, and the zone in which it occurs, it is now possible to see how these same conditions affect the manner in which this change occurs or, in other 168 THE ELECTEO-METALLtJEGY OF STEEL words, the crystalline structure of the solid steel. Referring to Fig. 82, assume a mass of hot liquid steel C to be poured on to a heavy iron chill plate D of indefinite area, and, taking a hypothetical case, assume that there is no heat loss from the upper surface of the liquid steel. The thermal conditions assumed are such as to promote solidification in well-defined zones. The rate at which heat is abstracted from the molten steel is variable; in the first instance, there is a rapid with- drawal of heat by the chill plate D dependent upon its thickness and, therefore, heat capacity, the exchange of heat being rapid by virtue of the high thermal conductivity of the cast-iron, which diffuses the heat throughout its mass. There must be, however, a gradual falling off in the rate of heat withdrawal as the chill plate becomes hotter and finally assumes a temperature of equilibrium, which occurs when the gain of heat from the steel by conduction is equal to the loss of heat by radiation. A thin layer of steel immediately next to the surface of the plate is subject to an intense chilling effect, and solidifies very rapidly with the formation of minute crystals, giving rise to a very close-grained crystalline structure. The thin layer of solid steel thus formed slightly lowers the rate at which heat is abstracted, and then allows solidification to proceed in well- defined isothermal zones, which favours a more regular and perfect growth of crystals. Since the heat travels from the steel to the chill plate in a definite direction, it follows that the crystals must grow regularly in the opposite direction, always presenting their uppermost end to the still liquid steel; this gives rise to a " needle "-like or columnar structure, as shown in Fig. 82. Crystals which grow in this manner will have a well- defined relative orientation, and are known as " chill " crystals. An ingot exhibiting this structure is often said to be " scorched," as it only results when the steel is cast very hot. While solidification proceeds in this way, the liquid steel is losing heat, and, should its mean temperature approach its freezing- point, solidification will then become irregular. If this occurs, crystallisation will not take place by the steady growth of the solidified surface, but may proceed by the formation of individual crystal grains within an ill-defined zone and remote from this surface. The crystallisation will then be irregular and the INGOT CASTING 169 solidified steel will be built up of crystal grains having no fixed orientation relative to one another. The columnar structure will give place to a granular structure consisting of so-called " equiaxed " crystals ; the proportion of "chill " to " equiaxed " crystals will naturally depend upon the relation between the speed of solidification, and the rate at which the still liquid steel loses heat by conduction. Slow heat abstraction will retard solidification and favour a gradual fall of temperature, and the proportion of " chill " to " equiaxed " crystals will become less, as indicated by the two sketches shown. The above theory may now be applied to demonstrate the crystalline character of a steel ingot cast in a square open-top ingot mould, standing on a heavy cast-iron bottom or chill plate. It has been explained how crystallisation proceeds uniformly in a direction at right angles to a chilling surface, so that, after the initial freez- ing of the envelope, the steel will solidify in suc- cessive layers parallel to one another, provided the rate of heat abstraction is Rapid Heat Abstraction." Slow Heat Abstract everywhere uniform. If F IG . 82. casting and cooling condi- tions are such as to favour the formation of " chill " crystals, it is evident that they will grow inwards from the four vertical sides and bottom of the mould; the crystals growing from any two adjacent chilling surfaces will be at right angles, and meet obliquely in a plane which lies at an angle of 45 to each surface, if the rate of growth should be uniform from both chill faces. Four such planes will also be formed inclining upwards from the bottom edges of the mould, and their lines of intersection will form a pyramid or truncated pyramid, according to whether the " chill " crystals penetrate to the centre of the ingot or not. In the case of small ingots, the growth of the chill crystals may be so rapid that they will penetrate to the heart of the ingot before the temperature of the steel at any time remaining fluid has fallen sufficiently low to admit of the growth of equiaxed crystals; this only occurs when the steel is cast very hot. The photograph of a broken 170 THE ELECTKO-METALLUEGY OF STEEL ingot shown in Fig. 83 indicates the three distinct forms of crystallisation illustrated in Fig. 82. If the walls of a mould vary in thickness, the rate of crystallisation will depend upon the heat-absorbing capacity at any section, and, by increasing the thickness of metal, the con- ditions which favour the growth of chill crystals are intensified. Molten steel, adjacent to the corner of an ingot mould having sides of uniform thickness at any horizontal section, is likewise subjected to more intense chilling than at points intermediate between two corners, so that chill crystals are more developed near the corners of an ingot. It has been well established that the planes marking the junction of chill crystals are planes of weakness, a fact that is demonstrated by the presence of longitudinal corner cracks, which sometimes occur in defective ingots. The solidification of steel, irrespective of the manner of crystallisation, is accompanied by a shrinkage, which must not be confused with the contraction shrinkage that follows later. The volume of liquid steel filling a mould is greater than the volume occupied by the solid steel, so that, unless solidification is accompanied by a depression of the liquid level to compensate for the difference in the two volumes, cavities will of necessity be formed in the ingot, and, moreover, at that point where solidification is finally completed. The shape of the zone marking the surface of demarcation between liquid and solid during the process of solidification, together with the correspond- ing shape of the shrinkage cavity, may cause defects in ingots that are, however, capable of being mitigated, if not almost prevented. Contraction cavities and gas cavities may also be formed under certain conditions, and give rise to defects which may only become apparent during the later stages of mechanical and physical treatment. Segregation of impurities, which only occurs to a minor extent in highly refined electric steel, must not be disregarded. The relationship between solidification, or more properly the mode of crystallisation promoted, and the chemical constitution of an ingot is dealt with later under the subject of segregation. Before describing actual methods of ingot casting, which will differ according to whether the mould is filled from the FIG. 83. [To face p. 170. INGOT CASTING 171 top or bottom, the design of the moulds and the use of certain special apparatus, it is more convenient to examine the character of the commoner ingot defects and the particular casting con- ditions responsible for their formation. In this way the merits of the various methods of casting may be better judged in so far as they avoid or minimise these harmful conditions. Ingot Defects. Piping. In the case of steel cast into chill moulds, the change of state from liquid to solid, which is ac- companied by shrinkage, has been briefly considered. Without using special precautions, the uppermost layer of molten steel in an ingot mould will become solid before solidification of the interior is complete, and since the crust so formed completes the solid envelope surrounding the still liquid portion, it follows that further shrinkage cannot be followed by a corresponding self-adjustment of the envelope, with the result that a cavity or series of cavities are formed in the interior. The shape of the main cavity depends upon the manner in which solidifica- tion has proceeded, and this again is influenced by the taper of the mould walls, the direction of taper of the mould itself, and the position occupied by the last portion of steel entering the mould. Such a cavity is generally called a " pipe," a term which is more literally descriptive of its character when it occurs as an elongated, inverted cone, with its base close to the ingot top. To study the formation of pipes it will be easiest to consider the process of solidification of liquid steel when cast into an open top mould with parallel sides. Fig. 84 represents the solidification of an ingot in successive stages. While the mould is slowly filling, the upper walls are becoming heated by radiation from the stream of steel and from the rising column of steel in the mould ; when the mould is full, the bottom of the ingot will have begun to solidify before the top has even begun to chill, so that the isothermal zone, which represents the surface of solidification, will, a few moments after teeming, be somewhat as shown in the sketch A. It is apparent that the body of steel immediately below the top crust formed will, on shrinking from it, be protected from further rapid loss of heat by radiation, and, since it was the last portion teemed, will tend to be the last portion to solidify and thus serve as a reservoir from which liquid steel is constantly drawn off as shrinkage 172 THE ELECTRO-METALLUKGY OF STEEL proceeds until finally exhausted. The other sketches shown in Fig. 84 illustrate how the pipe is formed by the thickening of the ingot wall from below the crust downwards, accompanied at the same time by constant depression of the still liquid steel. When solidification is on the point of completion, the ingrowing walls of solid steel at the middle and lower end of the ingot may be almost parallel, and if these walls should meet at certain points between which liquid steel still remains, it is obvious that shrinkage of those isolated portions cannot be met by draw- ing off from the larger reservoir above. Therefore, at those points there will also be found long, narrow cavities F (see sketch FIG. 84. Solidification in parallel moulds. D), which are known as "secondary pipes," and are en- tirely disconnected from the primary or main pipe E above. Frequently the main pipe is bridged across by one or more crusts of solid steel, which have formed at different levels, allow- ing the still molten steel below to recede and keep pace with shrinkage. The shape of the pipe formed may clearly be modified by either retarding or hastening the freezing of the steel at the upper end relatively to the lower. By retarding the rate of cooling at the top end, the zone of complete solidification from wall to wall will reach a higher position in the ingot before the INGOT CASTING 173 reservoir of liquid steel has been finally exhausted ; in this way the pipe may be considerably shortened, and will have greater lateral dimensions. It will be mentioned later how this may be effected in practice. By more rapid solidification of the upper ingot walls, the reservoir of liquid steel will rapidly diminish in volume before solidification of the lower portion of the ingot is complete ; under such conditions, those portions of the ingot still remaining liquid will, on solidification, produce elongated shrink- age cavities, or " secondary pipes," in a far more marked degree. In some cases, a solid crust may not be formed, and the pipe caused by the continual shrinkage of the steel will then be ex- posed to the air and become coated with oxide. Even supposing such an oxide coated pipe were capable of welding, the steel in the immediate vicinity of the weld would be partly decarburised and generally inferior to the rest of the billet. Pipes formed below solid crusts are free from a coating of oxide, but even then piped ingots of certain steels will not weld up perfectly during the forging or rolling operations. The presence and extent of piping are influenced by different methods of casting, to be dealt with later. In certain heavy engineering work, where safety and relia- bility is of primary importance, the pipe is either removed by trepanning or by rejection of the upper portion of the ingot in which it is situated. Kejection of the top is only effective in the absence of secondary pipes. Segregation. Liquid steel may be regarded as a complex mixture of iron, carbon, silicon, manganese, phosphorus, sulphur, etc., in which the metalloids probably exist dissolved in the iron in a colloidal state. It is now generally accepted that, at the moment of incipient solidification, pure iron begins to crystallise out from the mother liquor in the form of dendrites, which may be regarded as minute, acicular or needle-shaped crystals. According to Stead these crystals shoot out branches at right angles corresponding to the axes of a cube, and the branches themselves undergo growth of crystallisation in like manner. The mother liquor, from which these so-called dendrites or crystallites grow, becomes enriched in sulphur and phosphorus, with a corresponding increase of fusibility. The bulk of this 174 THE ELECTBO-METALLUKGY OF STEEL more fusible liquid becomes entrapped in the ever-multiplying crystal branches, which eventually become closely interlocked. If, however, sulphur and phosphorus are present in sufficiently large quantities in the mother liquor, they will form fusible compounds of low specific gravity, and these minute particles, owing to their fluidity, coalesce and are then either entrapped or pushed forward to that zone of the ingot where solidification last takes place. By virtue of their low specific gravity, the sulphur and phosphorus compounds will at the same time tend to rise through the mother liquor, and it is partly for this reason that drillings taken from an ingot in the neighbour- hood of a primary pipe contain more sulphur and phosphorus than elsewhere. This local concentration of the impurities is called " segregation," and is favoured by high casting tempera- tures and rapid cooling. These latter conditions, it has been explained, promote the growth of " chill crystals " which entrap less mother liquor as crystallisation proceeds. The mother liquor, therefore, becomes more and more enriched in impurities, and should any fusible compounds separate out as segregates, they are pushed forward by the slowly advancing chill crystals. For this reason, as Brearley has pointed out, a ring of segregates is often found lying at the boundary of the chill and equiaxed crystal zones. Equiaxed crystals, on the other hand, which result from rapid irregular crystallisation, entrap the mother liquor in situ, and so prevent pronounced segregation. Car- bon and manganese also segregate, but in both cases the per- centage enrichment in the zones of segregation is very small compared with that of sulphur and phosphorus. Sulphur segregates in the form of a fusible manganese sulphide, so that sulphur and manganese enrichments are always coincident. Segregation is usually regarded as a distinct fault, but in cases where the upper third of a large ingot is rejected on account of a pipe, or where the centre of an ingot is removed by tre- panning, it then becomes a virtue to be encouraged. Ghost Lines. " Ghost" is the name given to a defect which only becomes apparent on machining. It appears as a whitish streak in the steel, always following the direction of elongation under forging or rolling treatment. The term "ghost" is applied to such flaws, because, owing to their usually extreme INGOT CASTING 175 thinness, they can be removed with ease on machining; un- fortunately, the presence of a ghost line, which may itself be easily removed, indicates the presence of others, which may be situated outside the range of mechanical removal and are a source of danger when the steel is under transverse stress. According to Stead, these white " ghost lines " are carbonless streaks of ferrite in which are usually embedded lenticular or drawn out inclusions of manganese sulphide, the presence of which also indicates the segregation of phosphorus in the same region. The magnitude of the ghost lines is more pronounced in the case of high phosphorus steels, and to this element he attaches the power of expelling carbon from the zone of segregation on slow cooling. Others consider that the primary existence of "ghost lines " is due to slag inclusions, which act as nuclei for the secondary crystallisation of ferrite, or for the deposition of hard cementite (Fe 3 C), according to whether the percentage of carbon is below or above '89 per cent., which is the percentage present in the saturated eutectoid or solid eutectic of iron and carbon. These slag particles are drawn out on forging, and cause repeated deposition of soft ferrite or hard iron-carbide on cooling from a temperature above the critical range. This theory also explains the persistence of "ghost lines" even after repeated annealing, unless the heat treatment is so prolonged as to cause an actual dispersal of the drawn out string of nuclei. Fig. 101 shows slag inclusions imbedded in soft ferrite areas in a sample of an annealed steel casting. Lapping and Folding. These terms apply to irregularities produced on the skin of an ingot whilst the level of liquid steel is rising in the mould. When an ingot is being top-poured, it sometimes happens, especially if the steel is teemed cold, that a pasty semi-solidified crust forms on the top of the molten steel and assumes a slightly convex form along the edges of contact with the mould ; this convexity increases, and the molten steel is held from the sides of the mould until the pres- sure becomes great enough for the steel to burst through the pasty crust. The molten steel, being released, immediately flows round a portion of the edge of the crust and fills the gap between its convex surface and the side of the mould. It is improbable that the steel flowing over the convex edges of the 176 THE ELECTEO-METALLUEGY OF STEEL oxidised crust ever properly welds to it ; in fact, unless the crust has been momentarily remelted by the steel flowing upon it, the surfaces of contact are likely to become sub- sequently detached on forging or rolling. Even in the event of fusion, the crust will carry into the steel a quantity of oxide, which gives rise to local unsoundness. In top-poured ingots the lap or fold, which f jllows the line of contact between the edge of the crust and the steel flowing over it, is. generally wavy and irregular in form. In the case of bottom-run ingots, the steel rises in the mould with less surface agitation and with uniform pressure, so that if a small annular crust should form, producing only a slight convexity where in contact with the mould, the hot steel on rising will flood the crust uniformly, so that any " laps" or " folds " exhibited on the skin will be close together and parallel. If, however, the steel is very cold and the crust that forms is thick, then the folds will be wavy as in the case of top-poured ingots ; the horizontal laps cannot be pro- duced in open top ingots of small section, owing to the com- motion caused by the stream striking the steel as it fills the mould. The objection to fusion of the crust by the steel flood- ing it does not apply so strongly to closed top moulds, where oxidation of the semi-solid crust can only be very slight. Some alloy steels, particularly those containing chromium, are more liable to lap, owing to the readiness with which oxidation crusts form. Folding, then, results either from teeming cold steel or from teeming too slowly, which allows the steel to cool off and form semi-solidified crusts while filling the mould. "Shell" or "Catch". If steel on entering an open top mould should either strike the side or splash up against it from the bottom plate, it will be immediately chilled and form a thin, irregular shaped strip. This will either adhere to the side of the mould or be enveloped and remelted by the liquid steel. In the former case, as filling proceeds, the liquid steel will lise and chill against this strip without effecting a proper weld. The irregular shaped strip remains on the surface of the ingot as a " catch," and will sometimes emit a hollow sound when tapped, indicating imperfect adhesion to the body of the ingot. In the latter case the ingot is said to be " shelled," and will usually be rejected as scrap. Even if the " catch " is not very pronounced, it will INGOT CASTING 177 cause local unsoundness and small cracks owing to unequal con- traction of the strip and the liquid steel which is chilled on to it. A "shell "or "catch" is sometimes caused when bottom- casting, owing to the steel becoming partly chilled in the runner bricks at the commencement of teeming. In such cases, the steel becomes pasty and at first resists the pressure of the liquid steel in the trumpet. As soon as the pressure is sufficient to overcome the resistance offered by the pasty steel, the latter is suddenly pushed forward into the mould and followed by a fountain of liquid steel, which may strike the side of the mould and leave a strip of solid steel adhering to it. Pulling. Steel, after solidification and while still at a high temperature, is incapable of resisting any great tensile stress. Cooling and contraction of an ingot must proceed simultaneously, and any tendency to resist this normal function will result in a " tear " or " pull ". An ingot that becomes fastened to the top of a mould through flooding or some other cause will shrink away from the bottom and then be forced to carry its own weight ; in such a case the tender walls may at some point be unable to carry the weight of the portion below and are so torn apart. Moulds that for any reason have been flooded should be immediately cleared from any " fash," so as to allow free contraction of the ingot from the top downwards. Some types of "dozzles " and such other rigid devices, if too firmly fixed to the mould, will sometimes hold the ingot head and cause pulling, while the same thing applies to " scabbed " moulds or any other cause which prevents free contraction. Pitting. A good ingot is often spoiled by an inferior skin, which may give rise to small " rokes " during the later stages of forging. It is not uncommon to find the skin of an ingot studded all over with small pit marks ; these small circular depressions are caused by evolution of gas at the moment of contact between the steel and the mould, resulting either from an excessively high casting temperature, or from iron oxide or moisture on the surface of the mould. Moulds reeked with tar, which has not been vapourised before teeming, will also produce the same result. Pitting is not a serious defect, except in very small ingots, and can be avoided by careful attention to casting temperature and the use of clean and properly reeked moulds. 12 178 THE ELECTRO-METALLURGY OF STEEL Clinks. A cold ingot, as it comes from the casting pit, is always under stress which is not equally distributed. In the case of certain steels, especially those belonging to the air- hardening class, the ingots should be re-heated very slowly in order to remove these unequal stresses. If the heating is con- ducted too rapidly, the ingot -may develop deep seated cracks, which is sometimes accompanied by an audible report. Small axial cavities, which may constitute a secondary pipe or be due to internal contraction flaws, may often serve as starting points for such internal cracks. Contraction Cavities. Messrs. Brearley, in a paper read before the Iron and Steel Institute in 1916, advanced an in- genious explanation for the formation of contraction cavities. Contraction will follow immediately after solidification, and is therefore irregular in different parts of an ingot ; the outer envelope cools first and can contract on to the liquid centre, until finally it becomes sufficiently rigid to resist distortion by forces exerted upon it from within. When solidification is complete in the centre of the ingot, the outer skin will resist the deformation necessary to reduce the volume in accordance with the internal contraction that follows. Since, then, the central portion of an ingot must contract, it can only do so by tearing apart along a central axis, where the steel is least able to resist tensile stress, and this results in the formation of con- traction cavities. This process of contraction may actually proceed in those portions that have just become solid before solidification of the entire ingot is complete. From this it follows that contraction cavities will also be situated at other points, particularly where the steel is least able to resist internal stresses, and they are actually found lying in the diagonal planes of weakness which mark the junction of chill crystals growing inwards from two adjacent faces of a mould. Surface Cracks. Skin cracks, which are frequently a source of considerable trouble, may be caused in several ways. Trans- verse cracks or tears, produced by " pulling," are generally very pronounced and always occur at right angles to the direction of pull ; there is no difficulty in recognising this type of crack, and the cause, when not at once obvious, is never difficult to find. Surface cracks, which at times can only be discovered by INGOT CASTING 179 careful examination, may result either from irregular contrac- tion of the ingot just after solidification, or from the internal pressure exerted by the still liquid portion of an ingot on the thin tender wall. The formation of cracks due to fluid pressure is promoted by a high casting temperature and rapid teeming, as under such conditions the ingot walls towards the bottom may not have attained sufficient thickness to withstand the pressure due to the column of liquid steel above. The thin and tender ingot walls begin to contract immediately after their solidification, so that an exceedingly thin gap will be left between the walls of the mould and the newly forming ingot. Internal pressure may then cause rupture of the ingot skin, as evidenced by irregular surface cracks, which assume a more or less vertical direction. Cracks caused by irregular contraction of the solidified steel are due in some measure to the un-uniform chilling effect of a mould. In square moulds the chilling effect is far greater in the corners than along the sides, and in many designs of moulds the sides are cast thicker towards their centre to counteract this effect. Contraction cracks will tend to develop along any natural planes of weakness, and it is not infrequent to find longitudinal corner cracks which follow the diagonal plane marking the junction of chill crystals. High casting tempera- ture and rapid teeming promote this latter mode of crystallisa- tion and, under these conditions of ingot pouring, surface cracks are almost certain to be produced as a result both of internal pressure and irregular contraction. It does not follow that ingots which show no apparent sign of cracks will forge perfectly, and the " dry ness " in this case may probably be due to the extension of small internal contraction cracks formed in the ingot while cooling, especially in the neighbourhood of any weak spots, such as contraction or shrink- age cavities. In many cases cracks that are superficial in the ingot will open out and extend inwards on forging. The defective portion can be removed from the cogged bloom by chipping or grinding, which, however, adds considerably to the cost of working down. Frequently a perfectly sound ingot may be rendered defective under the hammer or press by improper 180 THE ELECTBO-METALLURGY OF STEEL reheating, or by forging at an unsuitable temperature, so that before condemning the original ingot it is important to ascertain whether it has been submitted to proper treatment. Blowholes. The presence of blowholes, distributed irregu- larly in the body of an ingot, is usually due to the condition of the steel when teeming rather than to improper methods of casting. Blowholes are usually formed as a result of chemical reaction between dissolved carbon and iron oxide, which may be either present in the steel before casting, or may be formed during transfer from the furnace to the mould by pouring over a damp spout, or by using an imperfectly dried ladle or bottom pouring trumpet. Steels which are low in silicon are more liable to be rendered "wild" by such contact with moisture. The evolution of gas, which occurs internally, is not observed until after solidification has begun. The oxide being in a state of dilution cannot react with the carbon in the molten steel, but, on separation of the ferrite crystals at the moment of crystal- lisation, the carbon and dissolved oxides are brought into closer contact by their resultant concentration, and then react with generation of carbon monoxide. Other theories are advanced which account for the delayed evolution of gas on the assump- tion that the reaction between the carbides and oxides only takes place at less elevated temperatures. In certain cases blowholes may be caused by the evolution of dissolved gases, which follows as a result of their reduced degree of solubility at lower temperatures. Excessive over- heating of steel in the furnace will often give rise to wildness which is sometimes impossible to eliminate by an increased addition of silicon and manganese ; it is difficult to explain this behaviour, which, however, has its analogy in acid bessemer practice, where hot blows frequently result in " wild heats " unusually high in silicon. For this reason careful control of temperature during the finishing period of a heat is necessary, and this is sometimes by no means an easy matter with small furnaces having a large power capacity. Occluded Gases. For many years past it has been thought that the presence of gases, dissolved or occluded in steel, has considerable influence on its physical properties. The actual determination of the quantity of gas dissolved and its chemical INGOT CASTING 181 composition requires most elaborate methods of sampling and analysis, and even now no definite rules governing the action of occluded gas have been established. At one time the addition of aluminium was favoured, owing to its supposed power of in- creasing the solubility of gas in steel, but it is more likely that the elimination of wildness is in this case primarily due to the removal of dissolved oxides and consequent prevention of their subsequent reaction with carbon accompanied by generation of gas. There is now little doubt that the mere act of holding molten steel in a tranquil condition prior to teeming does have a beneficial effect, which is generally accounted for by the opportunity afforded of expelling dissolved gas, and of allowing suspended slag particles to rise and so pass out of the steel. The electric furnace serves as an ideal receptacle for applying such treatment, which proceeds simultaneously with the final chemical refining stages of both the acid and basic electric processes. Ladles, their Use and Manipulation. Ladles may be divided into two classes : (a) Bottom-teeming, (6) Lip-pouring. The former type is universally adopted for ingot casting, and in foundry practice when basic-lined furnaces are used. The second type is only used for foundry casting, and then only when it is an easy matter to remove the bulk of the slag from the ladle and hold back the remainder while pouring, this being only feasible in the case of sticky acid slags. Bottom-Teeming Ladles. There are several types of bottom- teeming ladles, which only differ slightly in mechanical design. The tilting ladle (Fig. 85) consists of a slightly conical steel vessel A, built of mild steel plates and mounted on trunnions B, one of which carries a worm wheel C in gear with a worm D. The worm is fastened to one of two heavy suspension rods E con- nected together by a yoke piece F, and can be driven through bevel gearing by means of a light hand wheel, which enables the ladle to be turned through a complete circle. Two small brackets G and H serve as guides for a sliding bar I, which can be raised or lowered by means of a lever, or held in a fixed position by a set screw. This slide terminates in a head-piece machined over its upper surface to carry a cross-bar J, the head is also provided with a screwed pin or cotter bolt, which 182 THE ELECTRO-METALLUEGY OF STEEL INGOT CASTING Wedge Ring .Sleeve enables the cross-bar to be firmly secured to the slide. The outer end of the cross-bar is slotted to receive the upper end of the stopper rod K, which in this case is provided with a collar and screwed pin or cotter bolt ; the stopper rod can then be firmly fastened to the cross-bar and so respond to any vertical movement of the slide. This method of attaching the cross- bar to both the slide and the stopper rod allows for lateral ad- justment of the latter when being set so as to close the nozzle correctly. The lower end of the stopper rod (Fig. 86), if made from a solid bar, is drilled or tapped to receive the stopper pin ; in the former case, the pin is slotted and attached to the stopper rod by a cotter. A hole, through which the nozzle brick passes, is cut in the bottom plate of the ladle, and concentric with it but on the under side is fixed the nozzle box, which supports the nozzle brick and fixes its position. The walls of the ladle are lined from the bottom upwards with special shaped fire- bricks, ganister or damp fire-clay being rammed to fill the narrow gap caused by rivet and bolt heads. Ladle " compo " or ganister is sometimes rammed to form the entire lining. The bottom plate is covered with a course of fire-brick, or by a bed of ganister rammed solid ; in either case a hole is left a little larger in diameter than the nozzle brick to be used. FIG. 86. After lining, the ladle should be strongly fired for twenty-four hours until all moisture has been entirely ex- pelled ; failure to do this properly may lead to the loss of a cast of steel. The stopper rod is protected by a covering of sleeve bricks which are carefully fitted together, while the stopper end (Fig. 86) is firmly attached to the stopper rod by a stopper pin. Should a slotted pin and cotter be used for this purpose, a rigid con- nection can only be secured by interposing thin washers between the stopper brick and stopper rod, and by selecting a suitable cotter from an assortment of slightly varying widths. In the case of stopper rods having a shoulder at the upper end, the necessary number of sleeves must be threaded on to it before r Cotter . Washers Stopper P'm 184 THE ELECTEO-METALLUEGY OF STEEL fastening the stopper end brick. The stopper end and sleeves are all mortised to prevent steel finding its way through the joints, which are made as close as possible with a little thin fire-clay. As each sleeve is joined to its neighbour below, the space surrounding the iron rod is filled with sand to prevent any lateral movement after building up. The sleeves are usually made fast by driving a small wedge between the rod and a loose washer ring resting on the top sleeve brick. Stopper rods when finished are sometimes given a light wash over with graphite at the lower end and then slowly but very thoroughly dried ; any moisture in the sleeve bricks may cause them to burst when suddenly heated. The correct setting of a stopper rod before pouring is an operation that calls for considerable care and skill, and is usually only entrusted to men who have had considerable experience as helpers, and who have seen the operation performed hundreds of times before attempting it themselves. The nozzle brick, after being placed in position, is firmly fixed by ramming loam or ganister into the space round it, so as to make it one solid piece with the rest of the bottom lining. The ladle is then warmed before setting the stopper rod. The stopper rod is connected to the cross-bar, and adjusted in the manner mentioned until the stopper end finds a perfect seating on the nozzle brick when lowered. Both the nozzle brick and stopper end are con- vex at their point of contact, but owing to surface irregularity and slight distortion of shape, seldom meet exactly to form a perfect joint. The stopper end is generally ground to fit the nozzle before fastening it to the stopper rod, this being done by grinding the two surfaces together with a little fine sand before fixing them into their respective positions. This practice of grind- ing in with sand does not find universal favour, as it is argued by some that removal of the hard skin of the fire-brick is liable to cause trouble after beginning to teem. After finally setting and fastening a stopper rod, the seating is tested by throwing a little fine, dry, white sand round the junction of the nozzle and stopper brick ; if, on lightly tapping, no sand leaks through, the stopper rod is considered sufficiently well set. Sometimes ladlemen purposely set their stopper rods a fraction of an inch out of centre towards the slide bar, so that the stopper end, on INGOT CASTING 185 being lowered, just strikes the near side of the nozzle and then slides into the correct position. Ladles are always heated before receiving a cast of steel, to prevent the steel chilling and stopper troubles that might otherwise result. Sometimes the ladle is warmed by lighting a small coal fire on the bottom after setting the nozzle brick in position, care being taken to cover the nozzle brick with a small bent plate to prevent any ash from adhering to the ground seating; this method is good enough for ordinary purposes, but there is far less risk of skulling and "hard stoppers" if Blast Pipe J Fi-3. 87. some system of heating by air blast is adopted. Two methods are commonly used ; in one, the ladle is inverted over a shallow open top fire-box built in solid masonry (Fig. 87) ; in the other a small fire is made on the bottom of the ladle, and a blast of air blown downwards upon it from a blast pipe that can be readily removed (Fig. 88). The blast pipe A is constructed in the form of an inverted U, and is suspended from a light jib that swings about the axis of one limb. This limb fits loosely in a fixed blast pipe B in which it is free to slide up and down. The ladle, containing a small coke or coal fire, is placed close to the jib, and the latter is then swung round and the blast pipe lowered, so that the nozzle end is centrally situated in the 186 THE ELECTRO-METALLURGY OF STEEL ladle. This arrangement is very convenient as no extra hand- ling of the ladle is required, and the heat is produced on the bottom where it is most wanted. Methods of Ingot Casting. There are two entirely distinct methods of casting ingots, depending upon whether the steel enters the mould from the top or from the bottom. The former is known as " top-casting," and the latter as " bottom-casting " or "bottom-running". In either case tapered moulds may be used with their larger section at the top or at the bottom, as may be preferred. Before, how- ever, describing these different methods of casting, the various types of moulds used must be considered. Ingot Moulds. Ingot moulds are of three kinds (a) open top and open bottom ; (6) open top and closed bottom ; (c) open bottom and closed top. Moulds should be made of a high grade grey haematite iron of coarse open grain and high in silicon and graphitic carbon, as the composition of the iron used has a marked influence on their FIG. 88. life. An ingot mould is always subjected to extreme changes of temperature, more especially when teeming, and, since the heat is only applied to the inner walls, the stresses set up by unequal expansion are considerable ; rough treatment in handling also demands a high degree of toughness, so that the use of a No. 1 or No. 2 grade pig-iron is essential for the pro- duction of a good mould which will withstand the severe con- ditions of service. Moulds up to 20 inches may be square, but beyond that size are either hexagonal or octagonal in form ; the latter shapes favour a more even rate of chilling, which equalises to some extent the internal stresses set up in the outer envelope of the ingot. The inner walls of a hexagonal or octagonal mould are slightly convex, which allows for slight INGOT CASTING 187 deformation of the ingot on cooling and lessens the possibility of cracks. The degree of taper and thickness of moulds are subject to variation according to the particular views of different users. The rate of chilling or speed of initial solidification, is depend- ent upon the thickness of the walls and, therefore, the capacity of the mould for absorbing heat from the liquid steel, so that the dimensions of the mould will influence the manner of crystallisation and also the position of any pipe. The taper of an inverted mould with the larger end uppermost is often in- creased so as to amplify those conditions which are favourable to the reduction of " pipes " or their localisation at the top of the ingot. Closed and open-top moulds used small end up should, on the other hand, have only sufficient taper for the purpose of stripping, so as to prevent as far as possible the formation of deep-seated pipes. New moulds often have a thin skin of oxide or dirt on the inside faces, and until they have received a few casts of steel the skin of the ingot will be somewhat impaired by pitting. The interior of the mould should always be carefully cleaned with a wire scratch brush, and all scale or rust removed from those which have been in disuse. Cleaning alone is not sufficient for the production of a perfect ingot skin, it being further necessary to "reek " the inner faces of the mould. The term "reek" was originally used to denote the practice of depositing soot on the inner surfaces of small crucible ingot moulds by exposing them to the smoky flame of burning coal tar ; now it is used in the wider sense of covering or painting over the faces with some material which leaves behind a deposit of carbon, either on drying or heating. The older method of smoking is the best whenever it can be used conveniently, but for large moulds it becomes impracticable. Two methods of " reeking " are commonly adopted : (a) The mould is brushed over with a thin plumbago wash while it is warm enough to drive off the water. This method is open to the objection that, unless the wash is quite thin, clots of plumbago will be left on the mould and produce surface irregularities and local unsoundness of the ingot. (b) The mould is painted over with a thin coating of tar, 188 THE ELECTRO-METALLURGY OF STEEL which on vaporising leaves a thin deposit of carbon adhering to the surfaces so coated. Sometimes the tar is not boiled off before teeming, in which case there is a sudden evolution of gas at the moment the liquid steel comes in contact with it ; at the same time the mouth of the mould will be filled with smoke which prevents the teeming operation being watched. When tar is used for reeking it is better to clean the mould while still hot from a previous cast, and then to paint it over so that all the volatile matter may be driven off before it is again required for use. Ingot Pit. Moulds are usually set in a specially prepared pit, so that their upper ends are at a convenient height above the shop floor level for teeming. Sometimes the moulds are set on the floor, in which case it is necessary to provide a platform from which the ladleman can manipulate the stopper rod lever. Ingot pits are preferable from every point of view, and should always be provided wherever possible. When moulds are set on the floor level there is a far greater risk of injury to workmen in the event of a running stopper or a break out at the bottom of a mould or trumpet, besides which the manipulation of the ladle in the event of a hard stopper is rendered far more diffi- cult and dangerous. Ingot pits are usually lined with fire-brick, and the walls slightly tapered and crowned with a cast-iron curb plate to facilitate the withdrawal of any steel skull that may have accidentally covered the bottom from side to side. Top- Casting. The method of filling an ingot mould from the top is by far the most commonly adopted, and is almost invari- ably used for casting ingots heavier than about 30 cwts. Some- times split moulds are used which have no longitudinal taper and form a closed bottom. Moulds that are not split must be tapered longitudinally to ensure ease of stripping, and are used with either the small or big end uppermost. Ingots that have been cast from the same mould but in reversed positions will not have similar internal structures, so that when selecting a method of casting, not only should the relative ease of handling the moulds be considered, but also the probable position and effect of the pipe produced. As regards handling, it is clear that moulds set small end up will have to be stripped from off the ingots, and then reset in INGOT CASTING 189 position for a subsequent cast ; this does not apply in the case of ingots cast large end up, which can be removed from the moulds without disturbing their position, provided small wrought-iron eyes are fixed in the head of the ingots just before setting. Apart from the question of handling, the direction of taper exerts a considerable influence on the position and magnitude of a pipe, which may be explained by reference to Figs. 89 and 90. Fig. 89, which shows the section through an ingot cast small end up, indicates a pipe extending half way down together with a small secondary pipe. On studying the progress of solidification in such a mould it is clearly apparent that the chilling effect at the head of the mould is greater than at lower sections owing to its contracted area, the rapid loss of heat from the liquid steel exposed, and the sometimes increased thickness of the mould walls towards the top. The top end of an ingot may then solidify almost as rapidly as the bottom, so that the shrinkage, which accompanies solidifica- tion in the lower portion of the ingot, cannot be fed sufficiently by liquid steel from above ; this results in the formation of an extended cavity or pipe. Clearly, any modification of condi- tions which will delay the solidification of the upper portion of an ingot relatively to the lower, will not only alter the ultimate shape of the cavity, but will also eliminate the extended portion together with the secondary "pipes". Such conditions are better fulfilled by reversing the direction of taper, when the shape of the pipes formed will resemble that shown by Fig. 90. The solidification of the upper portion of an ingot may be further retarded by in- creasing the taper of an inverted mould, so that in this case taper can be used as a beneficial influence within limits, whereas for moulds used small end up the result would be still more disastrous. Moulds, specially made for casting large end up, are either provided with a solid bottom or must be machined all over the lower open end, so that perfect contact with a flat FIG. 89. Fio. 90. 190 THE ELECTRO-METALLUEGY OF STEEL surfaced chill plate may be ensured. Any gap between the mould and chill plate would immediately be filled with steel when teeming, which would prevent the ingot being stripped without the added labour and expense of removing the " fash ". Moulds of this type are often cast with their outside faces parallel, so that the thickness of the walls at the bottom end is greater than at the top ; this still further tends to accelerate solidification of the bottom portion. So far, then, as the use of moulds is con- cerned, those conditions which tend to reduce the extent of piping are best satisfied by top-casting into moulds, large end uppermost, when the effect of wall thickness and degree of taper may also be used to good advantage. A closed-bottom inverted mould is usually dished at the bottom, and, provided care is taken when teeming, its life may be as long as that of an open- bottom mould. An inexperienced teenier can, however, easily ruin a mould by opening out the nozzle too sharply at the start, and so burning the bottom to such an extent that " stickers " will result in subsequent casts ; this applies especially when casting high carbon steels. The bottom or " chill " plate used with open-bottom inverted moulds are cast with a deep concave depression in the centre, so that the splash caused by the stream striking the bottom is prevented from touching the walls of the mould, and from penetrating any open joint between the mould and the bottom plate. The depression in the chill plate is also rapidly filled with steel, which then breaks the force of the stream and so further prevents splashing. Bottom -Casting. This method is one that is most usually and conveniently used for casting a large number of ingots of small dimensions. If a large number of small ingots have to be cast separately from a ladle, there is every possibility that, owing to the very slow rate of teeming necessary, the steel will become chilled and too cold before the moulds are all filled. By adopting a system of bottom-casting a large number of moulds may be filled simultaneously, so that the rate of filling each mould is ex- tremely slow compared to the actual rate of teeming ; the ladle is thus emptied in a far shorter time and the proper condition of slowly filling each mould is satisfied. Closed-top moulds, having only a small conical gas vent, are frequently used, and INGOT CASTING 191 have certain advantages over the open-top moulds. A group of such moulds is shown in Fig. 91, where they are mounted on a bottom runner plate suitable for casting four ingots. The bottom plate is usually cast with four or six lateral recesses radiating from a central recess, which is accordingly either square or hexagonal. The central recess contains a centre brick which is simply fitted within it by a packing of dry sand, while the lateral recesses are suitably di- mensioned to hold the runner-bricks, which are mortised to fit into each face of the centre brick. If the centre of the mould is far from the centre of the plate an extension runner-brick of suitable length is used. Fig. 92 shows a four-way and six- way centre brick. Both the centre and the runner- bricks must be carefully set in the recessed plate with sand, so as to be flush with the top of it. The mould, or at least one side of it, will rest across the brick, so that any ^ difference in level between FIG. 91. the plate and the brick may cause the steel to run out and the ingots to " bleed". The 192 THE ELECTRO-METALLURQY OF STEEL runner-brick is closed at one end, but is provided with an orifice that is best surrounded by an annular flange, which helps to de- flect the stream of steel vertically upwards. It is most important that the moulds are set on the plate exactly central with this orifice, otherwise the side of the mould nearest the stream of steel may be washed and burnt ; careless setting may also cause cracked ingots, due to the unequal distribution of hot steel in the mould (Brearley). The steel is fed into the centre brick from a vertical fire-brick pipe, enclosed in a cast-iron or steel frame called a ''trumpet". Fig. 91 shows a section through such a trumpet, each half of which is built up of semicylindrical sections fastened together by bolts or cotter pins ; in the figure it is shown in one length, the two halves being cottered together. The pipe bricks, 0:4 t-WQj FIG. 92. FIG. 93. FIG. 94. (Fig. 93) are carefully laid and fitted together in one half of the trumpet on a bed of " compo " or ganister, the bottom brick, which is sometimes mortised at its lower end to fit the centre brick, being exactly flush with the end of the trumpet. The bell section is finally set in position, and the entire pipe daubed over with ganister or " compo". The other half of the trumpet is then laid upon it and cottered to the under half, squeezing out any excess of the bedding material. The built-up trumpet is carefully dried before use to prevent the brick pipe bursting through sudden generation of steam. The lowest section or bottom end of the trumpet has a flange, which is either bolted on to the bottom plate or held down by weights, as shown in Fig. 91 ; this is done to prevent the steel finding its way under INGOT CASTING 193 the trumpet and lifting it owing to the pressure of the column of steel in the trumpet pipe. When closed-top moulds are used, the steel rises up until it reaches the base of the vent, where it then normally chills, but if teeming at this point is too rapid and the steel very hot it may sometimes spurt out; any "fash" so formed round the vent hole is immediately removed from the mould top to pre- vent the steel being held fast in the vent. Should the latter frequently occur the vent becomes worn and enlarged, causing constant trouble with " stickers ". A plan commonly adopted is to build a small mound of loam above the vent hole and then prick through with a large nail, in this way the conical plug of steel filling the vent cannot become fastened to the mould top. Care is always taken to dry the runner bricks and moulds thoroughly, as, apart from the possibility of the bricks bursting, there is sometimes a danger of the moulds lifting through any sudden generation of steam. After the steel has become solid in the vent, the ingots may be constantly fed under pressure from the steel in the trumpet, since the bell will be at least 12 inches above the ingot top. In the case of open-top moulds, the bell will be at the same level as the ingot head, and con- sequently no feeding is possible from the trumpet ; "dozzles " and " cheek " bricks may be fitted, however, in the top of the mould to delay solidification of the ingot head, and so reduce the extent of " piping ". The solidification of a bottom-cast ingot takes place in a different manner from one top-cast. In the latter case hot steel is constantly fed from the top, so that, at the moment of filling, the steel in the head of the mould will be hotter than elsewhere ; in the former case these conditions are reversed, so that the region of piping in the ingot will be lowered, following upon the slower cooling of the lower portion of the ingot relatively to the top ; at the same time, the chilling effect at the head of the mould is greater than at the bottom, which further accentuates the tendency towards the formation of an extended primary pipe and small secondary pipes. This method of bottom-casting small ingots in groups may be safely adopted, provided there is no danger to be anticipated from the presence of pipes, which are always most pronounced. 13 194 THE ELECTEO-METALLUEGY OF STEEL There is yet one other modification of the method of bottom- casting, which entails the use of a special mould, having a partially closed bottom and used large end uppermost. Such an arrangement is shown by Fig. 94. The bottom of the mould is thick, and has a central conical hole into which fits a fire- brick plug designed to make a morticed connection with the orifice of the runner-brick. The stream of steel is guided vertically upwards without risk of washing the sides of the mould. The bottom of the ingot is exposed to a greater chilling effect, and, since the upper portion of the ingot is wider, the rate of solidification is there retarded. These altered conditions in the process of solidification tend to reduce the length of pipe and eliminate secondary pipes, so that the arrangement pre- sents distinct advantages over other methods of bottom-casting. There is, however, one serious objection, which lies in the preparation and correct setting of the mould before casting. It is obvious that any open joint between the fire-brick plug and the mould will become filled with steel and prevent withdrawal of the ingot ; unfortunately, it is not always possible to rely upon this joint being properly filled up with clay, so that the success of the method will depend entirely upon the human element. All methods of bottom-casting demand considerable care from pitmen, as any loose dirt left in the runner-bricks will be washed away and probably become entrapped in the ingot. Dozzles, Cheek = bricks, and Sinking Head. It was realised in the early days of crucible ingot casting that the formation of a pipe might be prevented by keeping the head of an ingot molten until solidification was complete at lower levels. This was accomplished by placing a strongly heated fire-brick, with a slightly conical central hole, in the head of the mould while teeming was momentarily stopped ; the steel quickly solidifies round its lower edge and holds it fast to the ingot walls. The central cavity in the brick is then filled with steel, which remains molten and serves as a reservoir from which liquid steel is con- stantly drained as shrinkage proceeds lower down. Such bricks are usually known as "dozzles" or "cores," and are made in a large variety of shapes and sizes (Fig. 95). It is noti always possible to pre-heat the largest sizes, and the beneficial effect is INGOT CASTING 195 then greatly lessened. " Cheek ' ' bricks are used for large moulds in place of the fire-clay " dozzle," and are moulded with a small lug on their upper edge which rests on the top of the mould (Fig. 96). As these bricks may sometimes become firmly attached to the ingot head, the lugs should be cracked after teeming to prevent any tendency to hold the ingot and cause it to "pull". "Sinking heads" are usually employed for large ingots, and consist of a light cast-iron or steel box, rammed up with loam, "compo," or moulding sand, leaving a central cavity shaped and tapered as desired. The "sinking head "rests on the mould top, and care must be taken, when setting in posi- tion, that the joint is well closed to prevent "fashes" being formed which will hold the ingot and cause it to pull. Some- times the upper part of a large mould is recessed all round, and FIG. 95. FIG. 96. the space filled up with " compo " or other material; in this way the sinking head becomes part of the mould itself. In many cases the use of such devices is not alone sufficient, so that charcoal is sometimes thrown on the liquid steel to further retard solidification by the heat of its combustion. Sir Eobert Hadfield advocated the use of a layer of fusible neutral slag between the steel and the charcoal, and also used an air blast to promote a higher temperature of combustion. With this pro- cess the steel sinks quite level in the " compo " or ganister-lined sinking head, leaving only a shell of steel adhering to the sides. Tun = dish Casting. The importance attached to the speed of teeming cannot be exaggerated, while the difficulty of casting a large number of small ingots under nearly similar conditions from a large ladle has been already mentioned. Group casting certainly provides one solution of the difficulty, but adds to the 196 THE ELECTEO-METALLURGY OF STEEL cost of casting, and does not produce such a reliable ingot as one top-cast large end up. It is only a few years since it was proposed to interpose an auxiliary receptacle oetween the ladle and the moulds for the purpose of dividing the stream from the ladle into several smaller streams that feed directly into the moulds. Such a re- ceptacle is known as a tun-dish, an example of which is shown in Fig. 97. The advantages of this system of casting have been well proved, and one instance may be quoted where the quantity of billets regularly scrapped on account of cracks was reduced from over 50 per cent, down to 5 per cent, and under. J. N. Kilby has given some conclusive figures proving the value of the tun-dish, which may be used even in large plants without causing any inconvenience in the shop. A separate tun-dish is usually supported above each group of moulds that it feeds, but sometimes it is attached to the under-side of the ladle and moves with it. They are generally designed to feed two moulds, but are also used for four. The tun-dish must itself be carefully fed, so that the stream of steel from the ladle is equally divided between the several nozzles. Failing to do this, the moulds will not fill at the same speed and ingots of unequal length will result, which may prove serious in cases where the moiilds are provided with dozzles or cheek-bricks. The dish is lined with either thin fire-brick splits, or a facing of rammed " compo," or ganister, and is usually deep enough to hold a reservoir of steel at least 8 inches in depth, which helps to equalise the pressure of steel above each nozzle, and, at the same time, prevents the stream of steel from the ladle splashing up from the bottom. Apart from the slow speed of filling, a further advantage is gained, since the streams from the tun-dish are steady and not forced through the nozzles under great pressure ; the commotion in the mould is therefore far less, and the skin of the ingot freer from the usual splash markings of top cast ingots. FIG. 97. CHAPTEE X. APPLICATION OF THE ELECTRIC FUENACE TO FOUNDRY PRACTICE. Introduction. The electric furnace has certain features that make it especially suitable, and in many respects superior to other steel-making plant, for the manufacture of light steel castings. It has, of course, disadvantages as well as advantages, and it is only by carefully studying the operating conditions, both technical and economic, that its suitability may be determined for individual cases. Prior to the introduction of electric melting into foundry practice, castings of thin section and high ductility were made by the crucible process, or by employing some form of small converter such as the Tropenas, Bessemer, and later, the Stock. The latter plants are satisfactory, provided that steel with a carbon content of less than *2 per cent, to '25 per cent, is not regularly required for castings of very thin section. There is no doubt that thin castings with a carbon content as low as *1 per cent, to '15 per cent, can be made, at any rate, from the Stock converter, but such blows cannot be repeated with absolute certainty under the always variable conditions of blowing. With the rapid development of motor engineering and the introduction of aeroplane construction, there has been a growing demand for steel castings of thin section, but not necessarily of small dimensions. The factor of safety is an element that has to be most carefully considered in these particular branches of the engineering industry, so that -the demand for light castings made from steel of a high degree of purity and having the necessary reliability in service has now become general. The chemical composition of crucible steel is dependent upon the quality of scrap melted, and the degree of carbon absorp- tion from the pot. Clay pots are not so satisfactory for foundry purposes as the tougher and more refractory graphite crucibles, (197) 198 THE ELECTRO-METALLURGY OF STEEL which may be allowed to cool off between melts. Apart from the high cost of the latter type, there is the disadvantage that unless special precaution is taken to add a little ore to the charge, the absorption of carbon by the steel will prevent a high degree of ductility being obtained. The iron ore is added in slight excess of the quantity required to boil out the carbon absorbed, so that the steel, when hot enough for pouring, will be oxidised and require somewhat drastic "killing" with aluminium before casting into moulds. It must not be as- sumed that castings made in this way from crucible steel are not of excellent quality, provided the raw materials are good, and, for a very small output of uniformly light castings, of which the heaviest can be poured from one or at the most two pots, there is a good deal to be said in favour of this method of steel melting. The production of mild steel from crucibles is almost entirely a matter of steel melting, and can hardly be expressed as "steel-making". When a crucible charge is melted and hot enough to pour, no attempt can be. made to control the carbon content, and it is only on examination of a casting itself that the suitability of the steel can be judged. When dead soft steel is melted in graphite crucibles, the final carbon will in most cases reach '2 per cent, and at times will fall within a wider range, reaching up to '6 per cent. Such cases as the latter are not, of course, frequent, but are mentioned merely to indi- cate that the crucible product is not under the same degree of control as converter or electric furnace steel. Prior to 1914 the converter at least held its own for foundry purposes, but the subsequent shortage of suitable raw material, and the increasing confidence in the economic performance of electric furnaces, caused the latter to gain favour for the manu- facture of light soft castings. The converter is restricted to the acid process, and consequently the composition of the final steel as regards sulphur and phosphorus is dependent solely upon the composition of the raw material used. The temperature of casting will likewise depend upon the silicon content of the charge, which must necessarily be high if a high temperature and degree of fluidity is required ; this imposes a limitation to the percentage of scrap in a charge, and to the quality of the pig-iron used. The acid process is well suited for foundry APPLICATION OF ELECTKIC FURNACE TO FOUNDRY PRACTICE 199 purposes, since the character of the slag renders its separation from the steel a matter of great ease, and therefore allows the system of lip-pouring to be adopted. The latter method certainly has advantages over bottom-teeming, which is more likely to cause " scabbing " and dirty castings, but, on the other hand, a higher casting temperature is necessary. Stopper troubles disappear and the steel always enters the mould at a steady, low pressure, due solely to the fall from the ladle lip and not to the pressure exerted by a head of steel be- hind the stream, as in bottom teeming. On the other hand, unless handshanks are used, lip- pouring is not suitable for very light work which requires a precise and rapid ladle control. The limitation of the converter to the acid process is, so far as the problem of casting is concerned, an advantage rather than otherwise, as both lip-pouring and bottom-teeming are open to choice according to the nature of the work being cast. The electric furnace offers far more scope to the steel-maker than the converter, and from a purely technical standpoint, the quality of the steel is almost independent of the class of steel scrap used for the furnace charge. It is optional to use either the basic or acid process, and the ultimate choice will depend upon the specification demanded and the quality of scrap avail- able. The temperature of casting is only limited by the fusing point of refractories, and the chemical composition of the steel is always under exact control. The relative advantages of the basic and acid processes, as conducted in the electric furnace, require further consideration. The basic process enables both phosphorus and sulphur to be reduced to very low limits, so that any chemical specification can be readily met, even if impure scrap high in both these elements is used. The deoxidation of the steel can also be con- ducted under highly reducing conditions, which certainly presents great advantages over the cruder method of " killing " wild steel by alloy or aluminium additions immediately before cast- ing. In cases where the amount of phosphorus and sulphur present in the raw material exceeds the specification limit of the casting, the basic process becomes obligatory. On the other hand, if the scrap available is sufficiently low in sulphur and phosphorus, and merely requires remelting with adjustment 200 THE ELECTRO-METALLURGY OF STEEL of carbon, silicon, and manganese, then the acid process, so far as the actual steel-making is concerned, will fulfil the purpose. If, then, it is not imperative to use the basic process, the choice of process will rest solely upon economic considerations based upon the prices of what may be called low grade and high grade scrap, and other minor advantages which each may present. The chemistry of steel-making with an acid slag is practically the same for both open hearth and electric furnaces, when applied to foundry practice, only in the latter case the entire or predominant portion of the charge will be steel scrap, which renders the possibility of over-oxidation of the bath more considerable. For this reason, the scrap should be selected and be free from rust, since the process of deoxidation cannot be carried to the same extent as under a basic slag. The acid process certainly shows a small saving in power consumption since the extra time occasioned by the use of a second slag is eliminated. The steel may be finished and poured when the correct casting temperature has been reached, assuming that the bath has been properly boiled down and is thus fairly free from oxides. The convenience of handling steel under an acid .slag has been already mentioned, and should not be disregarded when deciding upon the process to be adopted. For cleanliness and convenience of operation the acid process is to be preferred, and will be found rather more economical, if scrap of suitable quality and price can be secured. The comparative advantages and disadvantages of the crucible, converter, and electric processes for the manufacture of light steel castings are summarised and tabulated in Table I. Early Development and Statistics. It was early recognised that the electric furnace offered special advantages in certain departments of foundry practice on account of the high casting temperature attainable. The Stassano furnace was at first favoured for this purpose, but chiefly owing to the severe punish- ment of the magnesite roof, furnaces of the direct arc type are now more widely used. In the year 1909 the firm of George Fisher in Switzerland had installed a small, single-phase Heroult furnace for the production of very light steel castings, which included the first type of hollow-spoke motor vehicle wheels used in Great Britain. The first electric furnace laid APPLICATION OF ELECTRIC FURNACE TO FOUNDRY PRACTICE "201 TABLE I. COMPARISON OF THE CRUCIBLE, CONVERTER, AND ELECTRIC FURNACE PROCESSES FOR FOUNDRY USE. Crucible. Side-blown Converter. Electric Furnace. Character of 100 per cent, wrought- It is possible to blow a A charge entirely com- the raw iron or steel scrap charge containing 30 posed of scrap can be materials commonly used. per cent, of foundry used. All the foundry used, and Small proportion of scrap, p.ovided a high scrap made can be re- proportion clean foundry scrap silicon haematite pig is melted, if desired. Usu- of scrap also melted in charge. used, but 20 per cent, is ally the foundry scrap melted. Impossible to re-melt a more usual figure. Can made is worth more all the foundry scrap use up the bulk of than the steel scrap made. Value of foundry scrap made. bought for melting, in foundry scrap less Value of the foundry which case it is sold at than the raw mater- scrap is less than the a figure that has appreci- ials used. average value of the ated by treatment in the metal before blowing. furnace. Melting Very small, and depen- Usually a 14 per cent, to Loss averages about 7 loss. dent upon the cleanli- 24 per cent, loss on con- per cent, to 10 per cent., ness of the scrap version of the clean raw when using fairly clean melted. Loss en- materials consisting of scrap in a form that is tirely chemical. pig and steel scrap. not liable to excessive Loss is both mechanical oxidation in the furnace. and chemical. Loss both mechanical and chemical in basic furnaces, and chemical only in acid furnaces. Fuel cost. Coke melting expensive Cost of coke or oil fuel The cost of melting is and consumption of used for melting the raw high and depends also fuel high under forced material is very low. upon the average load draught. Gas, when Fuel cost per ton steel factor or output over a used, is cheaper. blown is almost inde- long period. Fuel cost per ton pendent of the output. melted is almost in- dependent of the out- put, when using coke. Size of unit Furnaces can be in- Minimum capacity limited The smallest satisfactory and out- stalled for any de- to ton, being the unit has a capacity of put pos- sired output, however smallest size commer- * ton, and cannot pro- sible. small. cially practicable. Pos- sible daily output much duce such a large daily output as a correspond- greater than with an ing converter plant. electric furnace of similar capacity. Suitable Generally very light Suitable for medium light Suitable for medium light types of castings only are and light cast ings. Very and light castings of any castings made. Possible to thin castings can be run, description. and tem- pour thin castings but not without risk perature since no appreciable of short-run wasters. limitation. heat loss on transfer Temperature of the steel to mould. depends upon the vari- able conlition of blow- ing, governed largely by the condition of the lining. 202 THE ELECTRO-METALLURGY OF STEEL TABLE I. Continued. Crucible. Side-blown Converter. Electric Furnace. Chemical Analysis rather irregu- Chemical composition Analysis of consecutive composi- lar, due to varying ab- under good control, giv- heats very consistent tion. sorption of carbon ing a consistent carbon owing to the exact from the pot ; sulphur percentage. Sulphur chemical control during and phosphorus de- and phosphorus in the refining and finishing pendent upon analy- charge increase slightly operations. Kemoval of sis of scrap. during the blow. Loss P and S possible by basic of Mn and Si to be al- process. Exact calcu- lowed for on addition of lated additions of Mn alloys. can be made without allowing for any oxi- dation loss. General pro- Crucible steel castings Liable to slag inclusions Case-hardens very well up perties of do not case-harden and dissolved oxides. to -2 per cent. C. Acid castings well if the carbon ex- Minute slag inclusions steel occasionally shows and nature ceeds -15 per cent. frequently cause test slag inclusions. Phy- of defects. Have a tendency to pieces to fail either un- sical tests are most con- tear or pull if the steel der the bend or tensile sistent and are not was highly oxidised test, which increases influenced by factors just before pouring. the number of wasters. other than ordinary Physical tests will This is, probably, the chemical composition only be consistent in most serious difficulty and heat treatment. so far as the chemical with converter steel. The physical tests show analyses are regular. The physical tests are a rather higher ductility not so consistent as than converter or cruc- might be expected from ible steel having the the chemical uniformity same resistance to ten- of consecutive casts for sile stress. Owing to the reasons above stated. the unlimited casting Castings are liable to temperature possible the tear on contraction risk of short-run cast- owing to the tenderness ings is reduced to a due to oxides ; this is minimum. also augmented by high sulphur. down for the production of castings in England was erected by the Braintree Castings Co., Ltd., in 1911. No considerable progress was made anywhere in the development of the electric furnaces for foundry purposes until 1914, when, owing to the rising price of hematite pig-iron and the availability of an increasing supply of cheap steel scrap, its special merits were more generally utilised for meeting the growing de- mand for light intricate castings. The early electric steel castings were far superior to the best malleable iron castings then being used in the motor engineering trade ; they were of APPLICATION OF ELECTBIC FURNACE TO FOUNDRY PRACTICE 203 uniform quality throughout, could be case-hardened, and did not suffer from the disadvantage of warping, as so commonly happens during the malleablising process. Drop forgings made in complex dies were at that time in their infancy, but are now used extensively in the place of small castings, when the shape of the article permits. Owing, however, to this latter limitation, the field for drop forgings must always be restricted. The marked development of the electric furnace for foundry purposes is shown by the following figures, which give the out- put of castings during the last few years in Great Britain and America. Previous to 1915 the returns for castings and ingots produced in Great Britain were not segregrated : 1912. 1913. 1914. 1915. 1916. 1917. 1918. 1919. Great Britain 2,000 9,288 15,600 44,901 30,000 America 4,162 9,207 8,551 23,064 42,870 64,911 58,000 42,000 The rapidly increasing use of the electric furnace in foundries is not due to any great economy in the production of the steel, but rather to a demand for a higher grade material, and to the considerably reduced loss occasioned by short-run, defective, and rejected castings. The latter include castings returned from machine shops and those rejected on failure to comply with the specified physical tests. Arrangement of Plant. The furnace installation and all ladle accessory plant are best situated at one end of a foundry ; all supplies may then be handled and the furnace manipulation conducted so as to cause the least interference with work on the foundry floor. The arrangement is also preferable from the furnacemen's point of view, as the working space around the furnace is less likely to be used as a dumping ground for boxes, sand, and other foundry material. In many shops there is one heavy crane for both furnace and casting use, and lighter ones for the manipulation of boxes on the foundry floor ; by the above arrangement of plant, the cranes are always situated in a position where they are mostly used, and are therefore less likely to interfere with one another in their operation. A battery of three 2-ton single-phase furnaces installed at the works of Messrs. Thwaites Bros., Bradford, is shown in Fig. 98. 204 THE ELECTRO-METALLUEGY OF STEEL The furnace plant is erected at one end of a long foundry bay and constitutes an entirely self-contained steel-making installation. Satisfactory means of heating the ladle are always provided, and, in cases where hand shanks are used for the transfer of steel from the ladle to the moulds, special apparatus is installed for their rapid and convenient heating, so as to minimise the production of "skulls". Gas burners are sometimes used for this purpose, but a simpler method consists of inverting each shank over a separate small coke or charcoal fire, built in an unlined steel fire-box provided with a perforated removable bottom and adjustable air blast. With this appliance shanks can be raised to correct temperature in half an hour. The floor plates, which cover the working floor space around the furnace, are sometimes built a few inches .above the shop floor level ; this more readily enables a general condition of cleanliness to be maintained round the furnace and slag pit. For the purpose of controlling the melting loss and general efficiency of the furnace, it is necessary to know the actual weight of steel poured into the ladle ; this is a difficult figure to arrive at accurately from the weight of finished castings, so that the use of a dial crane hook weighing machine is always to be recommended in foundry practice. Choice of Furnace Capacity. At the present time the electric furnace is generally used for the production of light castings of about 3 cwts. or less, and has not yet competed in the heavier trade of open-hearth castings in Great Britain. The number of moulds laid down for every ton of steel cast will usually be con- siderable, thirty to forty being quite a common figure. Owing also to the light character of the castings, the moulds will be proportionately large, so that the floor space and the number of moulds required for a ton of steel castings will be greater than in most converter and open-hearth plants. When a large number of moulds are laid down for one cast, the majority will be set and remain on the floor long before they are actually filled, so that the boxes are not in constant use. It is obvious that, by laying down and casting a fewer number of moulds which could be promptly knocked out and returned to the moulder for further use, an economy of floor space and moulding box plant would naturally result. Therefore, for a given daily APPLICATION OF ELECTRIC FURNACE TO FOUNDRY PRACTICE 205 output it is better to install furnaces of small capacity, which may be worked so as to be ready for casting at regular intervals, and so collectively provide a frequent supply of steel to the foundry in small quantities. The cost of a steel-making plant consisting of several small units will naturally be higher than one large unit having the same daily output, but the numerous advantages otherwise gained will more than compensate for the increased establishment and labour charges. It may well be expected that the electric furnace, if operated in conjunction with a basic open-hearth or converter, will find an extended field of application in the production of heavy castings ; in this event, furnaces of large capacity would be installed to meet the demands of the heavy engineering trade. Specification and Mechanical Tests. Castings are generally ordered to fulfil a physical test specification, no restrictions being placed on analysis other than for phosphorus and sulphur ; this rule allows the steel-maker far more scope in the choice of analysis, which can be varied according to the heat treatment to which the castings are submitted. The standard specifications for steel castings adopted by the American Society for Testing Materials in 1912 are given below : Hard.' Medium. Soft. Tensile stress per sq. inch 36-0 tons 31-0 tons 27-0 tons Yield point 16-0 14-0 13-0 Elongation on 2 inches . 15 per cent. 18 per cent. 22 per cent. Reduction of area .... 20 25 30 Cold bend 1 inch x inch 90 120 All castings to be annealed and slowly cooled. Some American Marine specifications are more exacting than the above, as shown by the following example: Tensile stress, 36*0 tons ; Yield, point, 18*0 tons; Elongation on 2 inches, 17 per cent. ; Reduction of area, 25 per cent. ; Bend, 90 (1 inch x ^ inch). The above specifications are intended for steel castings with phosphorus and sulphur below '05 per cent, and are easily met by electric steel made either from cold scrap charges or by the duplex process. In America all Govern- ment castings are ordered in the annealed condition. 206 THE ELECTKIOMETALLUBGY OF STEEL The specifications adopted by the British Engineering Standards Committee for marine castings are very similar to the above, but do not generally apply to the light intricate variety of automobile and other such castings. When specifications are given, it will be seen from the figures given below how easily they can be met and, moreover, outclassed to an extent which places electric steel castings in a special category. The following are typical tests of electric steel castings made by the basic process from cold scrap charges : Max. Stress, Yield Point, Per Cent. Per Cent, tons per tons per Elongation Reduction sq. inch. sq. inch. on 2 inches. of Area. I. Analysis, C -15 ; Mn -1 ; Si -21. (a) As cast 34 3 23-0 26 31-1 21-8 28-5 II. Analysis, C -2--25; Mm -5--5S; Si -3--3S. (a) As cast 30-0 16 (6) Annealed at 900 C. to 950 C. and slowly cooled. 31-71 22-14 34 52-7 35-0 23-18 23 34-6 (c) Same treatment as (b) but from a large casting. 31-07 17-73 33-25 52-2 27-01 18-75 38-5 57-8 (d) Annealed, water quenched from 750 C. and tempered at 550 C. 39-05 27-93 22-0 38-3 III. Analysis, C -4 per cent. ; Mn -5 per cent. ; Si -25 to '35. (a) Annealed and cooled in air. 40-09 27-01 17-25 21-1 42-6 24-4 20-0 Acid steel made from scrap low in phosphorus and sulphur will give somewhat similar physical test results. At one British foundry, equipped with several acid lined furnaces, the steel is practically standardised and made to the following analysis: C . . . . '25 - *3 per cent. Mn -8 Si -3 P and S under '04 ,, each. The castings, which are very light and intricate, are all annealed at 900 C. for 1-| hours and air cooled ; test pieces submitted to this treatment regularly give the following physical test results : APPLICATION OF ELECTRIC FURNACE TO FOUNDRY PRACTICE 207 Max. stress ...... 30/35 tons. Yield point .... . 18/21 Per cent, elongation .... 17/20 ,, reduction area . . .30 per cent. Annealing:. The simplest process of annealing castings con- sists of: 1 . Slow reheating to a temperature above the highest arrest or Ac point. 2. Soaking at that temperature, so as to break down the coarse crystalline structure. 3. Slow cooling through the recalescence points to about 200 C. to 300 C., after which rapid cooling in air is permissible. The above treatment is sometimes varied by rapidly cooling the casting from the soaking temperature to that of the lowest recalescence point, which means a drop of about 150 C. for mild steel, and then slowly cooling from that temperature ; this treatment gives a finer structure than if the steel were allowed to cool slowly from the soaking temperature down to its low- est recalescence point. If an unannealed casting is examined microscopically, it will be found that the structure is either coarsely granular or else exhibits a large irregular, triangular shaped pattern. Both these structures are due to the prolonged high temperature to which the steel is exposed after solidification, and differ only by reason of slow or rapid cooling respectively from the tem- perature of the highest to that of the lowest recalescence point. This coarse structure is a source of weakness, as the ferrite areas are more easily torn apart than when irregular in form., more finely disseminated, and more closely interlinked. By soaking such cast steel at a temperature just above the highest arrest or Ac point these coarse crystalline grains are broken down, and complete diffusion of the carbon results. Slow cooling from this temperature allows the excess of ferrite to fall out of solution again, with total elimination of the original structure. The microphotographs of similar magnification shown in Figs. 99 and 100 illustrate the marked change brought about in the crystalline structure of a mild steel casting by an annealing 208 THE ELECTRIC-METALLURGY OF STEEL treatment. The steel contained *2 per cent. C ; '22 per cent. Si ; '53 per cent. Mn, and it will be seen how the coarse crystalline structure of the steel as cast has been entirely changed to one consisting of finely disseminated and irregular shaped patches of ferrite and pearlite. The marked improve- ment in the ductile properties of this particular cast of steel resulting from this change of structure is shown by the follow- ing test results: As Cast. After Amiealing . Ultimate stress per sq. in. 3O46 tons. 30*9 tons. Yield point . . . 18'8 Elongation on 2 ins. . 18*0 per cent. 34'0 per cent. Per cent, reduction of area 20'0 ,, 40'0 ,, It is obvious from these figures that a steel casting in an annealed condition will be far more resistive to sudden fracture by shock than when in the " green " state. The grain size of annealed steel will depend upon (1) the soaking temperature to which the steel is raised ; (2) the rate of cooling from the soaking temperature to that of the lowest recalescence point ; (3) the rate of cooling from the lowest re- calescence point to atmospheric temperature. The temperature of annealing rises from 800 C. to 850 C. for medium carbon steels up to 900 C. to 950 C. for mild steel, while higher soaking temperatures tend to produce a coarser grain. Importance is not usually attached to the effect of rapid cooling from the soaking temperature to that of the lowest recalescence point, but in some cases the practice is followed of opening the annealing stove doors to cause rapid cooling, and then closing them up again until the castings are ready for re- moval. It is generally acknowledged that the finer the grain of an annealed sample of steel the better will be its physical properties, and to secure this in practice the castings are often withdrawn from the annealing stove and cooled in air. Excellent tests may be obtained in this way, but unfortunately there are risks of setting up internal stresses in castings which vary in thickness, owing to the rapid and irregular rate of cooling. In America this point is considered of sufficient importance to warrant a general stipulation that all annealed castings shall be FIG. 99. FIG. 100. \To face p. 208. APPLICATION OF ELECTBIC FURNACE TO FOUNDRY PRACTICE 209 slowly cooled. Castings cooled slowly will give a poorer test generally, but will be more reliable in service. The process of annealing also removes the unequal contrac- tion stresses set up in a casting during its initial cooling ; this applies particularly to castings that vary considerably in thick- ness, and which may show " pulls " due to this cause. There is considerable diversity of opinion as regards the advisability of annealing dead soft steel castings containing less than '^5 per cent, carbon. It cannot be denied that annealing will im- prove the tensile, yield point, and elongation figures of all carbon steel castings, by converting the original coarse struc- ture into one of finer grain which produces a fibrous fracture. However, rough test pieces cast \ inch thick from such low carbon steel should bend double when cold, so that, so far as ductility and resistance to shock are concerned, castings made in this low carbon material will, in the " green " state, fulfil the usual requirements. Therefore, unless annealing is actually specified there seems little reason to incur the added expense of improving the quality beyond what is required by the user for the particular purpose intended. The question of annealing dead soft castings must then be left to individual manufacturers to base their decision according to the demand. Unannealed castings, as has been stated, will always be unequally stressed, and their quality in this respect will be improved by heating to a few hundred degrees centigrade. Defects of Steel Castings. The production of sound steel castings depends far more upon the art of moulding than of steel-making, and for the purpose of remedying defects a careful distinction must be drawn between their possible causes. De- fects in castings may be sometimes due to an improper condition of the steel when cast, but are more generally caused by unsuitable methods of casting, more especially in respect of the nature and construction of the sand mould. Careful examina- tion of a defective casting will usually provide definite evidence as to the cause of the defect, which in certain cases can be remedied by suitable modification of the composition and physical condition of the steel when cast. Blowholes. The term " bio whole" is generally applied to cavities formed by the generation of gas, which accompanies 14 210 THE ELECTRO-METALLURGY OF STEEL the chemical reaction between the carbon constituent of the steel and dissolved iron oxide that takes place on solidification. In the case of ingots such unsoundness is generally due to the im- perfect removal of the dissolved oxides from the steel before cast- ing. This does not always apply in the case of castings, since it is quite possible for sufficient oxide to be formed and absorbed by the liquid steel in the moulds, owing to generation of steam from a damp sand skin. The blowholes formed by imperfectly de- oxidised steel will be found distributed irregularly throughout the casting from its skin inwards, and will usually be lined with a thin colour film of iron oxide. An unsound casting will have unusually sharp edges, owing to the expansion of the outside skin, which is caused by the internal pressure of the gases generated as solidification proceeds. Steel that may produce a perfectly sound casting in dry sand will sometimes blow when cast into imperfectly skin-dried "green sand" moulds; this is due to the steam being generated from the moisture present, which causes local oxidation unless there is sufficient silicon and aluminium in the steel to neutralise its oxidising influence. Blowholes, which are thus due to the action of water vapour, are generally confined to the region of the skin, and do not persist to the centre of the casting in other than exceptionally bad cases. If the trouble cannot be remedied by ensuring that every mould is properly dried, it may be greatly mitigated by a more liberal addition of aluminium to the steel. Silicon will also assist in the same way as aluminium, but its action is not so certain and rapid. For the above reason alone it is quite. a common practice to add one pound of aluminium to a ton of steel in the ladle ; this applies equally to electric, crucible and converter steels. Short-run Castings. When steel is poured into a sand mould, it may happen that, owing to insufficient temperature or to the thinness of the pattern, it fails to fill all parts of the mould entirely and produces a " short-run " casting. For certain engineering purposes lightness may be of prime im- portance, so that it is only by increasing the casting tempera- ture, and not by slightly thickening the pattern, that this difficulty can be met. To obtain a very high casting tempera- ture may entail prolonged heating with increased power con- APPLICATION OF ELECTEIC FUENACE TO FOUNDEY PEACTICE 211 sumption, so that it is generally more economical to reserve as many as possible of the lightest castings for one heat, which may be cast specially hot for the purpose. Gas Cavities. Insufficient venting or porosity of the sand may prevent the rapid and complete displacement of air from the moulds while filling, in which case bubbles of gas may become entrapped, and remain under the skin on the " cope " side of the casting. Large and isolated cavities resulting from this cause cannot be confused with blowholes formed by oxidation of the steel before solidification. Brittleness. This defect is entirely independent of the method of moulding and pouring, and is solely due to the con- dition of the steel when cast. Brittleness of a cold casting may be due to a moderately high carbon content, in which case annealing will often suffice to remedy entirely a defect which is then only apparent in the unannealed condition. Occasionally brittleness will be exhibited in a mild steel casting, and is then characterised by a bright coarsely crystalline fracture. This condition is usually due to a high silicon content, which occurs more especially in acid electric steel owing to the reduction of silicon from the slag under highly reducing conditions. Anneal- ing will generally suffice to remedy this fault, which, however, can be prevented by careful control of the silicon alloy additions and the condition of the slag during the finishing operations. Slag Inclusions. Slag inclusions are sometimes present in both acid and basic steel as very minute particles, which under the microscope are found lying along the boundaries of the crystal grains. The slag particles usually appear strung together like a chain of beads, three such chains usually meet- ing in a common point of intersection. The grains, whose boundary lines they mark, are those which result from the primary formation of pure equiaxed iron crystals during the process of solidification, and which continue to grow outwardly in all directions until they meet one another. During the process of cooling, subsequent to solidification and diffusion of the carbon into the iron crystals, these slag particles exert a selective attraction for the ferrite constituent, so that in low carbon steels, they are found embedded in an area of pure iron from which the carbon bearing constituent "pearlite" is totally 212 THE ELECTEO-METALLUEGY OF STEEL absent ; Fig. 101 shows their appearance at a magnification of 100 diameters. These chains of slag inclusions cause planes of weakness, since the metal that divides their particles and inter- links the neighbouring grains consists of weak ferrite. Steel containing these inclusions frequently fails under mechanical test in spite of satisfactory chemical analysis and heat treat- ment. These minute slag inclusions owe their origin to silicates, which exist in the liquid steel either in a state of solution or more probably in an emulsified form at the time of casting. During the process of solidification the silicates are expelled from the liquid steel, and coalesce to form minute chains of segregates. Highly siliceous compounds have considerable power of coalescence, and were it not for this property, it is doubtful whether these slag inclusions would ever become microscopically visible as such definitely arranged segregates. It is also possible that part of the silicates are not formed until crystallisation of the ferrite begins, which then allows a reaction between the silicon and any dissolved metallic oxides to proceed. In this way the formation and segregation of the silicates would be simultaneous. Kepeated annealing is sometimes sufficient to break down these ferrite areas, but only after actual dis- placement of the slag particles. Shrinkage Cavities or " Draws ". Cavities due to shrinkage of the steel during solidification in the moulds are analogous to " pipes " in ingots, and are generally governed by the same laws of formation. The prevention of shrinkage cavities is a question of moulding, but may be assisted to a slight extent by reducing the casting temperature and the percentages of manganese and silicon to a minimum, consistent with other necessary factors. Tears, Cracks, or Pulls. This particular form of defect is due to failure of the steel to resist rupture under the tensile stresses developed during either equal or unequal contraction on cooling. Here again, the remedy usually lies in the method of moulding or alteration of the pattern, when possible. Naturally, steel that is tough at a high temperature will best resist fracture under tensile stress, and will then either yield uniformly itself, or compress portions of the sand mould and cores so as to accommodate itself freely to reduced dimensions FIG. 101. [To face p. 212. APPLICATION OF ELECTRIC FURNACE TO FOUNDRY PRACTICE 213 without distortion. Steel that is most resistant to this type of defect will have a high degree of chemical purity, being as free as possible from dissolved oxides and sulphur, both of which cause "red-shortness" or brittleness at high temperatures. It is, therefore, also in the hands of the steel-maker to obviate these defects, which are least pronounced in basic electric steel. CHAPTEK XI. CHARACTERISTIC FEATURES AND PRINCIPLES OF FURNACE DESIGN. THE design of electric arc furnaces for steel-making involves a careful study of the special metallurgical conditions required, and of the electrical and mechanical means by which these con- ditions may best be fulfilled with simplicity, economy, and regularity of operation. An electric furnace for steel-making is above all a metal- lurgical appliance, and as such must be designed as far as possible in accordance with certain conditions imposed by the particular process adopted. It is obviously impossible to embody in any furnace every feature that is desirable from a metallurgical standpoint, as this could only be done by sacrific- ing certain fundamental principles of the mechanical and electrical design. The result then must invariably be a com- promise, in which the mechanical and electrical features are to a certain extent subordinated to and ruled by the metallurgical conditions imposed. For this reason it will be useful to outline the chemical, physical, electrical and mechanical conditions, which are either purposely or unavoidably produced during the basic or acid process of steel-melting and refining, and to point out to what extent these conditions influence the general con- struction and lining of a furnace installation. Chemical Conditions. (a) The various chemical reactions between basic slag and liquid steel proceed in most furnaces under the influence of intense local heat, which, especially under reducing conditions, causes slight volatilisation of the slag con- stituents ; the basic fumes so formed have a marked tendency to flux the acid portion of the furnace lining, namely, the silica roof and walls. (6) A direct arc striking downwards on to a slag blanket at an inclined angle frequently causes rotation about the axis of (214) FEATURES AND PRINCIPLES OF FURNACE DESIGN 215 the electrode, and in small furnaces, where the walls are usually close to the arc, this circulation may cause considerable erosion of the hearth at the slag line. This action is, of course, greatly influenced by the temperature and the corrosive power of the slag. (c) Certain operations, notably that of carburising, must be performed when the bath of metal is free from any slag cover- ing, and it is also necessary in many cases to remove one slag prior to the formation of another. The removal of slag by skimming would be an almost impossible operation to perform in a fixed furnace, more especially as the level of the slag line is not always the same in successive heats. (d) Chemical erosion of the bottom is sometimes severe, owing to the close proximity of the arc zones during the melt- ing down stage and the local generation of heat in the hearth of conductive bottom furnaces. The bottom has, then, a tendency to become deeper, so that it would be impossible always to drain a furnace provided with a fixed taphole ; this would occasion serious difficulties owing to the possible contaminating influence of a residual quantity of an alloy steel on a subsequent charge, and also to the impossibility of effecting any repair to the bottom, so long as any steel covered the worn or damaged portion. (e) Burnt-lime, which is generally used as a flux in the basic process, always contains a quantity of slaked powder. On charging this flux into a hot furnace, the light dust rises with the natural upward current of air, and so comes into contact with the roof, which, being almost invariably made of silica brick, is liable to be fluxed. To avoid this action, limestone has fre- quently been substituted for lime, although certain disadvantages that are then introduced hardly justify its use. Experience also shows that the erosion of the silica brick is worst at that point where these dust-laden gases escape from the furnace around the electrodes ; here the annular opening is constricted and the increased velocity of the gas escaping under pressure accentuates the fluxing action. Any means by which these gases could be prevented from so escaping would prolong the life of that part of the lining. (/) The various metallurgical processes depend upon de- finite chemical reactions, which must be allowed to proceed 216 THE ELECTRO-METALLURGY OF STEEL uninfluenced by chemical conditions other than those purposely introduced. Mechanical disintegration of a furnace hearth or roof will invariably exert a contaminating influence upon the metal and slag, and in certain cases will render it impossible to maintain the desired chemical conditions. Such disintegration and failure of refractory materials may also be caused by im- proper treatment during their preliminary heating, as has been explained elsewhere. Care must therefore be exercised in the choice of suitable materials and in the method of employing them for the hearth and roof construction. The highly reducing properties of certain slags will be vitiated, if not destroyed, by the oxidising influence of air, when allowed to enter the furnace too freely. This demands attention to door construction and to the restriction, as far as possible, of the annular openings surrounding the electrodes, as the free escape of gases at this point induces natural convection currents of air through the furnace. (y rigid bronze plates, which are grooved to prevent lateral slip on movement of the cables ; these clamps should be frequently examined, and any slack, caused by spread- ing of the cables between the plates, taken up. When socketed cable lugs are used great care must be taken to secure a perfectly sweated joint, and it is also advisable to pass a set screw through the wall of each thimble or use some other locking device as a further precautionary measure. Under no circumstances should sweated socket joints be used, unless only a moderate current density through the conductors, cables, and especially their joints, is allowed for. Local heating, when once it occurs, leads to considerable trouble with this type of connection, which, how- ever, is the most convenient and practicable if properly designed and made. The length of conductors should be the minimum possible, so as to reduce copper resistance losses, besides lowering their initial cost. Lengthy conductors, carrying 6000 amperes and over, introduce considerable reactance into the circuit, par- 224 THE ELECTKO-METALLUKGY OF STEEL ticularly when they lie close to steelwork. This may be a good fault in some respects, but it is usually preferred to cut down this uncontrollable reactance as far as possible, and introduce external reactance coils of definite design. The furnace trans- former sub-station is best located as near to the furnace as possible, the transformers themselves being arranged at such a height that their low tension terminals are at a convenient level to receive the flexible cables from the furnace. Electrode holders frequently serve as an integral part of the load circuit, and are then usually made of bronze and rigidly connected to the copper conductor bars. Sometimes they are constructed to grip the ends of the copper conductors firmly against the electrodes, and do not themselves carry any current. Furnace conductors are always exposed to radiated heat, and are usually designed for a normal current density not exceeding 1000 amperes per square inch. Should the conductors be very large and their skin effect considerable, the current density should then be reduced propor- tionately. With this current density the resistance losses are very small, and the circuit connections on the furnace will not be over- heated, if properly made. All insulating bushes, washers and plates should be made of a material that remains unchanged at the high temperatures above and around the furnace, as con- stant trouble is caused by short circuits when the binding com- position of the material softens. Transformer Capacity. The power capacity of steel furnaces is a matter that very considerably influences the cost of power used per ton of steel and the quantity of steel produced. It has been explained in Chapter VI. how the ratio between the power wasted as radiated heat and the useful power available for melt- ing will affect the power consumption per ton of steel and the total output. Those arguments in favour of a large ratio apply in all cases where the maximum possible daily load factor and out- put from a given furnace is in no way limited by shop conditions. They are open, however, to modification in cases where a fur- nace, having a required holding capacity, is installed to operate intermittently, and therefore at a low weekly load factor. Under such conditions, it may be an advantage to reduce this ratio by providing less power than usual : this procedure, of course, increases the power consumption per ton of steel, but, FEATURES AND PRINCIPLES OF FURNACE DESIGN 225 at the same time, lowers the maximum demand or flat rate payment to an extent that will show a small advantage on balance. It has been already shown (Fig. 77) how the cost per unit rapidly rises under a reduced load factor, and for this reason, if the maximum demand or flat rate charges were lowered, with a corresponding rise in monthly load factor, the net result would be a saving in the power bill, notwithstanding a small increased power consumption and lengthened heats. The tendency is, therefore, to provide rather less power for furnaces of a given capacity operating intermittently and at very low daily load factors, as this, besides reducing the power bill when partly charged upon a flat rate or maximum demand rate, also reduces the initial cost of the furnace plant. Heat Conversion of Electrical Energy. Direct arc furnaces may be divided into two distinct classes, according to the man- ner by which the electrical energy is converted into heat. I. Furnaces in which a conductive hearth, or metallic con- ductors imbedded in the hearth, become an integral part of one of the load circuits, which is generally a neutral return con- ductor. II. Furnaces in which the heat is developed in direct arc circuits entirely independent of the furnace lining. Opinions are divided upon the relative merits of these two distinct types, and there is no doubt that equally good steel can be produced with either. From purely technical standpoints both designs have their relative advantages and disadvan- tages. The several types of conductive hearth furnaces of small capacity, operated by either two-phase or three-phase current, are provided with only two electrode circuits. In comparison, then, with furnaces of the three-phase three-arc type of similar capacity, the two-arc furnaces should show a reduced electrode consumption, and have the advantage of a simplified electrode controlling gear. A more solid roof construction is also possible, the roof being usually arched about one horizontal axis in con- formity to the rectangular-shaped body generally adopted. A rec- tangular form certainly causes a greater radiation loss for a given internal capacity than one approximating to a sphere, but has an advantage in reduced constructional cost and simplicity of design. 15 226 THE ELECTEO-METALLUEGY OF STEEL It has been repeatedly stated that a certain amount of heat is generated in a conductive furnace hearth by simple resistance, which greatly assists in the manufacture of alloy steels by preventing the formation of a frozen layer of metal on the bottom, due to the chilling effect of the cold alloys added. This bottom heating is usually equivalent to about 5 per cent, to 8 per cent, of the full power input, or 27 kw. to 43 kw. respectively for a furnace operating under a load of 600 K.V.A. at a power factor of *9. If such a hearth is homogeneous and at a uniform tem- perature at any horizontal section, the current density and the resistance heating developed will be uniformly distributed. But, although these hearths are generally constructed of layers of material having a progressively lower conductivity towards the top, it is doubtful whether the greater part of the heat is generated in a region near the bath, since the conductivity of the top layers increases very considerably in proportion to that of the bottom layers at high temperatures. It is, therefore, impossible to know exactly in which part of the hearth heat is generated by resistance ; it is equally clear that the total heat generated is not entirely absorbed by the bath of steel, but is partially lost as a result of increased radiation from the bottom shell plates. Should, however, the distribution of current through the hearth not be uniform, then there will be more in- tense heating at certain spots in the bottom, but this is not likely to occur if a well-constructed hearth is at a uniform temperature, as would be the case under a covering bath of steel. J. Bibby in a paper contributed at a joint meeting of the Institution of Electrical Engineers and the Iron and Steel Institute in 1919, has gone further than this, and gives his opinion that the bulk of the heat generated is dissipated by radiation outside the furnace, rather than any being absorbed by the metal. It seems very doubtful, therefore, whether the conductive hearth furnace can offer any advantage from the point of view of bottom heat- ing over the top arc heating furnace. Steels containing 20 per cent, and more of alloyed metals can be regularly made in the latter type without any steel chilling on the bottom, the natural precaution of making small additions at a time followed by vigorous stirring, which is always necessary for mixing alone, being amply sufficient to secure this result. FEATURES ATD PRINCIPLES OF FURNACE DESIGN 227 Bath circulation or auto-mixing is another advantage claimed for these furnaces. Bath circulation can only be due to either heat convection currents or to electro-magnetic effects set up in the bath itself. It is obvious that a bath of steel heated by direct arcs will be hottest at the top, and, unless the bottom can be heated to a still higher temperature, convection currents cannot possibly be set up. Electro-magnetic circulation can in no way be due to the result of magnetic fields dependent upon the high permeability of iron, which becomes non-magnetic above about 750 C. Weak magnetic fields of varying intensity and polarity are, however, set up by fluid conductors carrying heavy alternating currents, and it is then theoretically possible for attraction and repulsion between different parts of a bath of metal to be caused by current traversing it in different directions. The mutually acting magnetic forces induced by solid conductors are visibly displayed by the movement of neighbouring cables of different phases carrying very heavy currents, but it is diffi- cult to say whether sufficient forces are actually developed to cause and maintain movement of heavy masses of molten steel as a result of currents traversing different paths through the bath. With certain types of conductive hearth furnaces consider- able difficulty is experienced in securing a conductive circuit through the bottom, when cold. In these cases it is necessary to use auxiliary gas or oil heating, otherwise the furnace can only be operated under an unbalanced and diminished load. To obviate this difficulty various modifications have been intro- duced embodying the use of an auxiliary upper electrode, which is connected to the conductive hearth cables and can be used to complete the load circuits, thus enabling a balanced load to be applied. Under continuous operation, however, no difficulty need be anticipated through failure of the hearth to conduct the full circuit current a few minutes after applying load. Experi- ence has proved that conductive hearth furnaces can produce excellent results, but, at the same time, the character of the hearth, owing to its use as an electrical conductor, is not so reliable under all conditions of service as those constructed of similar material but independent of the electrical circuits. The main advantage of the other class of arc furnaces, apart 228 THE ELECTKO-METALLUBGY OF STEEL from the ease with which they may at all times be heated electrically, lies in the durability of the hearth, which, if properly constructed, should never cause any metallurgical or other difficulties by premature and sudden failure under any normal conditions. I'he disadvantages are those due to the increased number of electrodes and their raising gear, higher electrode consumption, weaker roof construction, and the greater com- plexity of the load regulation of three-phase three-arc circuits. These disadvantages, of course, only apply when comparison is made with conductive hearth furnaces having not more than two top electrodes. Although this comparison is more especially ap- plicable to direct arc furnaces alone, yet practically the same arguments apply when comparing conductive hearth furnaces with those of the indirect arc type. Power Factor. The question of power factor is of great importance in electric furnace design, and must be considered in its relation both to the reactance of the load circuits and to the inherent reactance of the transformers or generating plant. According to most power contracts the flat rate or maximum demand rate charge is based on a K.V.A. and not on a K.W. input, so that payment is made on a figure which does not re- present the true maximum rate of power absorbed. Since the ratio of K.W. to K.V.A. is proportional to the power factor, the nearer the latter approaches unity, the better it will be for the con- sumer. At the same time, there is a clause in most contracts by which the consumer guarantees that the average power factor shall not be less than '8 or '85, so that careful attention must be given to the design of the transformers and the load circuits to the furnace electrodes. The relation of power factor to the capa- city of transformers for doing useful work has already been men- tioned, and determines the initial cost per K.W. capacity of plant installed in contradistinction to K.V.A. capacity, which does not truly indicate capacity for doing useful work. The objectionable features of heavy load fluctuations, and the extent by which they may be reduced by reactance coils, has been fully dealt with in Chapter IV. Where, however, such reactance coils are introduced into the load circuits, they should be designed so as not to reduce the power factor seriously at or below normal full-load current, but only on heavy current over- FEATURES AND PRINCIPLES OF FURNACE DESIGN 229 loads. Eeactance coils designed to fulfil these conditions will considerably reduce heavy power or K,W. overloads. Certain furnaces have been designed in which the amount of reactive resistance introduced into the load circuits is such that, at normal load, the power factor is about '7. Under these con- ditions the reactive effect is so great that on dead short circuit the power in K.W. is considerably reduced below normal full load, and the normal full load current only increased by 41 per cent. In this way a practically automatic load control, resulting in excellent load factors, can be obtained at the expense of power factor. Furnace design, from the point of view of power factor, will then depend to a great extent upon the limiting figure allowed by the power companies, the increased initial outlay for larger transformers, and the cost of electric energy as purchased on the basis of either K.V.A. or K.W. demand. The power factor of furnaces will be influenced by the nature and relative disposition of the circuits between the transformer terminals and the electrodes. Any individual circuit carrying an alternating current is influenced by the magnetic field set up, but if two or more such circuits carrying currents that are out of phase are brought close to one another, the effect of the magnetic field set up by one will be partly counteracted by the magnetic field due to the others ; for this reason the resultant reactive voltage induced in each circuit will be very much reduced, and the power factor considerably less affected. There- fore, when high power factors are desired in furnace construction, it is preferable to keep the several conductors close together, to avoid a closed iron circuit around any one set, and to support them as far as possible away from all steel parts. The length of a circuit also affects power factor by increasing the self- induction. Mechanical Features. Furnace Mounting. It has been previously indicated that provision must be made for tilting electric furnaces owing to the necessity of skimming a bath and completely draining the furnace hearth. The earliest types of both arc and induction furnaces were fixed, but the necessity for tilting and emptying furnaces for steel melting soon became evident. 230 THE ELECTRO-METALLURGY OF STEEL There are two usual methods of mounting furnace bodies for tilting : 1. The furnace body is carried on rocker castings, which either roll forward on a flat base plate, or are supported on sets of rollers which allow the furnace to roll about a horizontal axis. 2. The furnace body is provided with trunnions mounted on trunnion bearings, and can be tilted by hand or mechanically driven gearing ; this method is generally confined to furnaces of small capacity. Tilting Gear. When hydraulic power is available, tilting may be effected by one or two rams situated under the rear side of the furnace ; this is undoubtedly the most reliable and least complicated method of tilting. Various mechanical methods have been used, none of which can be said to give entire satisfaction. The tilting bar may consist of a heavy screw, fed forward or backwards by a rotating nut, or of a straight or curved rack engaged by a pinion ; in the latter case the rack is fixed to the furnace body concentric with the rocker castings supported on roller mountings. The screw and nut method, which is almost universal in Great Britain, is likely to give trouble through failure of the screw thread or ball races, unless the gearing is carefully cleaned and greased at frequent intervals ; this, however, is often neglected as the tilting gear is usually in a position that is not easy of access. A rocking arm or connecting link is also used for tilting ; in this case a plain tilting bar is attached by a swivel joint to the furnace body, and connected at the other end to a crank, which is made to slowly revolve ; this method is very simple, and has proved most satisfactory for small furnaces of 2 tons capacity and under. In the case of any electrically driven tilting gear, limit switches should be provided to prevent the furnace being tilted too far either way, which might cause disengagement of the tilting bar or strain on the tilting mechanism. Electrode Regulating Gear. The gearing used for adjusting the electrodes is operated either electrically or by hand, pro- vision being usually made for operating by either means at will, by introducing a simple clutch device. Various methods of FEATURES AND PRINCIPLES OF FURNACE DESIGN 231 gearing have been employed, which utilise either a rack and pinion drive, a nut and screw, feed, or a simple rope and winch hoist. The downward movement of any electrode should only be possible so long as the full weight of the electrode and its mounting is carried by the lifting gear, and, as soon as resist- ance is offered by the charge or furnace bottom to further movement of the electrode, the mechanical gearing should be thrown out of action ; in this way no excessive strain can be thrown on to the gearing. The same applies to the limit of upward travel, and for this purpose limit switches and clutches are also used. Power driven gearing is frequently operated in conjunction with automatic regulators, and is then designed with a sufficient braking action to prevent any tendency of the motors to over-run, which would cause incessant hunting. The tendency to hunt is more pronounced where a rack and pinion lifting gear is used, but, on the other hand, a screw and nut feed is more liable to failure through troubles arising from wear of the screw threads. Electrode raising mechanisms are always exposed to heat and dirt, and should, therefore, be heavily constructed and enclosed as far as possible. Electrically driven gearing is necessarily heavy and cumbersome, and difficult to operate manually owing to its low mechanical efficiency, especially when using heavy electrodes. Air pressure has also been employed for adjusting light electrodes, but has not been developed to any extent. Hydraulic control is now being introduced in place of electrically driven gearing, and is being satisfactorily developed for automatic regulation. The raising gear for all electrodes may be mounted together on one side of a furnace shell, or may be divided and attached to two or more sides. When this latter arrangement is adopted the raising gear is always set to one side of the plane of tilting, so that the furnace may be provided with pouring spouts and charging dpors both back and front ; this is certainly an ad- vantage as all skimming operations can be performed over one spout, while the pouring spout is reserved for casting. When the several raising gears are set side by side, they are either attached to the back of the furnace or to one side of the plane of tilting so as to utilise the double-spout construction. 232 THE ELECTRO-METALLURGY OF STEEL Electrode Holders. There is considerable variation in the construction and design of electrode holders, which may be roughly classified according to whether the holder (i) is itself the conductor, (ii) clamps the conductor to the electrode and carries its weight, or (iii) acts as a support for some independent clamping device which is really an integral part of the whole. A holder belonging to the first class should be made of metal of high conductivity, be water-cooled or specially designed for air cooling, and possess sufficient flexibility to permit of rapid opening or closing with minimum risk of fracture. A maximum degree of flexibility has been obtained by hinging two portions of a holder together ; the rigid portion is firmly fixed to the electrode arm, the hinged portion being thus alone free to move. Such holders are generally made of bronze, but steel has also been used successfully. The same conditions apply to holders of the second class, only in this case there is no necessity for using a gun metal or bronze of high conductivity, since the holder itself is not called upon to carry current to the electrode. In the case of the third class the copper plate conductors are gripped to the electrode by an independent flexible clamp, which is supported by a fixed annular collar fastened to the movable carriage or gallows arm. The flexible clamp in this case is only under lateral tension and vertical compression, and is not subjected to any bending or twisting forces. The supporting collar, which actually carries the weight of the electrode, is not required to open and close, so that its construction can be greatly simplified. This division of a holder of the first class into three distinct parts certainly simplifies the construction of each part individually, but renders the whole less compact. A perfect holder has yet to be designed which will combine flexibility, strength, electric conductivity, and rapid means of clamping, and at the same time preserve these characteristics under all normal conditions of working. The use of graphite electrodes greatly simplifies the construction of holders, and the required degree of flexibility is not sacrificed to the same extent as for large diameter amorphous electrodes where it is necessary to have great strength and rigidity. Furnace Doors. A furnace door of good design should com- FEATURES AND PRINCIPLES OF FURNACE DESIGN 233 bine as far as possible the following features : (a) It should be tight-fitting but allow the free escape of gases under pressure from the furnace interior ; (b) it should be capable of easy and rapid movement; (c) it should be possible to open it slightly for inspection purposes ; (d) it should prevent undue loss of heat through the door opening in the furnace lining. The simple lift-up door which fulfils all these conditions has been severely criticised, but, if proper care is taken to keep the furnace door jambs in good condition, the electrode consumption due to in-draught of air is not materially increased. The various designs of close-fitting swinging doors are, from a mechanical point of view, quite satisfactory, but are not suitable as inspection and working doors, since they cannot be slightly opened for spoon sampling and other manipulative operations. At least one door opening should be large enough for the re- moval of a full-diameter piece of electrode, and all others large enough for the purpose of charging uniformly and fettling. CHAPTEK XII. MODERN TYPES OF ELECTRIC STEEL FURNACES. THE several types of electric steel furnaces now in use may best be studied in the order in which they have been suc- cessively introduced. In this way the introduction of novel features peculiar to any particular design will be more readily understood. It is also necessary to divide electric arc furnaces into two distinct classes : (a) Indirect arc furnaces. (b) Direct arc furnaces. Indirect Arc Furnaces. This class includes all arc furnaces in which the arc strikes between electrodes, so that the furnace charge is entirely independent of the arc circuits and receives heat by radiation and reflection alone. This type was originated by Siemens, whose furnace is illustrated in Fig. 2. Stassano Furnace. Stassano was the first to use a single indirect arc for metallurgical purposes conducted on a com- mercial scale, and in 1898 built his first furnace, which was intended for the direct production of steel from iron ore. This furnace did not meet with economic success, so that Stassano ultimately modified its construction for melting steel scrap. The outstanding feature of the modern Stassano furnace lies in the mechanical method of mixing the molten or semi- molten charge in order to utilize the heat radiated by the arc to the best possible advantage. There are also several less im- portant features embodied in this design, which are nevertheless characteristic : (i) Fixed orientation and inclination of the electrodes. (ii) Special hydraulic electrode regulating mechanism, oper- ated by low pressure water circulating in cooling jackets, which carry the electrode holders. (234) MODERN TYPES OF ELECTRIC STEEL FURNACES 235 (iii) The melting chamber x is lined with magnesite bricks, and assumes the form of a hollow segment of a sphere, or of an ellipsoid for large furnaces (Figs. 102 and 103). A vertical section of the furnace as shown in Fig. 103 clearly illustrates these special fea- tures of the construction. The furnace body is pro- vided with trunnions, which rest on bearings carried by a ring encircling the furnace body. This ring also carries trunnions, which are supported on fixed pedestal bearings. The axes of the two sets of trunnions are set at 90 to each other, so that the furnace is free to swing in every direc- tion just like a compass mounted in a gimbal. A pivot is fixed centrally to the underside of the bottom plate, and is displaced >7 f fr3 ' s.C FIG. 103. to one side of the normal vertical axis of the furnace by a large bevel wheel, which engages the pivot through an adjustable ball and socket bearing. The bevel wheel is axial to the trun. nion ring and the normal axis of the furnace when vertical, so 236 THE ELECTRO-METALLTJKGY OF STEEL that rotation causes the pivot to describe a circle about the normal vertical axis, and imparts an oscillating movement to the furnace body ; the degree of oscillation can be easily adjusted by altering the eccentricity of the pivot. This construction is a considerable departure from the earlier types, in which the furnace body slowly rotated about a slightly inclined axis, a design which necessitated the supply of power to the electrodes through rubbing contacts. The furnace lining is composed of magnesite brick, which is surrounded by a heat insulating backing of either brick or special refractory earths, the shape of the melting chamber being de- signed to reflect the heat downwards on to the charge or bath. The lining is naturally exposed to a very intense heat, and for this reason only magnesite brick can be successfully used, and then only when certain brands are available, which not only stand up to the intense temperature, but resist " spalling " to a most marked degree. A single charging door is provided, to- gether with a small inspection hole, both of which are closed when highly reducing conditions are required. A closed tap- hole is used and any slag skimming has to be done through the charging door. Two furnaces of the above described type were installed in the north of England for the manufacture of light intricate steel castings. They were of 1 ton capacity and designed for three-phase operation, each being equipped with a 300 K.V.A. transformer supplying three-phase current at a line voltage of either 150 or 100 volts. The high voltage was used more especially for melting, and the low after fusion of the charge was complete. Such high voltage arcs are always considerably drawn out, especially in a hot furnace, and care has therefore to be exercised when charging in fresh scrap, so as not to break an arc and thus interrupt the proper electrical conditions. The method of electrode mounting used on these particular furnaces, as shown in Fig. 103, enabled the electrodes to be rapidly re- moved or adjusted in their holders. Three cylindrical water jackets, fixed to the furnace shell at an inclined angle and con- verging to a common centre at angles of 120, served as guide boxes for the electrodes. Each jacket also carried two pro- jecting guide rods upon which the base of the holder was free MODERN TYPES OF ELECTRIC STEEL FURNACES. 237 to slide, the rods being so spaced that the axis of the holder was central to the cooling jacket. The holder was indirectly con- nected to the end of a piston rod, operating in a small cylinder fastened to the underside of the cooling jacket, and was thus capable of axial movement and rapid removal. These furnaces of 1 ton capacity are reputed to have made 80 to 85 heats before requiring to be relined, and used about 1100 units for each heat. The furnace load is controlled with the aid of ammeters which indicate the current flowing through each electrode circuit, the electrodes being moved either inwards or outwards until the current flowing through each is at the desired value. If the arc between one pair of electrodes is shorter than either of the other arcs, then the current flowing through either elec- trode of that pair will be greater than the current flowing through the third. Balance of current, therefore, is only possible when the arc lengths are equal, and when the electrode tips form the apices of an equilateral triangle. The arcs them- selves are mesh connected, so that the current flowing through each arc equals the line current -f- 1*73. If A is the current flowing through each electrode circuit and V equals the line voltage, then the power can be calculated from the equation w _AxVx3x power factor 1-73 x 1000 Rennerfelt Furnace. The outstanding feature of this fur- nace lies in a special arrangement of the electrodes, whereby the arcs are forced to take the shape of a flame that is strongly deviated downwards in the form of an arrow head. The heat is in this way more concentrated on those zones where it is required for melting and refining purposes, and, at the same time, the roof and upper walls of the lining are not exposed to the same intense heat of uncontrolled indirect arcs that always have a natural tendency to flame upwards. Unlike other in- direct arc furnaces there is also a shading effect from a vertical electrode, which is to some extent comparable to that of a direct arc furnace. The arc zones can also be moved in a vertical plane, so that their distance from the charge can be kept con- stant as melting proceeds. 238 THE ELECTRO-METALLURGY OF STEEL Electrical Design. The furnace as generally constructed operates on a low tension two-phase system, the current being conveyed to the melting chamber by three circuits connected to adjustable electrodes. One circuit serves as a neutral return for the current flowing through the two phases, and is connected to a vertical electrode passing centrally through the roof. The outer terminals of each phase are connected to horizontal elec- trodes, the axes of which, together with that of the neutral electrode, lie in the same vertical plane. The arcs strike between the tips of the horizontal and vertical electrodes, and are de- flected downwards by the resultant magnetic effect of the fields set up by each arc. A complete diagram of the power supply, furnace, and in- strument connections is shown in Fig. 104. The power supply MODERN TYPES OF ELECTRIC STEEL FURNACES 239 is three-phase, as indicated by the three line wires L, which are brought into an automatic tripping oil switch 1. The high tension cables then pass to a set of choking coils 2, each of which is divided into two unequal parts in the ratio of 1 to 2 ; circuit breakers enable either portion to be short-circuited, so that three different values of choking effect may be obtained. These choking coils are frequently introduced into the low tension circuits. The high tension current is transformed down to a suitable voltage by a Scott-connected group of transformers 4, primary tappings and selector switches 6 being provided for secondary voltage variation. Auto-transformers installed on the secondary side of the power transformers have also been used for this purpose. The low tension circuits Ph. I. and Ph. II. are shown, together with the neutral return conductor Ph. I. and Ph. II. The current transformers 8 operate the various controlling instruments and the automatic regulators, if used. The usual voltages available between the horizontal outer electrodes and the vertical neutral are 80 and 100. When choking coils are used on the secondary side, the voltage across each phase is about 150, which allows for a considerable arc voltage drop on normal full load. The load is regulated by moving the two side electrodes either towards or away from the vertical neutral electrode, which is always so adjusted that the tips of all three are in line. Electrode adjustment is effected either by hand or automatic control. When the load is equally balanced between the two arcs, the current through the neutral is 1'41 times the current flowing through each phase. The side electrodes are capable of being tilted downwards, so that it is also possible to strike two entirely distinct direct arcs on to a molten charge, provided the neutral electrode is likewise in contact with the slag or dips into it. Three sets of bus bars are brought out horizontally from the transformer house at a point well above the furnace, flexible cables being then employed for connecting these bus bars to the three electrode holders. The transformer ratings for various furnace capacities are given in the following table : 240 THE ELECTRO -METALLURGY OF STEEL Furnace Capacity. Transformer Rating. 4 cwts 75 K.V.A. 7 . ... 125 15 .... 250 14 tons 400 3-3* . . . . 800 4-44 . 1000 The power factor is normally about '90 at normal full load which allows for sufficient circuit reactance to prevent very heavy fluctuations or short-circuit currents. Structural Features. The furnace body of the most modern type is built in the form of a vertical cylinder with a flat bottom, and is covered by a detachable circular roof. The furnace is mounted either on trunnions or on rockers to permit tilting, which is done by hand in the case of the smaller sizes up to about l^ tons capacity. A half section front elevation of a trunnion mounted furnace is shown in Fig. 105. The electrodes pass through cylindrical cooling jackets, which are pivoted on brackets fastened to the furnace shell. These jackets are also rigidly connected to the electrode carrying frame, which can be tilted by means of the hand wheels shown on the extreme sides of the drawing. The electrode holders are adjusted by a nut and screw feed, driven either by hand or motor as shown. The motors are either fixed under the carriers or on brackets bolted to the furnace shell. The rectangular furnace (Fig. 106) has been designed for capacities of 5 tons and over, and embodies a complete duplica- tion of the low tension furnace circuits operating in parallel. A furnace of this type of 4 to 5 tons capacity with a power input of 750 K.V.A. has been in use for two years at the works of Stridsberg and Biorck, at Trollhatten, for making high class carbon steels. Furnace Lining. Acid and basic linings are both employed, the walls being in either case 14 inches thick, which includes a 44 inch backing of fire-brick. A section through a rectangular basic lined furnace is shown in Fig. 107. The bottom is covered with two courses of fire-brick, above which is laid a single course of magnesite bricks placed on edge : the fire-brick is shown MODERN TYPES OF ELECTRIC STEEL FURNACES 241 stepped up towards the sides and carried to the top as a backing to the magnesite and silica wall bricks. The magnesite bricks are also stepped up and carried to a point a few inches above the slag line, and from there the walls are built of silica 16 242 THE ELECTEO-METALLUEGY OF STEEL bricks. The hearth is built of a mixture of calcined mag- nesite and basic slag, which can be sintered in layers by the heat of the arcs. In the modern circular form of body the walls are built up to the top of the steel shell, and form a seating for a circular domed roof, which is lined with special 9-inch silica bricks ; the sintered hearth is also about double the thickness of that shown in Fig. 107. In the case of acid lined furnaces, silica brick and ganister are used in place of magnesite brick and the sintered basic hearth mixture. Electrodes, Graphite electrodes are preferred, and are generally loaded up to 150 to 220 amps, per sq. in. Small sized electrodes have, however, been loaded as high as 400 amps, per sq. in. The vertical neutral electrode is larger in diameter than the side electrodes, owing to the heavier current carried. The electrode consumption has been carefully ascertained in terms of Ibs. consumed during each hour of operation under definite conditions. This is certainly a convenient and accurate way of expressing electrode consumption, which enables the consumption per ton of steel to be approximately estimated for intermittent or continuous operation. The following table gives actual figures of electrode consump- MODEKN TYPES OF ELECTKIC STEEL FUENACES 243 tion for furnaces operating with graphite electrodes of various diameters for the manufacture of tool steel and castings : Diameter of the Side Electrodes. Diameter of Vertical Electrode. Lb. per Hour. Period Averaged. 1 in. graphite 1 in. graphite 5 2 3 66 1224 hours 3 3 1-5 several months 3i 4 2-36 5 5 2-2 480 hours 4 5 i 1-65 130 2 sets of 5 2 sets of 6 6-4 130 The Kennerfelt furnace, apart from its use for making tool steel and castings has been employed for melting ferro-alloys and special grades of pig-iron. It has found considerable favour in the United States of America and Scandinavia, but is at present not widely known in Great Britain. Furnace Operation. The furnace may be preheated simply by means of the free burning arcs, which can be kept at any desired distance from the bottom. The usual practice is to start melting a cold charge of scrap by means of the free burning arcs, which are gradually lowered as the charge melts. When the charge has been completely melted, and it is desired to obtain strongly reducing slag conditions, the side electrodes are tilted downwards until direct arcs strike on to the slag, the neutral electrode being at the same time lowered to make contact with the bath. Under these conditions the ordinary deoxidising and desulphurising carbide slag can be maintained. Since the charge is quite independent of the arc circuits, it follows that the load is not so subject to fluctuation during the melting period as in furnaces of the direct arc type. The circuit reactance prevents heavy overloads when striking the arc, and tends to steady the current, especially when starting to heat up a cold furnace. The chief difficulty of manipulation lies in charging the scrap. Great care must obviously be taken to prevent the side electrodes arcing on to the charge when they are being tilted downwards as melting proceeds ; the same diffi- culty applies when charging scrap into the bath. Heavy small 244 THE ELECTRO-METALLURGY OF STEEL scrap will give less trouble than bulky light scrap, which requires more constant feeding. The Kennerfelt furnace has, so far, only been used for melt- ing cold scrap charges up to 5 tons in weight. Larger units are being developed, which will be more suitable for refining liquid steel. DIRECT ARC FURNACES. Heroult Furnace. In this furnace the principle of direct arc heating was first commercially applied to the metallurgy of steel. Direct arc furnaces had been used for many years before the introduction of Heroult's modified form, and were always pro- vided with a carbon-lined bottom, which was connected to one terminal of a single-phase power circuit. It was essential to eliminate this carbon- conducting bottom to prevent carbon absorption by the steel, and this Heroult accomplished by splitting the single direct arc, as hitherto used, into two direct arcs in series, using the metallic charge to complete the circuit between the two arcs. This enabled any suitable refractory material to be used for the hearth lining, and burnt dolomite was chosen for that purpose. Heroult's chief aim was to produce a furnace of simple design, in which the basic open hearth process of steel-making could be practised by merely substituting electric heating for gas. This aim was actually realised, but it was found too expensive to use the electric furnace for boiling out carbon from pig-iron for conversion to steel. For this reason its application was later confined to melting and refining mixed charges of scrap iron and steel, in which the carbon was not sufficiently high to prolong the refining operations. It was also found that internal electric heating enabled a highly reducing atmosphere, and consequently reducing slag conditions, to be maintained within the furnace, and this resulted in the discovery of further refining powers in the nature of sulphur and oxygen removal. A bottom metallic electrode, imbedded in a refractory hearth, was also tried as a substitute for the carbon bottom and, although made the subject of a Belgian patent application in 1902, was abandoned in favour of keeping the arc circuits entirely in- dependent of the furnace lining. This early decision of Heroult MODEEN TYPES OF ELECTRIC STEEL FURNACES 245 has been firmly upheld to the present day, so that the out-stand- ing feature of the original Heroult steel furnace still remains. The single-phase furnace design was not suitable for operation on polyphase systems, owing to the high cost of motor-generator sets coupled with their poor electrical efficiency. For this reason the furnace was redesigned to operate on a three-phase supply, being provided with three electrodes for striking star-connected arcs on to the metallic charge or bath, which serves as a star point. Features of the Electrical Equipment. Three single-phase transformers are connected in either delta-delta or star-delta fashion, and with this simple method of grouping, the trans- formers are identically the same, so that it is only necessary to keep one spare in case of emergency. The connections of the primary windings can also be made readily interchangeable from star to delta or vice versa, when considerable variation of the secondary voltage is desired. Flexible cables are taken direct from the secondary terminals, where the mesh connection is made, to the cable clamps attached to the furnace conductor bars, so that the least possible Length of cable is used. The primary windings are always provided with one or two tappings for effecting voltage variation across the secondary circuits, this being done by means of special switches, as already described in Chapter IV. The usual line voltages employed for basic working are 84 and 72 volts at normal full load, the open circuit voltage being somewhat higher according to the reactance of the circuits. For working the acid process, a considerably higher line voltage is necessary, owing to the high electrical resistance of the siliceous slag, and for this purpose a line voltage of 110 at normal full load is usually provided, corresponding to an arc voltage of approximately 63 volts. The transformers suitable for various furnace capacities are generally rated as follows : Furnace Capacity. Transformer Capacity in K.V.A. lOcwts. . . . 200-400 li tons- . . . 450-600 2 .- V . 600 3 '; \ . . 600-900 6-7 .; . 1200-1800 10 . ". . 1800-2400 for melting cold scrap- 246 THE ELECTRO-METALLURGY OF STEEL The power factor of the furnace load is invariably higher than the guaranteed figure of '85, which some power companies demand, and installations are working for which the average monthly K.V.A. maximum demand is calculated on a carefully, recorded average power factor of '90 to '92. The power circuits are always designed with sufficient reactance to prevent very heavy current overloads, and when reactance coils are introduced into the low tension circuits they are designed to produce only a small reactance drop at normal full load, w r hich, however, rises very rapidly on overloads. In this way the power factor is hardly affected at normal full load current. The amount of current flowing through each electrode is indicated by an ammeter, three of which are usually mounted on a panel fixed to the back framework, just above the electrode raising gear. The panel also carries three lamps, each of which is connected between one set of cables and a common point connected with the furnace hearth. When these lamps are of equal brilliancy it is an indication of balance, since the arc voltages, and therefore the current through each electrode, must be equal to produce this effect. The luminosity of each lamp, being dependent upon the voltage between each electrode and the furnace charge, is bright unless the electrode touches the charge, when the lamp is extinguished. In this connection it should be noted that one electrode can be forcibly lowered on to a charge of scrap without causing any current to flow until one of the other two completes the circuit, and for this reason a lamp which indicates contact is an exceedingly useful accessory to an ammeter, and prevents breakage of fragile graphite electrodes. With a constantly breaking load they are also most useful, as they enable the electrodes to be rapidly adjusted with less risk of causing heavy overloads on again striking arcs. An indicating wattmeter and voltmeter are usually mounted on a separate panel, which is hinged to a wall bracket and can be swung outwards into a prominent position. The wattmeter connections are made to current transformers placed in each electrode circuit, and to the three furnace conductor bars. The voltmeter is arranged to indicate by suitable plug connections the line voltage and any of the three arc voltages. MODERN TYPES OF ELECTRIC STEEL FURNACES 247 Besides the foregoing instruments, used for regulating the furnace load, a graphic recording wattmeter and an integrating watt-hour meter are generally installed, both operating off the low tension power transformer circuits. Furnace Design. The modern furnaces are three-phase and, except for the largest sizes, the same general design (Fig. 108) is adopted for all. The furnace body is octagonal to conform as far as possible to the circular form of the melting chamber ; at the same time this shape simplifies the construction of the doors and the attachment of the electrode columns and raising gear. The bottom plate is bent to a slight curve and is also more easily constructed than would be the case for a cylindrical furnace shell. The octagonal form of the shell for a given holding capacity reduces the surface of radiation to a minimum. The shell is bolted on to two rocker castings, which roll forward on a cast-iron bed plate on tilting, these rockers being rigidly braced together by two cast-iron separators and one steel casting to which the tilting screw is connected. Three door openings are provided, one at each side and one in the front wall im- mediately above the pouring spout, so that every part of the furnace hearth is readily accessible both for charging and fettling operations. The shell plates are strengthened at each door opening by a cast-steel stiffener through which the furnace doors are raised and lowered, the doors being suspended by chains from a rocking arm, pivoted on an angle support and balanced by counter- weights. The steel framework which carries the electrode carriages and gallows arms consists of three pairs of channels, each pair being set with their flanges facing so as to form a long rectangular guide-box for two pairs of rollers attached, one at each end, to the steel electrode carriage. The latter is free, then, to move up and down between these channels with only sufficient lateral movement to ensure ease of working. Two projecting lugs are cast on the back of each carriage, between which a rack, guided by means of straps, is free to slide. The racks are meshed in with pinions which are driven through reduction gearing and strongly mounted on brackets attached to the back framework. In the event of the downward movement of the electrode being 248 THE ELECTKO-METALLUEGY OF STEEL FIG. 108. 6-Ton Heroult Furnace. MODERN TYPES OF ELECTRIC STEEL FURNACES 249 resisted, the rack will no longer carry the weight of the arm, and, by moving away from the top projecting lug, opens a switch which automatically stops the motors. This device is a safe- guard against damage to the electrodes or the raising mechanism. The steel electrode carriages are cast with short arms, to which are fixed extension pieces of steel channel carrying the holders, from which they are carefully insulated. The arms of the two outer carriages are slightly set inwards, so that the centre lines of the channel extensions pass through the centres fixed for the electrode axes ; in this way the three holders can be made identical and therefore interchangeable. A space is provided between the back shell plate and the channel-guide framework, which are rigidly connected together by steel plates at each end so as to form a narrow rectangular chamber. The conductor bars pass down through this chamber, and are guided by insulated gun-metal boxes, which bridge across both the top and bottom. With this construction there is no complete iron circuit surrounding any one phase, and the heating effect of eddy currents and a reduced power factor are avoided. The electrode holder is built in two halves hinged together and water cooled, the water connection from one half to the other being made by means of a short loop of copper tube. A lug of ample dimensions is cast on to the rigid half of the holder, and is machine-faced for connection to the conductor bars. These bars are firmly held in place by insulated gun-metal brackets attached to the upper side of the electrode carriage. Water circulating pipes to and from the holders are clipped to the conductor bars, and terminate alongside the cable clamps. The roof frame is circular, and when bricked up rests upon the body lining, four lugs being riveted to the framework for purposes of bolting down to angle plates fixed to the shell. Cast-iron coolers, split so as to break magnetic circuits, rest upon the roof brickwork, and are connected to water circulation pipes which are grouped close together and terminate in flexible hose pipes. The tilting gear is of the screw feed type, which is clearly shown in Fig. 108. A heavy tilting screw is connected by a pin joint to the steel casting which separates the two rockers, 250 THE ELECTRO-METALLURGY OF STEEL and works in a heavy phosphor-bronze nut, journalled in a cast steel trunnion box. This nut is bolted to a large bevel wheel, which is rotated by a small bevel pinion driven by motor through reduction gearing. The end thrust on the nut is taken on a heavy ball race, another ball race being provided on the under side to prevent any possible axial movement in an upward direction. A telescopic dust guard covers the screw. This type of tilting gear should be cleaned, oiled, and greased at regular intervals to prevent excessive wear of the screw threads. This method of tilting is widely adopted for other types of furnaces in Great Britain. The l|-ton furnace, shown in Fig. 109, embodies the same general principles of construction as above described, but a special feature is introduced by the provision of two swivel arms for carrying the ladle. By this means an overhead casting crane in the furnace bay can be dispensed with. This arrangement is used in conjunction with a special bogie, which serves two pur- poses, according to whether ingots or castings are being made : (a) for the purpose of teeming ingots, the transfer bogie is mounted on rails, which are supported above and on either side of the ingot pit. The ingot moulds are set carefully in line with the teeming nozzle, and can be filled successively by carefully controlling the travelling movement of the bogie, which is effected by a spur wheel and pinion drive ; (b) for foundry purposes the bogie is merely used for trans- ferring the ladle from the furnace to the casting bay, where the ladle is then handled by a casting crane. The method of using this ladle carriage for the transfer of the ladle to and from the furnace is as follows : The removable bogie rails which span the ladle pit are placed in position in readiness for pouring. The ladle, which is pro- vided with extended double trunnions, is slung on the bogie and run up to the position shown in the figure. The ladle arms are swung inwards, and the furnace is slightly tilted backwards, so that the arms lift the ladle clear of the bogie trunnion bear- ings. The bogie is then moved backwards clear of the ladle pit, and the detachable pieces of bogie rails removed. The furnace can then be tilted forwards and poured, and again brought back to its original position. The rails are replaced, MODERN TYPES OF ELECTRIC STEEL FURNACES 251 , 252 THE ELECTRO-METALLURGY OF STEEL the bogie again brought up, and the previous cycle of operations reversed. The bogie then carries the ladle of steel and is free to serve either of the above purposes. Furnace Lining. The furnace is equally suited for either the acid or basic process, as the hearth is not called upon -to carry any current. The methods adopted for lining with either acid or basic material are those which are fully described in Chapter XIV. Electrodes. Both amorphous and graphite electrodes are used, the latter being the more suitable for furnaces of 2 ton capacity and under. Amorphous electrodes are, at present, almost exclusively used for the larger furnaces, the diameters varying from 14 to 20 inches. Economisers of special design are described in Chapter XV. Qirod Furnace. The original design of the Girod furnace was characterised by metallic electrodes, which penetrated the FIG. 110. hearth and electrically connected the furnace charge to one of the line conductors. Electrical Features. The furnaces are designed to operate on either single or three-phase low tension systems. The dia- grams in Fig. 110 show three methods that have bean used for supplying single-phase current to the furnace electrodes. In the first two instances the bottom electrodes were insulated from the furnace body, whereas according to the latest method they are electrically connected to the steel shell plates, to which is directly attached one set of the conductors. The small arrows indicate the direction in which the arc is deflected in each case, the deflection being due to magnetic fields set up in the steel shell by the heavy alternating current traversing the bus bars in close proximity to it. At the Gutehoffnungshutte the local destruction of the furnace walls was so considerable that it was eventually found cheaper to adopt the third and MODEEN TYPES OF ELECTBIC STEEL FUENACES 253 more symmetrical method of bus bar arrangement, which en- tailed the use of extra copper and resulted in a rather lower power factor. The actual arrangement of the conductor bars is more clearly shown in Fig. ill. The bars are brought interleaved from the generator to a point "US" underneath the furnace, whence they are split into two separate sets ; each set consists of con- ductors of opposite polarity similarly interleaved, which are FIG. 111. carried up to a point level with the rolling axis of the furnace. Here, either cables or flexible strips are used for making the short connections both to the electrode bus bars and to the furnace body. The steel shell, below the point where the flexible connections are made, is under the influence of alter- nating currents of similar magnitude and opposite phase, so that the magnetic effects are neutralised. In this way, only very slight, rotating magnetic fields are set up around the 254 THE ELECTRO-METALLURGY OF STEEL carbon electrode, which results in more uniform heating of the furnace charge and lining. The system of connections used for the three-phase furnace is exceedingly simple. The three low tension phases are star connected, the outer terminal of each phase being connected to an upper adjustable carbon electrode, while the star point is connected to a series of metallic pole pieces fixed to the bottom plate and embedded in a conductive hearth. With this arrange- ment, when the load is equally balanced between the three arc circuits, no current will flow through the bottom electrodes and the return conductor ; the phase or open circuit arc voltage is generally about 65 volts. Furnace Design. The 3-ton single-phase furnace, as used at the Gutehoffnungshiitte, is shown in section in Fig. Ill, the body in this case being square. The six bottom electrodes are electri- cally connected by means of a copper ring and plate with each other and with the furnace body. The furnace shell is mounted on rockers resting upon roller mountings. The ratio of the cross-section of the steel electrodes to the rest of the bottom area is as 1 to 16. These electrodes are 4 inches in diameter at their upper, and 6J inches at their lower extremities, which project about 8 inches below the furnace bottom. The projecting portion has a cylindrical cavity 5J inches long, through which water circulates to prevent excessive melting of the electrode at its upper exposed end. 1 A 10-ton furnace, for melting and refining cold scrap charges, was put into commission at the works of the Bethlehem Steel Company, U.S.A., in 1916. Furnaces of this capacity are designed for three-phase operation and are constructed circular in shape. The furnace shell is 5 feet in depth and 15 feet in diameter. A large, single charging door,, sliding in a water- cooled frame, is provided on one side of the furnace and im- mediately opposite the pouring spout. The furnace can be tilted either forwards or backwards, so that slag can be poured off through a notch in the charging door sill. Fourteen soft steel electrodes, about 3- inches in diameter, are electrically connected to the furnace bottom plate, the lower ends being 1 American Electro Chemical Society. MODERN TYPES OF ELECTRIC STEEL FURNACES 255 water-cooled as usual. The electrode carriers are mounted in structural columns, which are fixed on opposite sides of the furnace and are raised and lowered by a screw and nut feed. Special care is taken to insulate the electrode bus bars, and to prevent induced currents in the shell and roof frame. The furnace is supplied with power from a group of three single-phase trans- formers having a total capacity of 700 K.V.A., each of which is protected by a reactance coil of 106 K.V.A. capacity. Furnace Lining. The single-phase furnace referred to above was originally lined with magnesite, which was later given up in favour of dolomite for both the hearth and wall construction. When the furnace was used for liquid refining, the hearth and walls would generally last about 120 heats, the hearth, originally 18 inches thick, dropping about two inches during this period. The roof was lined with silica bricks springing from magnesite skewbacks to admit of its easy detachment from the dolomite walls, which was generally found necessary after 60 or 70 heats. The heat loss, due to water cooling the bottom electrodes, was carefully determined by measuring the quantity of water flowing and its temperature before and after passage through them. The loss was equivalent to an energy consumption of only 2*9 K.W. hours per ton of steel, and is small compared to the loss of 10*5 K.W. hours per ton of steel measured at the electrode cooling jacket. The amount of water required for cooling the bottom electrodes was only "20 cubic meters per ton of steel as compared with "65 cubic meters required for the top electrode cooling ring. With the proper degree of water-cooling the steel electrodes should only melt to a depth of about one or two inches below the hearth level, so that no serious erosion of the dolomite results. In the larger three-phase basic furnace the wall lining is built of magnesite brick up to the roof, from which it is separated by asbestos plates ; in other respects the lining does not differ from standard practice. Electro -Metals Furnace. The original feature of the Electro- Metals furnace was the application of a three-wire two-phase system of low tension connections. This design enables either two or three-phase high tension current supplies to be used without the aid of motor generators, which were formerly 256 THE ELECTEO-METALLUEGY OF STEEL necessary for single-phase furnace operation. A conductive hearth is still an essential characteristic of this furnace. Electrical Features. The method by which two-phase current is supplied to this furnace has been fully described in Chapter III., and does not require further explanation. A diagram of the entire electrical equipment is shown in Fig. 112. Here, the two-phase low tension current is transformed down from a three-phase high tension supply by Scott-connected transformers. Tappings are taken out from the primary windings to give either 90, 80, or 70 volts across each of the low tension phases on open circuit, and the various connections can be made by two selector switches A and B, which are interlocked with one another, and with the main oil switch O.S.A. In the 7^-ton furnace rather higher open circuit voltages are used, namely, 100, 85, and 75. The neutral conductor cables are con- nected to copper bars placed side by side on a course of bricks laid on the bottom plate ; details of this method of conveying current to the hearth are given in Chapter XIV. (Fig. 126). A furnace operating on a four-phase low tension system has been recently designed for large capacities. The special method of transformer grouping adopted for supplying such four-phase low tension current from a three-phase supply has also been dealt with in Chapter III. The large power inputs required for furnaces exceeding 10 tons capacity cannot be satisfactorily carried by only two elec- trodes, to which number the two-phase pattern is limited, and it is chiefly for this reason that four-phase current requiring four upper electrodes is employed. This system requires five separate sets of conductors from the transformer group, four being con- nected to the upper electrode bus bars, and the other to copper bars imbedded in the furnace hearth in a manner similar to the two-phase furnace. The hearth is only called upon to carry rather more than the current flowing through any one elec- trode when all are equally balanced, and the current density is one-third of that of the two-phase three wire type. The load is controlled and balanced by ammeters, which register the current in each arc circuit. Voltage variation is also provided for by tappings taken from the high tension transformer windings in the usual way. p ; . . . . 232 regulating gears . . .. . '-.-.- . . 230,231 Furnace doors . ' -. . . 233 mountings * ' . .. .; . '''. . . . 229,230 Tilting gears . . ... . . . . . . 230 Physical conditions and their influence upon Heat distribution ..... .. ... ... . 217 loss by radiation . . . . . . . . . 218, 219 Intensity of arc heating .... .. . . . . . 216,217 Temperature variation . . ... . . . . 217 Direct arc furnaces- See Booth-Hall, Dixon, Electro-metals, Giffre, Girod, Greaves- Etchells, Heroult, Keller, Ludlum, Siemens, and Stobie furnaces. Dixon furnace Four-phase type . ... . ... . . . .70,71 Patents . . ......' . . . 40, 58, 70, 71 Two-phase type ..... . . . 58 Dolomite . . ..... . . . . . . . 283 Dozzles . . . . . . . . 194 Duplex processes Girod's patents 25 Methods of applying . . . . . . . . 163-164 EFFECTIVE current, definition of 44 voltage, 44 Electrical degree, definition of 43 Electrodes Amorphous, chemical and physical characteristics of . . . 312-315 , comparison with graphite electrodes . . . . 321 , current carrying capacity of 320 .defects of 310-312 , economises for 324-329 , joining of 317-319 , manufacture of 308-310 , materials used in manufacture of .... 306-307 , storage of 320 Graphite, artificial 307, 308, 316-319 , comparison with amorphous electrodes .... 321 , current carrying capacity of 320 , economisers for 324-329 346 THE ELECTRO-METALLURGY OF STEEL Electrodes (cont.) PAGES Graphite, joining of 319 , manufacture of 316 , natural 307 Holders for, design of 232 Regulating gear for . . 230, 231 Electro-metals furnace Constructional features of 257 Electrical features of 38, 55, 71-73 Electrodes used in .......".... 259 Four-phase type 71-73 Linings of 258 Operation and manipulation of 259 Patents 38 FERRANTI furnace 3, 4 Ferro-alloy additions to liquid steel In acid process, calculation of 156 In basic 139-147 Fettling, method of 148, 149 Fire-clay 290, 291 Flat rate system, purchase of power on 103 Fluctuation of load, effect on load factor Ill, 112 Fluxes, used in basic process 122, 133, 134 Foundry practice Application of acid process to 198-200 basic 199 Electric melting in, advantages and disadvantages of . . . 197-202 Furnace capacity, choice of 204, 205 Lay-out of furnace plant 203, 204 Statistics of castings production 200, 202, 203 Four-phase circuits, electro-metals, grouping of 71 , mesh connection of 71 , star connection of 70 current, application of 69-73 furnaces, Dixon types 70-71 , Electro-metals type 71-73 Frequency, definition of 42 Frick furnace 13, 14 Furnace capacity, in relation to foundry practice . . . . 204, 205 , radiation loss . . . . . 218, 219 Furnace linings- Acid 303 Baking in basic hearths 299-303 Basic, with bottom electrodes 225, 226, 296-299 - , without bottom electrodes 292-296 Drying-out 299-303 Life of 305 Repairs to 304, 305 Roof construction 299 GANISTEB 289, 290 Giffre furnace, single-phase type 52 , three-phase type 67 Girod furnace Bethlehem Steel Co., U.S.A., description of 254 Bottom electrode losses in 255 Constructional features of 254, 255 Early types of 33-35 Electrical features of 56, 67, 252-254 Linings of 255 Patents for . 33-36 INDEX 347 Girod furnace (cont.) PAGES Single-phase type 252-253 Two-phase type 56 Three phase type 67, 254 Greaves-Etchells furnace Constructional features of 273 Electrical features of 68, 270-272 Hearth construction 274 resistance 273 Lining of 274 Patents for 41 HEAT Distribution of, in arc furnaces 217 Radiation, loss of 218, 219 , in relation to furnace design 218 , thermal efficiency . . . 108-109 Heroult furnace Construction of modern types of 247, 249, 250, 252 Development of 244 Early types and history of 20, 21 Electrical features of modern types .66, 245-247 Electrodes used in 252 Linings of . . . . 252 Patents for 21-25 Power factor of 246 Single-phase type 20, 25, 55 Two-phase type 55 Hiorth furnace 15, 16 INDIRECT arc furnaces Bassanese furnace 40 Rennerfelt furnace 39, 237-244 Siemens furnace 2 Snyder furnace 39, 263-270 Stassano, ore smelting furnace 17-19 , steel melting furnace 19, 234-237 Induction and induced currents, definition of 45 Induction furnaces Colby furnace 4-6 Ferranti furnace , 3-4 Frick furnace 13-14 Hiorth furnace 15, 16 Kjellin furnace 6-9 Rochling-Rodenhauser furnace . . 9-13 Ingots Bottom casting of 190-194 Defects of 171-181 Blowholes 180 Clinks 178 Contraction cavities . . . . . . . . . 178 Folding and lapping . . . . . " . . . 175,176 Occluded gases . . . . . . . . . 180, 181 Pipes 171-173 Pulls . . . . . . ... . . . 177 Segregation .......'... 173-175 Shell .--... . 176,177 Surface cracks . . . . . . . *.- 178-180 pitting . . . ., . . - .-. . . . 177 Theory of formation . . ... . . . . 165-171 Top-casting of . . . '.- . i . . ... 188-190 Tun-dish casting of . . . . . * . . . 195,196 348 THE ELECTBO-METALLUBGY OF STEEL. PAGES Ingot moulds, grey iron for casting ......... 186 , influence of shape on ingot formation .... 186-190 P^ 188 KELLER furnace Conductive hearth type 31, 32 Early single-phase type 26 Patents for 28-31 Single phase, four electrode 26-28 Three-phase 67 Kjellin furnace 6-9 LADLES, bottom teeming Heating appliances for 185, 186 Use and manipulation of 181-185 Lag, definition of 45 Lead, definition of 45 Linings, see " Furnace linings ". Liquid refining, Heroult's patents for . ' 25 See also " Acid process of " and " Basic process of ". Load factor Definition of 51 Effect of fluctuating current upon Ill, 112 , on cost of power 105-107 , on energy consumption 110, 111 , on output 110, 111 Maximum demand, in relation to 107 Ludlum furnace . . . 280 MAGNESITE 284 Manganese, determination of, in steel 334, 335 , removal of, by oxidising slag 127 Maximum demand Contract for power based upon 104, 105 Definition of 105 Eelation of, to cost of power 108 , to load factor 107 Meter rate system, purchase of power on 103 NICKEL, determination of, in steel . . . . . . 336-338 PERIOD, definition of 43 Phase, definition of 43 Phosphorus, determination of, in steel 333-334 , removal of, in basic process 126-127 Power Consumption of (energy), for melting and refining .... 112 , influence of load factor upon . . . 112 , power input upon . . . 109 Contracts for, basis of 102, 103 , on " flat rate " system 103 , on " maximum demand " system .... 104,105 , on " meter-rate " system 103 Cost of, effect of load factor upon 105-107 , maximum demand upon 108 Power factor Definition of . . - 48 Heroult furnace, power factor of 228, 229 Relation of, to furnace design 84, 85 , to reactance voltage 266 Snyder furnace, power factor of 246 INDEX 349 RADIATION, heat loss by PAGES Relation of, to furnace design 218, 219 , to load input 108, 109 , to useful energy input 109 Reactance Coils 83-85 , voltage variation by means of 83 Definition of 47 Drop, definition of 48 Internal, of transformers 77 Voltage 48, 84 Refractory materials Acid, analysis of 290 Alundum 287 Bauxite 286 Chromite 287 Classification of 281 Conditions of service of, in electric furnaces .... 281-283 Dolomite 283 Fire-clay 290-291 Ganister . ... 289, 290 Magnesite ...... 284 Silica . ., . 287-290 Regulation Automatic, applied to polyphase arc furnaces 88, 89 , single -phase constant pressure circuits . . 88 , direct arc furnaces . . 22, 86-88 , indirect arc furnaces . ; . 86 , principles of . ..'. . " . ,' . * . . . 86-89 Regulators Combined current and voltage, General Electric Co., U.S.A., type ..- 95 , Watford type ..... 95 Thury, description of.. . . . . . . . .90 , diagram of connections . . . ...... ~ 94 Voltage, Watford type ..." .- . . ... . .93,94 Rennerfelt furnace Constructional features of . . ." . 240 Electrical features of . .... . . . ... 57,237-240 Electrodes used in . . . . ... . . . 241,242 , consumption of 243 Linings of . 240, 241 Operation of . . . . 243 Patents for . . . . . , . .39 Rochling-Rodenhauser furnace ...... , . . . . 9-13 Roofs of furnaces 299 SAMPLES, methods of taking . . . . . . - 330, 338, 339 , preparation of . . ... . . < . 331 Scrap- Choice of, for acid process . . . . . .. . 152 , for basic process . . . . . . . 117-123 , method of charging . . . . - . . 120-122, 152 Self-induction, definition of . . . . . ... . 47 Siemens' direct arc furnace ..'._. 2 indirect arc furnace . . . . . . . Silica bricks, cement and sand . . f . . . . . * 287-290 Single phase- Current, applications of . ..-. . . . . .. 52 , definition of. ..'. 50 generators . . . . . ... . . 75 , Miles- Walker type of . ... . . . . 76 , supply of t . M . . 74 350 THE ELECTEO-METALLUEGY OF STEEL Single-phase (cont.) PAGES Furnaces (arc type) See Bassanese, Girod, Heroult, Keller, Siemens, Snyder, and Stassano furnaces. Furnaces (induction type) See Colby, Ferranti, Frick, Hiorth, Kjellin, and Rochling- Rodenhauser furnaces. Transformers (static) 77 Sink-head 195 Slag- Acid, analysis of . 156 , characteristics of 155 , formation of 153, 154 , function of 153, 154 Basic oxidising, character of 123 , composition of 123 , fluxes for 122 , formation of 124 , function of 122 , reactions of 125 , removal of carbon by 125, 126 , manganese and silicon by . . . . 127 , phosphorus by 126, 127 , sulphur by 127 reducing, analysis of 135 , character of . . 135, 136 , deoxidising powers of 136-138 , fluxes used for 133, 134 , formation of 136 , function of 132, 133 , reactions of 138, 139 Slag inclusions 211, 212 Snyder furnace Constructional features of 267 269 Electrical features of . 263-267 Lining of 269 Load input, current and voltage curves of 263-265 Operation and manipulation of 269, 270 Patents for 39, 40 Power factor of 266 Specifications, for castings 205, 206 Stassano furnace Ore smelting types 17-19 Steel melting types, constructional features of .... 234, 235 , electrical features of 66, 236, 237 , patents for 19 Stobie furnace Constructional features of 260-262 Electrical features of 57, 68, 260 Electrode economisers of 262 Electrodes used in 262, 263 Lining of 262 Patents for 38 Sulphur Removal from steel, by Heroult's process 22, 24, 25 , by oxidising basic slags 127 , by reducing .... 138, 139 Switch gear, change voltage ......... 81 - , high tension 99-101 TEMPERATURE, control of 130, 140 , methods of testing bath 140 INDEX 351 Three-phase PAGES Circuits, " delta " or " mesh " connection of 62-64 , four wire star connection of 67, 68 , inverted star connection of 65 , nomenclature of . 62 , three-wire star connection of 60 Current, application* of 59-69 , definition of . . . . - 50 Furnaces (arc type) See Giffre, Girod, Greaves-Etchells, Heroult, Keller, Ludlura, and Stobie furnaces. Furnaces (induction type) See Rochling-Rodenhauser furnace. Tilting gears 230 Tools, furnace 148 Transformers (static) Capacity of, in relation to furnace capacity . . . . . , . 224 , load factor . . . . ,. . -'.. 224 Current ' ... . . . . . . . 97 Grouping of, three-phase to three-phase . . ... . 79 , to two-phase . . . . ... 80 , two -phase to three-phase . . . . . . 80 Heating of . ; . . . . .... . 77 Internal reactance of . - .. . . . . .. ... . ... 77 Primary tappings of . .... . . . . . . 78 Secondary voltage, variation of . . . . ' . . . . . 80 Single-phase . . ... . . . . . . . 77 Trumpet bricks . . ... . . . . . . 192 Tun-dish . , . . .... . 196 Tungsten, determination of, in steel . . . . ... . 338 Two-phase Current, application of . . . . . . . . 53-59 , definition of . . . . .. . . . . . 50 Furnaces See Booth-Hall, Dixon, Electro-metals, Girod, Heroult, Eennerfelt, and Stobie furnaces. VOLTAGE, circuit or line 44 , effective ' . 62,84 , reactive . .- . 48, 84 , resistance 84 Volt-ampere and K.V.A., definition of . . . ' 48 WATT- and K.W., definition of . . . ; . . . . . 49 Wattless current . . . . . . . . . . 48 Watt-hour meter, use of . . . -. * . . . . . . 99 Wattmeter- Graphic recording, use of . . . . . - . . . . 98 Indicating, use of 98 Waveform 42 ABERDEEN: THE UNIVERSITY PRESS D.VAN NOSTRAND COMPANY are prepared to supply, either from their complete stock or at short notice, Any Technical or Scientific Book In addition to publishing a very large and varied number of SCIENTIFIC AND ENGINEERING BOOKS, D. Van Nostrand Company have on hand the largest assortment in the United States of such books issued by American and foreign publishers. All inquiries are cheerfully and care- fully answered and complete catalogs sent free on request. 8 WARREN STREET - NEW YORK UNIVERSITY OF CALIFORNIA LIBRARY, BERKELEY THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW Books not. returned on time are subject to a fine of 50c per volume after the third day overdue, increasing to $1.00 per volume after the sixth day. Books not in demand may be renewed if application is made before expiration of loan period. '*!* MAR -3 23 K lOm-4,'23 YC 18731 500142 UNIVERSITY OF CALIFORNIA LIBRARY