Me ctiani c s De ID t . The Penton Publishing Co. Cleveland, O., U. S. A. Publishers of The Foundry The Iron Trade Review Marine Review Daily Metal Trade Power Boating Abrasive Industry AMERICAN MALLEABLE CAST IRON By H. A. Schwartz First Edition Published by The Penton Publishing Co. Cleveland, Ohio 1922 Library Copyright in the United States and Entered at Stationers' Hall, London 1922 The Penton Publishing Co. Cleveland, Ohio THE PENTON PRESS CO., CLEVELAND THE MEMORY OF ALLEN SMITH BIXBY Wi:OSE INSTRUCTION, CO-OPERATION AND ADVICE I OWE THE EARLY OPPORTUNITIES WHICH MADE THIS BOOK POSSIBLE IT IS AFFECTIONATELY DEDICATED 588565 PREFACE THE literature of malleable cast iron, in the American sense of that term, is limited to a single book first issued about 10 years ago and now out of print, and to a series of articles of great diversity of character and quality in the technical publi- cations of this country and Europe. Much of the most valuable scientific matter is buried in the purely scientific press, frequently under titles which do not suggest its application to any one not a specialist in metallurgy. Under these circumstances the preparation of a new book dealing with American malleable cast iron in theory and practice may serve a useful purpose as summarizing and rec- ording, so far as any book can, the contemporary state of the art in the metallurgy of this product. If in the following pages the specialist finds much which ap- pears to him elemental or trivial or the non technical reader finds matter which appears too complex, the author must plead in ex- tenuation his desire to prepare a book to suit many kinds of readers. This has necessitated the inclusion of much elementary matter both in metallurgy and mechanics which will be useful only in acquainting -the lay reader with the interpretation of terms and data which form the every day vocabulary of the technician. On the other hand it has seemed that in order that the reader might secure full value from a reading of these pages no known fact or theory should be excluded merely in the interest of simplicity. Feeling that no single individual is justified in the belief that his own views are final in so complex a subject the author has not hesitated to refer freely to the literature and even to record opinions contrary to his own. So far as possible due credit has been given in all such cases. Guided by the injunction of Leonardo da Vinci, "Con- firm your statements by examples and not by assertions", it has VII been the author's constant effort to- record facts rather than opinions wherever possible. This has been particularly true in the chapters dealing with manufacturing methods. In these chapters the record is one of what has been rather than of what might be accomplished. Much of the experimental work re- ferred to is the work of the author's associates. In this connec- tion special recognition must be given to the very unusual micro- graphs which are the work of Harrie R. Payne, chief chemist and metallographer of the author's laboratory. Many of the author's friends within the malleable industry, in the organization of which he has the honor to be a member, and among the business connections of that association have contributed valuable information. Whenever possible credit has been assigned. In some cases where for obvious reasons it was- improper to identify the in- formation the latter has consented to the anonymous presenta- tion of his material. The co-operation of the American Malle- able Castings Association in furnishing statistics and of the late Thos. Devlin of Philadelphia, and Alfred E. Hammer, Branford, Conn., in contributing historic matter from their long experience is especially worthy of grateful acknowledgment. If the following pages contain any information calculated to dispel the many misconceptions as to malleable cast iron and to acquaint the interested reader accurately with the proper- ties and methods of manufacture of this interesting, valuable and characteristically American product, the author's labor will have been richly repaid. H. A. SCHWARTZ VIII CONTENTS CHAPTER I Early History of Ironmaking 1 CHAPTER II Development of Malleable Industry in the United States 15 CHAPTER III Metallurgy of Malleable Iron 41 CHAPTER IV General Manufacturing and Plant 71 CHAPTER V Melting Stock 91 CHAPTER VI Fuel and Refractories 109 CHAPTER VII Air Furnace Melting 135 CHAPTER VIII Electric Furnace Melting 159 CHAPTER IX Cupola and Open-hearth Melting 175 CHAPTER X Annealing Practice 189 CHAPTER XI Principles of Annealing 213 CHAPTER XII Molding and Patternmaking .* 233 CHAPTER XIII Cleaning and Finishing 249 CHAPTER XIV Inspecting and Testing " 267 CHAPTER XV Tensile Properties 287 CHAPTER XVI Compression, Bending and Shear 303 CHAPTER. XVII Fatigue, Impact, Hardness and Wear 315 CHAPTER XVIII Plastic Deformation 339 CHAPTER XIX Thermal and Electrical Properties 371 Selected Bibliography 385 Index 403 IX LIST OF ILLUSTRATIONS PAGE Fig. 1 A meteorite - in the American Museum of Natural History, New York, brought from Greenland by Admiral Peary 2 Fig. 2 A primitive furnace, 1500 B. C. The illustration was re- produced from an Egyptian wall painting Fig. 3 One of the earliest blast furnaces 4 Fig. 4 An early American blast furnace 7 Fig. 5 Reaumur's foundry in 1724. One furnace has just been emptied and the blast is being applied to the other Fig. 6 Statue of Seth Boyden, erected in the city park of Newark, N. J M by citizens in memory of the man who laid the founda- tion for the malleable industry in the United States 12 Fig. 7 Seth Boyden 16 Fig. 8 J. H. Barlow, Boyden's successor 16 Fig. 9 Distribution of malleable iron foundries in the United States. The dots represent the location of malleable foundries according to data compiled for government use : . . . . 20 Fig. 10 Map showing location of principle sellers of malleable iron castings in the United States 22 Fig. 11 Comparison of production of steel and malleable iron castings 24 Fig. 12 Familiar figures in the development of the malleable in- dustry in the United States 28 Fig. 13 Three metallurgists who have been closely identified with the technical advancement of the industry 32 Fig. 14 The names of these men are linked with the rise of the American malleable industry 34 Fig. 15 Austenite and ledeburite in manganiferous white cast iron 42 Fig. 16 Martensite in quenched white cast iron 42 Fig. 17 Troostite in steel 43 Fig. 18 Pearlite in incompletely annealed malleable 43 Fig. 19 Spheroidized pearlite 44 Fig. 20 Graphite in gray iron 44 Fig. 21 Soft gray cast iron 45 Fig. 22 Malleable cast iron 45 Fig. 23 Benedict's diagram recording the equilibrium conditions in terms of temperature and agraphic (non graphitic) carbon. It is based on Benedict's principle, somewhat modified 47 Fig. 24 A further modification of Benedict's diagram indicating the results of recent research work 51 Fig. 25 Graphite crystals in malleable made from hard iron con- taining graphite 54 Fig. 26 Unannealed hard iron. The structure is always dendritic but varies slightly with the carbon content 55 XI LIST OF ILLUSTRATIONS Continued PAGE Fig. 27 Effect of silicon in relation to carbon on malleable. This graph is based on data from Thrasher's determinations 56 Fig. 28 Beginning of graphitization after one half hour at 1700 degrees Fahr 58 Fig. 29 Progress of graphitization after \ l / 2 hours at 1700 degrees Fahr 58 Fig. 30 Progress of graphitization after 3 l / 2 hours at 1700 de- grees Fahr 60 Fig. 31 Equilibrium at 1700 degrees Fahr. after 70 hours 60 Fig. 32 Imperfect attainment of equilibrium below A due to too short a time 62 Fig. 33 Normal malleable iron, metastable equilibrium below A t . . 62 Fig. 34 Graphite crystals produced by annealing at 2100 degrees Fahr , 64 Fig. 35 Manganese sulphide in a malleable cast iron. The arrows point to MnS 64 Fig. 36 Chart showing conversion of combined carbon into temper carbon 66 Fig. 37 Changes of metallographic composition during the freezing and annealing of white iron 69 Fig. 38 Organization chart for malleable foundry 72 Fig. 39 A good example of the approved style of architecture for a malleable foundry built a generation ago 74 Fig. 40 Exterior view of a large malleable plant built about 1917 76 Fig. 41 Coreroom of a modern malleable plant showing roof con- struction designed to facilitate removal of fumes and gases and to afford good natural lighting 78 Fig. 42 Interior of the annealing department of a modern malleable foundry 80 Fig. 43 Chart showing cycle of principal operations in a malle- able plant 83 Fig. 44 Chart showing division of labor in a typical foundry .... 84 Fig. 45 Molding floor in a well organized American malleable foundry 86 Fig. 45 The stock yard usually is served by a traveling crane 92 Fig. 47 Map showing location of principal ore fields, and coke and charcoal blast furnace producing malleable pig iron 95 Fig. 48 An open pit iron ore mine on the Mesabi range. Ores in this district are suitable for making malleable pig 98 Fig. 49 An ore loading dock at one of the ports on Lake Superior 100 XII LIST OF ILLUSTRATIONS Continued PAGE Fig. 50 An ore unloading dock at a Lake Erie port, where the ore is transferred from ore carrier to railroad car 102 Fig. 51 A charcoal blast furnace in Michigan where malleable pig iron is made 104 Fig. 52 A typical coke blast furnace in the Mahoning valley 106 Fig. 53 Map showing location of principal resources of metallurgi- cal fuel in the United States .' 110 Fig. 54 A modern coal tipple in West Virginia 112 Fig. 55 Picking table in a coal tipple showing facilities for remov- ing slate, sulphur, etc., by hand 114 Fig. 56 Adjustable loading boom which places coal in car without breakage 114 Fig. 57 A modern by-product coke plant which is engaged in making foundry fuel 116 Fig. 58 A typical scene at a beehive coke oven plant in- the Con- nellsville region 118 Fig. 59 Cross section of a modern gas producer 120 Fig. 60 A scene in an important oil field in Oklahoma 122 Fig. 61 Operations in a molding sand pit 125 Fig. 62 Hauling sand from a pit 125 Fig. 63 Map showing the principal sources of molding sand, fire- clay and brick in the United States 126 Fig. 64 Open fireclay pit covering over 10 acres and with bed of clay from 25 to 40 feet thick '. 128 Fig. 65 A plant in Missouri showing round, down-draft kilns, factory and stock sheds 130 Fig. 66 A repress room in a Missouri firebrick plant, showing machines in which stiff mud firebrick are made 132 Fig. 67 Firebrick and special fireclay shapes in kiln ready to be burned 133 Fig. 68 Sectional drawings showing construction of typical air furnace 136 Fig. 69 Graph showing recombination of carbon in pig iron,... 139 Fig. 70 The roof of the modern air furnace is almost straight.. 140 Fig. 71 A waste heat boiler connected to two air furnaces. Note that coal for auxiliary firing is on hand 142 Fig. 72 Gray sprue 148 Fig. 73 Gray sprue showing white patches. Characteristic of less but still excessive carbon and silicon 148 Fig. 74 Moderately mottled sprue characteristic of carbon, silicon and temperature suited to very small work 148 XIII LIST OF ILLUSTRATIONS Continued PAGE Fig. 75 Normal sprue for metal of the higher carbon ranges of specification metal in average work 149 Fig. 76 Similar to Fig. 74 but lower in carbon 149 Fig. 77 Similar to Fig. 76 but quite low carbon 149 Fig. 78 ''High" iron i.e. metal low in carbon, silicon and man- ganese. Fracture granular throughout and edge showing blow- holes 149 Fig. 79 Changes of metal after tapping 151 Fig. 80 A powdered coal atatchment for an air furnace 155 Fig. 81 Cupola producing molten iron the starting point of the Kranz triplex process 160 Fig. 82 Two-ton side-blow converter producing liquid steel from cupola metal in triplex process 162 Fig. 83. Transfer train consisting of electric motor car and trailer with' crane ladle. This equipment is used in carrying cupola and converter metal to the electric furnaces 165 Fig. 84 Heroult electric furnace in which cupola and converter metal is charged for final step in triplex process 168 Fig. 85 Heroult furnace tilted for pouring 170 Fig. 86 Pouring side of open-hearth furnace for malleable iron.. 176 Fig. 87 Charging side of open-hearth furnace in malleable plant . . 178 Fig. 88 Design of a modern, stationary open-hearth steel furnace. 180-181 Fig. 89 Separator plate designed to eliminate use of packing with annealing pots 190 Fig. 90 A view of the annealing department in a modern malleable castings plant 194 Fig. 91 Charging trucks facilitate the handling of pots to and from the annealing furnaces 195 Fig. 92 The interior of the powdered coal mill of a modern malle- able plant 197 Fig. 93 While most of the plants in the United States employ annealing furnaces similar to those shown in Fig. 90, a few plants use the pit type, illustrated above 198 Fig. 94 Diagram showing the distribution of heat in a continuous- type annealing furnace 200 Fig. 95 Interior of continuous-type annealing furnace looking toward the entrance end 200 Fig. 96 Single section of combustion chamber of continuous type annealing furnace 201 Fig. 97 A sectional plan and elevation of a double-chamber, car- type tunnel kiln for annealing malleable iron castings. The fir- XIV LIST OF ILLUSTRATIONS Continued PAGE ing zones are diagonally opposite each other 202 Fig. 98 Rim of a casting containing most of the usual defects due to annealing. Etched with picric acid, magnified 100 diam- ' eters and subsequently reduced one-fourth on erfgraving ....216,217 Fig. 99 Increase in carbon content at increasing depths below' the surface of malleable cast iron .........'... 223 Fig. 100 Graph showing effect of removing one-sixteenth inch, decarborized surface in specimens of various diameters on the tensile properties of the metal 225 Fig. 101 Graph showing effect of varying degrees of decarburi- zation on tensile properties of malleable cast iron 226 Fig. 102 Equilibrium curves illustrating the reactions between car- bon, iron and oxygen, after the data of Matsubara 230 Fig. 103 Methods of mounting patterns 234 Fig. 104 Squeezer-type molding machine and mold and pattern equipment in place 236 Fig. 105 Stripper and rollover -type molding machines 238 Fig. 106 Curve showing contraction in cooling from solidification to room temperature 238 Fig. 107 Graph showing the per cent of contraction of malleable from pattern size 240 Fig. 108 Graphs showing relation of annealing upon the density of the metal 241 Fig. 109 Casting with thin disk and thick hub, showing probable point of rupture 242 Fig. 1-10 Type of casting with thin disk center and thick rim.. 242 Fig. Ill Dendrite (about half size) from shrink in hard iron ingot 8 inches in . diameter by 20 inches high which was poured without feeding 244 Fig. 112 Typical gate for malleable castings showing strainer, core and skimmer gates for furnishing clean metal for feeders and producing sound castings 246 Fig. 113 Tumbling barrels are used for cleaning castings 250 Fig. 114 Sand blast equipment is used for removing sand from castings 252 Fig. 115 Sorting and inspecting small castings are important opera- tions in many plants 252 Fig. 116 When machine center and casting center are not concen- tric apparent hard spots may be found .-...- 255 Fig. 117 (left) Cementite persisting near a shrink. The metal in porous areas is somewhat oxidized 257 XV LIST OF ILLUSTRATIONS Continued PAGE Fig. 118 (Right) Hard slag inclusions just below the surface which may dull cutting tools rapidly 257 Fig. 119 Malleable casting effectively arc welded with Swedish iron. The changes visible microscopically were insufficient to make notable difference in metal. Area A is soft iron but very slightly recarburized from the malleable; B is an oxide or slag film, and C is the malleable showing but little resolution of carbon due to close confinement 258, 259 Fig. 120 Hard iron casting successfully acetylene welded with hard iron and then annealed. Note metallurgical homogeneity of casting except for presence of slag. A is the original casting, B the slag, C the material of weld as noted by larger grain size, and D the material of weld as noted by persistence of a little pearlite due to decarburization 258, 259 Fig. 121 Ineffective hard weld of malleable casting using ingot iron wire and acetylene method. Neither material has its original structure. A is the soft iron filler converted into hard iron by migration of carbon from the malleable. B is the original malleable iron, the background of which has become sorbitic due to recombination of carbon at temperature the metal reached in welding 258, 259 Fig. 122 Photomicrograph showing heavy pearlitic rim which may cause machining difficulties 262 Fig. 123 (Left) An effective acetylene weld, malleable becoming sorbitic due to resolution of carbon. A is gray iron converted into white cast iron by remelting. B is malleable 263 Fig. 124 (Right) Tobin bronze weld in malleable. Note absence of oxides and slag in weld and absence of recombination of carbon due to relatively low melting point of breeze. A is . bronze, B is malleable 263 Fig. 125 Analytical laboratory in malleable plant 268 Fig. 126 Apparatus for determining carbon 269 Fig. 127 Inverted types of metallographic microscope 272 Fig. 128 Detail of inverted type of metallographic microscope (Bausch & Lomb) 272 Fig. 129 A. S. T. M. tension test specimen 274 Fig. 130 Dimensions of proposed tension test bar 276 Fig. 131 A 200,000-pound Olsen universal testing machine 278 Fig. 132 Ewing-type extensometer for determining elongation under load ...".... 279 Fig. 133 Olsen-type torsion testing machine 280 Fig. 134 Leeds & Northrup Co. apparatus for determining critical XVI LIST OF ILLUSTRATIONS Continued PAGE points by Roberts-Austin method 281 Fig. 135 Apparatus for measuring magnetic properties of metal.. 281 Fig. 136 Farmer fatigue testing machine 282 Fig. 137 Charpy hammer for impact tests 283 Fig. 138 Brinell hardness tester 284 Fig. 139 Stress-strain diagram of malleable cast iron in tension 289 Fig. 140 Tensile strength and elongation plotted from specimens submitted by members of American Malleable Castings As- sociation 291 Fig. 141 Effect of carbon on tensile properties of malleable iron.. 293 Fig. 142 Relation between tensile strength and elongation of malle- able cast iron 295 Fig. 143 Comparison of tensile properties of machined and cast specimens of equal diameters 296 Fig. 144 Results of tests on specimens not machined 298 Fig. 145 V groove in bar 300 Fig. 146 Necked specimens of steel (left) and malleable (right) 301 Fig. 147 Stress strain diagram of malleable cast iron in com- pression 304 Fig. 148 Malleable (center) and cast iron (right) in compression each specimen before testing was of the size and shape shown at the left 305 Fig. 149 Diagram of stresses in cross bending of malleable iron 308 Fig. 150 Displacement of planes by linear shear and (at right) by torsional shear 310 Fig. 151 Stress strain diagram of malleable cast iron in torsion.... 311 Fig. 152 Diagram showing factors to be considered in deter- mining torsion stresses 312 Fig. 153 Effect of elongation of specimen on the resistance to dynamic tensile loads 319 Fig. 154 Walker test wedges 321 Fig. 155 Behavior of malleable iron under fatigue as a rotating beam 322 Fig. 156 Separation of grains by repeated cross bendings 323 Fig. 157 Relation between Brinell number and strength of malleable iron specimens 325 Fig. 158 Graph showing comparison of Brinell and Shore numbers indicating relation between them is not definite 326 Fig. 159 Tests of machining properties of malleable cast iron 330 Fig. 160 Graph showing values of a in drilling formula 332 XVII LIST OF ILLUSTRATIONS Continued PAGE Fig. 161 Graph showing values of'fc in drilling formula 333 Fig. 162 Relation of torque and thrust to ultimate strength 334 Fig. 163 Relation of torque and thrust to Brinell number 335 Fig. 164 Slip bands in ferrite of malleable iron 341 Fig. 165 Intragranular fracture of a ferrite grain in malleable.. 342 Fig. 166 Intergranular failure of malleable 343 Fig. 167 Ferrite grains in malleable, showing slip in two planes at right angles 344 Fig. 168 Slip bands due to plastic compression in malleable iron 345 Fig. 169 Plastic deformation of malleable in compression 345 Fig. 170. Same specimen as shown in Fig. 169 347 Fig. 171 Path of cross bending rupture through malleable 348 Fig. 172 Malleable iron compressed about one half. Annealed 5 hours at 650 degrees Cent 349 Fig. 173. Stress strain diagram of malleable iron in tension for two rates of loading 352 Fig. 174 Changes of strain with time at small increments of stress 353 Fig. 175 Changes of strain with time under considerable increment of stress (about 70 per cent of ultimate strength) 354 Fig. 176 Stress strain diagram of malleable iron in repeated tension under increasing loads 355 Fig. 178 Behavior of malleable under cyclic cross bending at con- stant maximum stress 358 Fig. 179 Maximum deflection and permanent set under cyclic cross bending at constant maximum stress 359 Fig. 180 Stress deflection diagram of malleable in cross bending with and without previous cold work 362 Fig. 181 Effect of torsional deformation upon subsequent tensile strength of malleable 363 Fig. 182 Absorption of energy from successive impacts 364 Fig. 183 Load deformation diagram of specimen subjected to al- ternate impact , 367 Fig. 184 Magnetization and permeability curves of malleable cast iron .' 373 Fig. 185 Magnetic properties of malleable cast iron 375 Fig. 186 Variation of electrical resistance of malleable cast iron with temperature 377 Fig. 187 Expansion of malleable cast iron 379 Fig. 188 Heat transfer from machined malleable to still water for various temperature differences 380 Fig. 189 Effect of temperature upon tensile properties of mal- leable . 382 Fig. 190 Thermal conductivity of malleable cast iron 383 XVIII American Malleable Cast Iron I EARLY HISTORY OF IRONMAKING SINCE the dawn of civilization man has continuously labored to use the natural resources of the world for his own well being. He first adapted to his needs the materials most easily obtained and as his knowledge and skill grew he sought to find or make other materials which would better suit his requirements. Copper and gold, being found in the metallic state in nature, were the first metals to attract his attention. More- over, being malleable, these metals were readily fashioned into the shapes desired. Far beyond even legendary his- tory the mound builders used copper utensils while the Incas and Montezumas used gold in domestic articles as well as in ornaments. Of the various metals found as compounds in na- ture, lead, silver and tin are fairly easily reduced from their ores ; hence prehistoric metallurgists soon added these to the list of available materials. Thus the age of copper was succeeded by the age of bronze. The only free iron found in nature is that of meteoric origin, usually existing in small fragments which easily rust away. However, in a few cases, notably the three large siderites brought from Greenland by Admiral Peary and now in the American Museum of Natural History in New York, meteoric iron has been put to industrial use. Peary's siderites, which are the largest ever discovered, constituted the only source of iron for the Esquimaux of northern Greenland. Approximately five thousand years ago, one of Pharoah's 2 c *" AnteriCcirt m ^M(ikletfble Cast Iron Courtesy of American Museum of Natural History Fig. 1 A meteorite in the American Museum of Natural History, New York, brought from Greenland by Admiral Peary masons carelessly left one of his tools lying on the masonry where a new stone was being set in building the pyramids. Thus packed in lime, this earliest known piece of man-made iron was preserved for posterity. The method doubtless used by the Egyptian iron masters still persists in many semicivilized communities. As shown in Fig. 2 it consisted of heating rather finely divided ore in a charcoal fire blown by a hand or foot bel- lows in a shallow basin in the ground. The charcoal acted both as fuel and as a reducing agent, liberating metallic iron. The temperature being low, the iron did not combine much with the carbon nor did it melt freely. The pasty bloom which accumulated in the hearth was removed and crudely hammered into the desired shape. Obviously the process was laborious, yet it was practiced on a considerable scale. It is believed that the famous pillar of Delhi was made by welding together blooms of the kind just described. , Metal of this kind possessed some of the properties of wrought iron or unusually soft steel of the present day. Early History of fronmaking However, it doubtless was variable in quality since the carbon content must have fluctuated considerably due to the changing and uncontrolled temperature conditions. Not- withstanding this lack of uniformity, it was decidedly a bet- ter metal for tools and arms than the copper and bronze preceding it. Still before the era of written history there lived a primitive Carnegie whose very name has been lost. This early steel master, probably a native of Greece, determined to engage in the quantity production of iron. He substituted a stack or shaft for the shallow hearth then in use with the hope of rendering the operation continuous instead of in- termittent. He introduced blast from the bellows at the bottom, started a fire of charcoal and then began to add alternate layers of charcoal and ore until the shaft was full. Presumably he expected to dig out blooms of iron from the bottom of the furnace at frequent intervals and to supply V ## m .r^. Fig. 2 A primitive furnace, 1500 B. C. The illustration was repro- duced from an Egyptian wall painting -I ntcr icon Malleable Cast Iron charcoal, ore and air continuously. Doubtless he was much surprised when on some occasion instead of iron blooms ap- pearing, molten metal ran from the opening in the stack. Such was the first production of cast iron. The better utilization of heat in the shaft furnace had produced a tem- perature high enough to more completely carburize the product. The decreased melting point, coupled with the higher temperature reached, produced a liquid metal prob- Fig. 3 One of the earliest blast furnaces \ ably of white or mottled fracture. Unconsciously this primitive artisan discovered the blast furnace. Even today the process of smelting iron ore is governed by the same general principles which obtained in the early days in Greece. Only the technique has been perfected. The earliest known blast furnace purposely to make pig iron is said to have operated in the Rhine provinces of Germany in 1311. The industry spread over the rest of Europe dur- ing the succeeding century. With the development of the crude blast furnace, one Early History of Irontnaking of which is illustrated in Fig. 3, there existed two kinds of iron. The one had to be forged to shape and was rather soft although not easily broken and the other, which could be cast into shape, was rather hard but too brittle and fragile to use. Obviously, a metal of either of these limita- tions was not exactly adapted to the making of swords, the manufacture of which constituted a most important pro- fession in the early days. Therefore, the most important metal for that age was one not soft enough to be bent and blunted by armor nor so brittle as to be shattered by a sharp blow. In the search for a material to better meet the requirements of the armorer some pioneer found that if the soft iron produced in the forge were heated in charcoal, the surface of the metal could be made harder in fact the metal could be hardened throughout if the treatment were continued long enough. It was learned that in this man- ner tools and weapons could be produced with a superior edge. For many centuries this "blister" or cementation steel was the only steel available. One of its principal shortcom- ings was its lack of uniformity across the section. How- ever, this was later overcome by remelting the carburized steel in crucibles, thus rendering it homogeneous. The crucible process also was modified by melting wrought iron mixed with sufficient charcoal or cast iron to give the desired properties to the metal. The amount added was determined em- pirically, for at that time chemical control from the viewpoint of carbon content was unkown. Thus at the beginning of the eighteenth century three kinds of iron were known to the world. These were wrought iron, soft and worked only by forging; cast iron, brittle and worked by casting; and crucible or cementation steel, some- times melted in the process of manufacture but always forged to shape, not brittle but hard enough to hold an edge and be tempered. Steel, however, could only be made from wrought iron, wrought iron only from ore, and neither could be made from the relatively cheap cast iron. The next forward step in the metallurgy of iron and in fact the first since the dark American Malleable Cast Iron ages, was the invention by Cort of the puddling furnace for converting molten cast iron into blooms of wrought iron by treatment with ore. This invention made possible the reduc- tion of the metal from its ore in the cheaply and efficiently operated blast furnace and its later conversion into mal- leable and ductile wrought iron. Steel was sitill made by using wrought iron, now ob- tained by puddling, as the raw material. This continued to be the only source of steel until the discovery of the bessemer process in the middle of the nineteenth century and the invention of the open-hearth furnace by Siemens about 15 years later. Both of these processes, which depend for their success on the increased temperatures available, pro- duce liquid steel of nearly any desired carbon content. The former process uses the carbon and silicon content of the molten pig iron for fuel, burning these within the charge by a blast of air. By the removal of the carbon, the cast iron becomes steel which is kept liquid by the heat of combus- tion of the carbon and silicon. Siemen's was practically a modified reverberatory fur- nace fired by gas, the fuel and air for combustion being heated in regenerators by the waste heat of the escaping products of combustion. The oxidation *of carbon was ac- complished, as in the puddling furnace, by the oxygen of the hematite iron ore added to the slag. The essential dif- ference between Cort's and Siemen's invention was that the latter worked at temperatures sufficiently high to keep the resulting product molten. A review of the industrial world at about the close of the American civil war indicates that five well established types of iron and steel were being used. Charcoal iron was made directly from ore and charcoal on the same principle used in pre- historic "times. This material resembled wrought iron and was practically obsolete from a production viewpoint. Wrought iron was made from cast iron in the puddling fur- nace. It was a pasty mass and was shaped by rolling and hammering only. This material was soft, malleable and ductile. The railroad iron of which the MONITOR'S armor Early History of Ironmaking was made was of this character. A third material was cast iron made in the blast furnace and cast to shape in molds. This iron was incapable of being bent without breaking. The fourth material was blister or cementation steel made from wrought iron in unimportant amounts. This steel had to be forged to the shape desired. The Fig. A An early American blast furnace fifth and most important metal was steel made in liquid form by the crucible, bessemer or open-hearth process from cast iron. This had so high a melting point that it was incapable of casting any but large molds, hence it was usually cast into the latter form and rolled or forged to shape. When desired it could be produced of a composition permitting of hardening and tempering. A sixth product, then just coming into use is the subject of this volume. It will be observed that in none of the first five products are combined the properties of malleability of wrought iron and fusibility as found in cast iron. In other words, no ma- terial has been described which could be cast into intricate American Malleable Cast Iron a bb Early History of Ironniaking 9 shapes and which would be in any degree malleable when complete. The problem of producing a malleable cast iron to fulfill these requirements had long occupied the minds of the iron masters. Since Cort had produced wrought iron by the use of ore, a modification of his process which would not involve the melting of the cast iron now seems to us a logical conclusion. In 1722 Reaumur, a French physicist, described a process, not necessarily original with him, for producing malleable cast iron by packing small castings of (presumably white) cast iron in pulverized hematite ore and heating them to bright redness for many days. This method evidently was suggested by the cementation process for making steel from wrought iron, substituting for the charcoal which adds carbon in that process, ore which removed carbon, the same reaction later discovered by Cort as applied to molten cast iron. Reaumur's discovery, or better disclosure, actually grew into an industry in Europe. It happened that European white cast irons, except in Sweden, were relatively low in man- ganese and high in sulphur, owing to the available fuels and ores. Being white, it also was low in silicon. Such conditions are all unfavorable to the formation of free carbon and consequently Reaumur's reaction was never complicated by the formation of temper carbon or graphite. In intention, at least, the annealing removed from his thin castings all the carbon which burned from the carbide of iron. The amount of the carbon originally present was im- material, in any event the resulting casting, if the anneal was successful, had only traces of carbon but contained all the other chemical elements originally present. Having been only moderately heated it retained its original cast form but approximated the chemical and physical properties of wrought iron. The shortcomings arose mainly from the fact that since carbon was removed through the surface, the process oould not be commercially applied to moderately thick sections owing to the prohibitive annealing time. Moreover, a casting having both thick and thin parts nat- American Malleable Cast Iron Urally would be completely decarburized in the former while still retaining much carbon in the center of the heavier por- tions. If the process were continued to completion in the thick sections, trouble from oxidation and scaling of the thinner parts would be encountered. Furthermore, it was difficult to be sure that the castings were annealed clear through, since the interior is not available for inspection. Any castings not annealed through would be brittle owing to the remaining undecarburized core. Hatfield in his "Cast Iron in the Light of Recent Research" says of this process as practiced in England : "Essentially, the materials used in Britain in the production of malleable castings, are high in sulphur, necessitating a somewhat lengthy anneal at a fairly high temperature with a view to annealing largely in decarburization. These re- marks apply also to the practice in France, Switzerland, Belgium and Germ.any." Production of "White Heart" Limited The industry thus was limited to comparatively small tonnages and hence to crude methods. As practiced then, and still practiced in England, Germany and France, the product is used largely for harness parts 'and small and un- important work. Melting is frequently done on a small scale either in crucibles or cupolas. The total volume of production is relatively insignificant in the iron production of Europe aUhough there are said to be 126 white heart malleable foundries in Great Britain. Reaumur's publication was productive of only the most meager commercial results from an American viewpoint. Boyden and his immediate successors attempted to anneal by decarburization. The metal made by the Philadelphia Hardware and Malleable Iron Works before the Civil war was "white heart", as was that of at least a number of its contemporaries. About 1861, however, the manufacture of this product in America practically ceased. A single job- bing manufacturer of white heart malleable continued op- erations until a few years ago, operating largely on European Early History of Ironmaking pig iron. At least one plow manufacturer continues to operate on the basis of European cupola practice and to turn out white heart malleable of high strength and low elongation. Many of the stock phrases regarding malleable which have gone the rounds for many years originated with "white heart" metal. For instance, the fairly widespread belief that malleablization takes place from the surface in, that the material is not annealed clear through and that the material cannot be used in heavy sections because of the unannealed center, are among the common fallacies handed down from Reaumur's time. Even though the "white heart" or Reaumur's process never has possessed any tonnage significance in the United States, and has been practically discontinued for 60 years, its faults have been frequently assumed to apply to the American or "black heart" metal by those not conversant with the facts. The art of making malleable castings, as that term is understood in America, was discovered probably uncon- sciously by Seth Boyden while attempting to practice Reaumur's method in Newark, N. J., in 1826. Boyden was a manufacturer rather than a scientist. Probably for this reason no formal announcement of a new discovery was made. It is presumed that in attempting to duplicate European practice with American pig iron, which is low in sulphur and high in manganese, he inadvertently discovered an alloy which when heated to produce deear- burization, graphitized instead. The product possessed all the properties of the best white heart metal and was more easily made and more uniform. Not realizing that he had discovered a new art, Boyden continued this work along the lines he found empirically most likely of success. Boyden left a diary covering his experiments from July 4, 1826, to Sept. 1, 1832. It shows that he was at- tempting to duplicate Reaumur's process. Under date of Oct. 20, 1826, he writes : "I have a piece so good it will not harden any more than copper". Yet from his third ex- 12 American Malleable Cast Iron Fig. 6 Statue of Seth Boyden, erected in the city park of Newark N. J., by citizens in memory of the man who laid the foun- dation for the malleable industry in the United States Early History of Ironmaking 13 periment on there are allusions to graphitization. In the report on the third experiment he states, "Much blacker in- side and not half so good". Again in Experiment No. 5 he refers to a piece "which had been done totally well before rendering dark in the middle". An entry on the eighth ex- periment is: "Quite gray; none of the above bend or are good for anything". In the ninth experiment he comments : "Hard iron melted in coal dust from the air received no change but in scoria and coal dust became soft gray iron. A piece of Sterling (grade of pig iron) without W (prob- ably wrought iron) in soft gray state done (annealed) eight times remains gray and unmalleable". Boyden had been unconsciously recording the first ob- servation of the formation of temper carbon and its dis- tinction from graphite. Being still convinced that he was striving to produce a steely decarburized iron he refers in Experiment No. 11 to the fact that "the iron was tough when broken and was rather too dark in color". Yet in the next experiment he writes, "Experiment in the foundry. Sterling the toughest but very dark. Sprues and Sterling dark and good". On Sept. 10, 1826, he notes that "some of the pieces were tough, gray and very good". On Oct. 20 of the same year he makes the peculiar observation that "the best piece I have ever seen.... was pale blue in the middle". For many years neither he nor his successors realized that decarburization was not essential to the process. He and his associates laid great stress on packing materials and their chemical effect upon the product. Inasmuch as the graphitizing reaction discovered by Boyden forms the metallurgical basis of the present indus- try, its consideration in detail will be reserved for a later chapter. Black heart or American malleable cast iron bears no metallurgical relation to the European product and its history begins not with Reaumur but with Seth Boyden. II DEVELOPMENT OF MALLEABLE INDUSTRY IN THE UNITED STATES SETH BOYDEN began business as an iron founder in 1820 at 26 Orange street, Newark, N. J. Being inter- ested in malleable castings, he attempted to duplicate European practice at a time when metallurgy was prac- tically unknown. After six years of continuous experiment he succeeded in producing malleable castings, but not of the kind he attempted to make. Due presumably to the raw material available, he hit upon the practical operation of the graphitizing anneal and thus founded a new industry. Boyden operated the plant under his own name until 1835 when it became known as the Boston Malleable Cast Iron and Steel Co. The foundry continued under this management for two years, after which it was operated under various firm names by Daniel Condit, J. H. Barlow and others, becoming in 1907 the Barlow Foundry Co. This company occupied the original site until May, 1914, when it removed to another location and the birthplace of black heart malleable was razed. Quite naturally the early development of the industry centered about its discoverer and its birthplace. At one time Newark had eight malleable foundries, and three of Boyden's brothers Otis, Alexander and Frank engaged in the malleable founder's art. Otis operated a foundry in Newark from 1835 until 1837, when it was absorbed by the Boston Malleable Cast Iron & Steel Co. Alexander and Frank engaged in the business in East Boston during the same interval, after which Alexander was employed by Frederick Fuller, of the Easton (Mass.) Iron Foundry, es- tablished in 1752. The business later came into the hands of Daniel Belcher and was continued by his descendents. Two plants were started in Elizabethport about 1840 and in 1841 David Meeker began to manufacture malleable 16 American Malleable Cast Iron to JH O u 10 Q> O O CO in 'G Q) O PQ , g _a o % *s r ~ (ji Q S C ^ -r >o ^- >o t~ i^ oo a-' 5\ ON ooooooooooccoococoooso 3 o 3 t t; ^ ^ ^ g'S v2 sT'^ "^ - -^ ^ ~^^~ $$. i &'i > &X&*?m$"$&&*5 Development of Malleable Industry 17 in the Hedenburgs Works. The New Jersey Malleable was founded at Newark in 1841. The information regarding the activities of the Boyden brothers and their contemporaries and associates is derived from a paper presented before the Philadelphia Foundry- men's association by George F. Davis. It would be exceedingly interesting to trace back to its beginnings the present, highly developed industry. Un- fortunately written records of the early days survive, if at all, only in the account books and the minutes of stock- holders' and directors' meetings of the older corporations. Such records are not open to public scrutiny and therefore it is difficult in sketching the early history of the industry to do full justice to all. The writer has been unable to trace in complete detail the early history of the industry, other . than through Boyden's activities. This may be due to the fact that these older plants did not survive or may be caused by in- adequate search. It seems to be of common knowledge that during the first half of the nineteenth century, a number of persons entered into the business, the plants being mainly located in New England and New York, at least one as far west as Buffalo. Thomas Devlin has informed the writer that when the Philadelphia Hardware & Malleable Iron Works, now the Thomas Devlin Mfg. Co., was 'founded in 1852, the com- pany officials knew of the existence of the Westmoreland (N. Y.) Malleable Works, of a plant in Worcester, Mass., and also of the M. Greenwood Co., of Cincinnati, which was founded in or possibly before 1850 and later was taken over by James L. Haven. In the early fifties, Isaac Johnson established a mal- leable foundry at Spuyten Duyvil. In 1872 he, together with J. H. Whittemore of NaugatUck and W. S. Nichols, a brother-in-law and representative of Walter Wood, organ- ized the Hoosick Malleable Iron Works at Hoosick Falls, New York. Some years later, Johnson also organized the malleable plant bearing his name in Indianapolis, which in 18 American Malleable Cast Iron 1883 passed into the control of the group which later be- came the National Malleable Castings Co. In the early eighties, the Walter Wood Mowing & Reaping Machine Co. absorbed the Hoosick Malleable Iron Works, enlarging the plant from time to time. The same organization under the style of the Walter Wood Har- vester Co. started the business in St. Paul which, after a failure during the panic of 1893 and one or two changes of ownership, became the Northern Malleable Iron Co. under Frank J. Otis. Much of the early development centered in New England, particularly in the state of Connecticut. Among the oldest malleable plants is what is now the Naugatuck works of the Eastern Malleable Iron Co. at Union City. Here the development work of J. H. Whittemore and B. B. Tuttle was done beginning in 1858. From that plant and that of the ^Bridgeport Malleable Iron Co. were recruited many of the executives who established the industry in the Middle West. The Naugatuck and Bridgeport plants, with those at Troy, Wilmington and New Britain became the present Eastern Malleable Iron Co. At a later date the village of Hoosick Falls, N. Y., also sent westward a group of mal- leable iron foundrymen. G. H. Thompson went to Colum- bus, John Haswell to Marion, and later to Dayton, O. Sidney Horsley, superintendent of the Northern Malleable Iron Co., and others also graduated from Hoosick Falls. In 1854 Duncan Forbes, a Scotchman who had previously resided in western New York, removed to Rockford, 111., and with his son Alexander Duncan Forbes, established a gray iron foundry. In 1859 Forbes installed an annealing oven and intermittently produced cupola malleable castings in connection with the production of gray iron stoves which constituted the larger part of this business. In 1864 the gray iron portion of the business was definitely abandoned in favor of malleable castings alone. Duncan Forbes, the first manufacturer of malleable castings west of Cincinnati, died in 1870. The business was __ Development of Malleable Industry 19 continued and enlarged by others of his family. In 1890 the company was incorporated as the Rockford Malleable Iron Works and in 1907 removed to a new location in Rockford, where it continues to be operated by descendants of the original founder. In 1866 Charles Newbold and Peter Loeb started a malleable and gray iron foundry in the east end of Day- ton, O., which was incorporated as the Dayton Malleable Iron Co. in 1869. In 1872 the business was removed to its present location on West Third street/ and from time to time the capital stock and plant equipment were increased. In 1916 the plant of the Ironton Malleable Iron Co. was purchased, and has since been operated' as the Ironton works of the Dayton Malleable Iron Co. In February, 1922, the Dayton Malleable also took over the foundry of the Timken Co. at Canton, Ohio. In August, 1868, the Cleveland Malleable Iron Co. was incorporated and in 1869 Alfred A. Pope became inter- ested in the business and immediately thereafter its presi- dent. In 1873 John C. Coonley, sometime of Louisville, and a number of men in the Cleveland company, started the Chicago Malleable Iron Co. The same organization, which in 1891 became the National Malleable Castings Co. ac- quired by purchase or construction, plants in Indianapolis, Toledo, O., and Cicero and East St. Louis, 111., besides steel plants whic'h are not of interest in the present con- nection. A. A. Pope and J. H. Whittemore were leading factors in the early growth of- the industry, the institutions over which they presided now being the two largest in the country. Many other manufacturers of malleable cast iron have honorable histories extending back into the sixties and seventies of the last century. The writer has not had the opportunity he could have wished to do full justice to the histories of some of these smaller companies. The industry has the distinction of numbering on its rolls a president of the United States, Mr. Harding having been one of the original stockholders of the American Mai- 20 American Malleable Cast Iron 111 " it ' cy o C *> ^ o II N C x 2 . - "^ Development of Malleable Industry 21 leable Castings Co., organized in 1905 under the leadership of Charles L. LaMarche. In this as in other industries, the growth has been largely in accord with the survival of the fittest. Many plants have been started on a moderate scale especially at times of great industrial activity. Some of these failed to survive the first period of depression encountered ; others, particularly in New England, continued a 'small but often prosperous existence, catering to a limited trade, usually in their immediate neighborhoods. A number of the organiza- tions grew in size and influence, and by sound business and technical methods, coupled with an aggressive policy, at- tained position of prominence in their fields. Another type of malleable foundry has sprung up in the history of the industry. This is the foundry which pro- duces a given specialty not for the open market but as a department of an organization manufacturing a finished product. Many of these foundries also have branched out into jobbing work when not fully employed for their own requirements, but are primarily operated to furnish castings for the product made by the parent company. In this class are the malleable foundries of the General Electric Co,., American Radiator Co., International Harvester Co. of America, Ebefhard Mfg. Co., Link-Belt Co., and a number of others." These foundries, having a definitely established outlet for their product and the financial and administrative support of a well organized industry, usually have survived and grown successful. However, in at least one case a found- ry of this character has been sold to a malleable founder in pereference to its continued operation by the consumer. The organization of a special foundry is only possible where the requirements of the parent company run up to a sufficient tonnage to make possible operations on a large enough scale to warrant the best operating conditions and supervision. The malleable industry, involving more expen- sive equipment and greater technical skill than the gray iron industry, cannot well be operated in small units on account of excessive overhead. Therefore, unless an in- dustry is large enough to operate'quite an extensive foundry, 22 American Malleable Cast Iron Development of Malleable Industry 23 castings of better quality usually can be obtained more cheaply by purchase from established jobbing foundries. The present extent and distribution of the malleable industry is shown in Fig. 9. Each dot on this map repre- sents the approximate location of a producer of malleable castings. The list" is as complete as possible, having been compiled from the data of the American Malleable Cast- ings association and from information gathered for the government during the war. There are a number of plants marked as producing malleable where there is reason to doubt whether they have actually done so. The most recent lists include between 20 and 30 more plants than are shown on the map. In part this may be due to incomplete returns, and to a less degree to new foundries of very small capacity. It is un- likely that any important plant has been omitted. It will be noted that the plants are largely in the territory north of the Ohio and east of the Mississippi rivers, their locations following closely the various divisions of the Pennsylvania, New York Central and New York, New Haven and Hart- ford lines, with an additional development near Milwaukee and in southern Michigan. These locations were largely determined by the fact that they are coincident with the important manufacturing districts of the country, present good shipping facilities and are conveniently near the sources of fuel and pig iron. A more interesting compilation from the viewpoint of the user of malleable cast iron is the map shown in Fig. 10, showing the principal sellers of malleable cast iron. This has been prepared from the previous map by the elimination of foundries primarily operated as departments of larger in- dustries producing finished products, as for example harness parts, pipe fittings, etc., and of foundries whose tonnage is not of sufficient magnitude to be an important consideration from the viewpoint of the consuming interests. It will be seen that the distribution is almost identical, although the number of plants has been considerably reduced. Annual Production of Malleable The plant capacity of the United States as of 1920 is estimated at 1,286,300 tons annually, divided by states as 24 American Malleable Cast Iron in Fig. 11 Comparison of production of steel and malleable iron castings in the United States The production of steel castings is charted from the statistics compiled by the American Iron and Steel institute. The production of malleable iron castings is carefuDy estimated on the basis of the known production of plants whose output constitutes the majority of the annual tonnage of the country. No definite figures ever have been compiled showing the actual production of malleable castings and in the absence of such information, it is believed that the above charted values are as accurate as any that can be obtained under existing circumstances. follows: Illinois, 297,700; Ohio, 202,700; New York, 167,- 500; Pennsylvania, 133,100; Wisconsin, 116,600; Michigan, 108,400; Indiana, 88,300; Connecticut, 58,200; and all others, 113,800. The most complete information at the writer's disposal indicates that there are between 176 and 204 manufacturers of malleable castings in the United States. In this list, however, are in- cluded a considerable number of manufacturers with whom the production of malleable castings is only incidental to other operations. Some of these produce malleable only intermittently or in small quantities, as their own need require, and are included here in the interest of complete- ness rather than: because of their importance to the job- Development of Malleable Industry 25 bing trade. About 85 per cent of the tonnage of the country is produced by 76 manufacturers having a capacity in ex- cess of 5000 tons annually each. Sixty-two and one-half per cent is produced by 33 owners having capacities over 10,000 tons per annum. No single manufacturer can be said to exercise anything approaching a monopoly, as the five largest interests together have only 28 per cent of the capacity of the country. Each of these five can produce 30,000 tons per annum or more. The eight additional manu- facturers, having individual capacities from 20,000 to 30,000 tons, account for an additional 14.2 per cent. Data as to the production of malleable castings in the United States go back only to 1913 when the American Malleable Castings association began the accumulation of statistics on this point. Fig. 11 shows the production of mal- leable castings by years since 1913, compared with the production of steel castings by years since 1904, as recorded by the American Iron and Steel institute. The production of malleable pig in 1913 was about 5 l /> times that in 1900. This would imply a production of only about 127,000 tons of malleable castings in that year. It will be noticed that the annual production is con- siderably below the annual plant capacity. The recent fig- ures on production of course are based on manufacturing operation in times of great industrial activity. Under these conditions, a deficit in production compared with capacity seems at first glance unlikely. This deficit is due to two causes ; first the fact that the reported capacities are doubt- less a little higher than the facts warrant and second that most of the malleable foundries had been unable to obtain either sufficient labor or fuel to permit the realization of their full production. In other words, plants apparently were built in excess of the available labor supply and no material increase in the country's maximum output of castings could be expected as a consequence of the erection of ad- ditional plants. In view of the lack of productivity in all lines of manu- facture before the 1921 depression, it is unlikely that the manpower of this or any other industry could be largely increased except by 26 American Malleable Cast Iron increasing the productivity of the individual employe. The only visible remedy seems to be an increase in tonnage per man by the introduction of every possible mechanical aid. A considerable improvement may be possible by some means tending toward a decreased loss of time by the in- dividual worker, and an increase in his skill. Data based on con- ditions since the summer of 1921 are of course valueless on account of the very small production in all lines. The commercial development of the industry was par- alleled by steady progress in the technical details of malle- able production. As has been stated earlier in the discus- sion, Boyden's discovery was not the result of a logical metallurgical development but was the accidental outgrowth "of an attempt to practice a theoretically distinct art. When it is realized that all of this work was done at a time when even the chemical analysis of iron was an unusual thing and that Boyden and his successors blazed the way without any knowledge of variations in raw material and product, save what might be gathered by the crudest of inspection, we cannot but marvel at their courage and persistence in estab- lishing the empirical basis for the present great industry. Boyden, however, having a truly technical mind, left behind complete notes of his experiments and the results attained. Some of his notes already have been quoted. He recog- nized the presence of carbon but only in the free state, be- lieving white iron to contain none. He made many experi- ments with various packings and under different annealing conditions, finally concluding that red iron ore was the best material. He records the belief that the annealing temper- ature should be at least the melting point of silver. He considered the presence of silicon and sulphur but knew nothing of analysis. Under date of Jan. 23, 1829, he records observations as to the effect of additions of phosphorus, clay, lead, zinc, tin and antimony. Boyden's brother, Alexander, seems to have been the earliest mystery monger in the trade, it being related by Davis, on the authority XD Horace Spaulding, the last sur- vivor of the Easton foundry, that Alexander had a little pump with which he squirted something into the stack Development of Malleable Industry 27 and also that he used to throw some metal into the furnace, creating a great volume of smoke and doubtless an equally great awe in the mind of the spectator. In 1872 Alfred E. Hammer, of the Malleable Iron Fit- tings Co., began to study the chemistry of black heart malleable at Branford, Conn. This company in 1864 had succeeded to the business of an earlier one, 'the Totoket Co. founded in 1854 for the purpose of practicing Boyden's method. In 1875 Mr. Hammer had established a chemical lab- oratory which was, so far as he is aware, the first in con- nection with the malleable iron industry. Writing of this laboratory he says : "I found that I was practically in an unknown country. For that reason, 'however, the work was not only interest- ing but positively exciting so much so that I had a mat- tress laid in my laboratory and with the aid of an alarm clock, I was able to follow the then tedious chemical oper- ations through the night without much loss of sleep." As a result of this work, he found it possible to lay down "a chemical ratio as between carbon and silicon, and manganese and sulphur," thus being the first one "to bring the malleable process down to a chemical proposition. " Among Hammer's difficulties, not the least was that at this period, pig iron was not made and sold by analysis. However, he soon applied his chemical ideas to the selection and mixing of irons, irrespective of brands and grades. The work of this pioneer metallurgist seems to have escaped adequate recognition since his conclusions \vere thought to be too valuable" trade"* secrets to 'warrant publi- ;* cation. He certainly came to correct quantitative conclusions as to manganese and sulphur at a date long before the the- oretical explanation was even thought of. His views as to carbon also seem to have been far in advance of later inves- tigators who gained much greater general recognition. While not a technically trained metallurgist, A. A. Pope from his earliest association, with the industry strove by every means in his power to collect and interpret ex- 28 American Malleable Cast Iron > *-> Development of Malleable Industry 29 perimental data bearing on the processes and products of his plants. These investigations were conducted by Em- merton, Benjamin and others and resulted in an accumulation of valuable data during the seventies and eighties. It is interesting to note that especially with reference to manganese and sulphur, Mr. Pope's conclusions were very similar in application to those reached at about the same time by Mr. Hammer. Among the most progressive of the malleable manufac- turers was the late B. J. Walker, of Erie, Pa., who pursued a most liberal policy with regard to the exchange of in- formation and did much to develop the industry. In 1893 McConway & Torley established in Pittsburgh what is frequently said to have been the first laboratory in the malleable industry. It was under the direction of Dr. Richard Moldenke, who not long thereafter severed his connection with that company to become associated with another in the Pittsburgh district. H. E. Diller, now metal- lurgical editor of The Foundry, was associated with Doctor Moldenke at this time. In the autumn of 1893, James Beckett, after a tour covering all of the malleable foundries then producing agri- cultural implement parts, found that none of them had es- tablished a chemical laboratory for works control. This statement does not apply necessarily to plants not engag- ing primarily in this specialty which were not visited by Mr. Beckett. In 1894 the Wood Mowing & Reaping Machine Co., Hoosick, Falls, N. Y. employed Enrique Touceda as a consultant and established a well equipped laboratory. In 1903 when the National Malleable Castings Co., established a works and experimental laboratory at Indian- apolis, the author was unable to find by diligent search of the literature available any adequate information of a definite and quantitative character regarding the chemical fundamentals of the process. Therefore it was decided to disregard precedent and to establish a sound theoretical basis for works control, using the information accumulated by Mr. Pope as a nucleus. In this connection the quantita- 30 American Malleable Cast Iron tive effect of carbon, a sine qua non in the works control of the product, was worked out in 1904, the conclusion reached being apparently new to a number of the best in- formed malleable men with whom it was discussed at the time. A little later the effect, or rather lack of effect of manganese sulphide was also worked out. This offers the theoretical explanation of the practical observations of Mr. Hammer and Mr. Pope. These facts were certainly discovered independently by other observers, including W. R. Bean now of the Eastern Malleable Iron Co. In the absence of contemporary publi- cation it is impossible to state whether these discoveries preceded or followed the Indianapolis investigations. So far as the writer has been able to learn the Indianapolis laboratory was the first to successfully exercise complete works control on the basis of the total carbon content being the determin- ing factor in the quality of the product. During all of this time the results of none of these in- vestigations became publicly available and therefore it is difficult to accurately chronicle the scientific development of the art. The organizations collecting scientific and re- search data of value did not feel it to be sound business pol- icy to make public disclosures of their work. Regardless of whether or not this policy was fundamentally sound from the manufacturers' viewpoint, it certainly proved a handi- cap to the consumer, who remained in ignorance both of the theoretical principles and practical applications of the manu- facture of malleable castings. Having again severed his business connection and es- tablished himself as a consultant. Dr. Moldenke began to contribute voluminously to the technical press. Unfortu- nately the only sources of information open to him seem to have been the work in which he personally participated. Furthermore he was presumably handicapped by the con- fidential character of his relations with his clients and ap- parently felt constrained to speak only in somewhat gen- eral terms. Nevertheless he did yeoman service in striving for a better interchange of ideas and information, and also in advocating suitable technical control of the industry. Development of Malleable Industry 31 His services in this direction are probably of even greater importance than the actual informative value of his literary output. The earlier literature of the subject was derived di- rectly or indirectly almost entirely from his publications. There still persisted in the engineer's handbooks and in the technical press a mass of ill-supported conceptions largely predicated on a confusion with the white heart process. For example, great weight was attached to the oxidizing action of the packing and its effect on the proper- ties of the product was unduly emphasized. Great dif- ferences also were supposed to exist between the heart and surface of the same casting. Similarly there was an im- pression that malleablizing proceeded from the surface in- ward and was complete at the surface before it had pro- gressed far at the center. A corollary to this belief was that very thick castings could not be annealed clear through. Since none of those who knew better felt called upon to publicly combat statements of this character, it is not surprising that the engineering public was left in ignorance and hence in distrust of the qualities of the material. More- over it is not surprising that in the absence of guidance by those better informed, some of the less intelligent and progressive manufacturers did not clearly understand the principles of the process they practiced and therefore pro- duced unsatisfactory castings. A few of the larger producers maintained adequate laboratory facilities to investigate and control their methods. The smaller manufacturers, however, had to get on as best they could with their own resources until the American Malleable Castings association undertook as one of its ac- tivities to carry on extensive research work for the benefit of its members. Prof. Enrique Touceda, of Rensselaer Polytechnic institute, was employed as consulting engineer and since 1913 has labored unceasingly to instruct the mem- bers of the association in sound practice and -correct funda- mental principles. This work was largely confidential in character and added little to the user's knowledge but con- 32 American Malleable Cast Iron Development of Malleable Industry 33 tributed immensely to his satisfaction in the use of the product. At about this time, Oliver Storey published the results of some research work at the University of Wisconsin, deal- ing with the fundamentals of the graphitizing reaction. In the writer's opinion this was the first scientific American contribution to the literature of- the metallurgy of malleable iron. The problem has been since investigated by Archer and White, Merica and by the writer. A few years earlier, Hatfield thoroughly investigated the less important subject of decarburization. The theoretical aspects of graphitization have been studied abroad rather thoroughly. In 1881 Forquignon published in the Annals de Chemie et de Physique a contribution dealing with his tests in the an- nealing of malleable iron and steel. Unfortunately the au- thor has been unable to familiarize himself with this pub- lication, which is said to have dealt very adequately with the subject. In 1902 Charpy and Grenet published a study of the graphitization of white cast iron which covers the ground very fully and accurately, even in the light of present knowledge. This publication seems to be almost if not en- tirely unknown in this country. Howe, in the Transactions of the American Institute of Mining and Metallurgical En- gineers in 1908, discussed critically and exhaustively the evidence then available bearing on graphitization. Hatfield in 1910 discussed the chemical physics of the precipitation of free carbon from iron carbon alloys in a paper before the Royal society. In 1911 Rueff and Goecke published a study of the solubility of carbon in iron and in the same year Ruer and Iljin discussed the stable system of iron carbon. Heyn summarized the contemporary knowledge of the iron carbon alloys at the New York congress of the International Society for Testing Materials in 1912. That these technical investigations have been of so little service to the American manufacturer seems a reflection upon the 34 American Malleable Cast Iron ined Carbon 1.002 \ 5 ^JlE^ 03 Time Scale Will Depend On Chemical Composition Fig. 36 Chart showing conversion of combined carbon into temper carbon The graphs show the relation between the carbon remaining combined and the lapse of time at each of five temperatures. Note the increasing velocity and higher carbon content of the conclusion at high temperatures as compared with low. complished toward the location of 5"' as affected by variations in other elements than carbon. These points are of very great academic interest, but from an operating viewpoint are inconsequential. No operating errors will be involved in considering the line to be straight and joining the two points mentioned. Fig. 36 show r s in diagrammatic form the decrease in com- Metallography of Malleable Iron 67 bined carbon according to the time of exposure to various temperatures. It will be noted graphs are given for each- 'of a number of temperatures. The horizontal or time ordinates have been plotted to scale ; however, the values given for this dimension are suggestive only, since the rate of graphitization and hence the time to attain equilibrium at various temperatures is dependent on the chemical composition with respect to other elements in addition to carbon. The figure is given as an ex- ample of what may happen rather than for quantitative inter- pretation. Speed is promoted by graphitizing the cementite at the highest possible temperature but to a certain extent at the ex- pense of quality. Temper carbon differs from graphite only in form. It has been pointed out that these differences of geo- metric form are due to the temperature of the metal in which the free carbon is formed. Accordingly, the two forms, temper and graphitic, shade over into each other by infinitesimal degrees and the temper carbon formed at high temperatures may grow so coarse and flaky as to be almost graphitic. Fig. 34 shows the carbon produced by graphitization at 2100 degrees Fahr. far above any commercially possible temperature. It will be seen that this carbon is purely graphitic and bears no resemblance to the temper form. Also the matrix of malleable iron is not a continuous mass but consists of an assemblage of individual grains as shown in Fig. 33. The character and size of this grain structure is influenced by changes of heat treatment, introducing another viewpoint for the selection of annealing temperatures. Moreover, high temperature may cause operating difficulties due to the deformation of castings, destruction of pots and fusing of packing material. An attempt to reduce the annealing period too far by a rise in temperature therefore is usually inadvisable. Commercial practice involves a maximum temperature of the castings of between 1500 and 1800 degrees Fahr. The time for maintaining the maximum temperature varies from 24 to American Malleable Cast Iron 60 hours, or even longer, the longer periods properly accompany- ing lower temperatures. The commercial rates of cooling are variable, ranging from 5 to 12 degrees per hour. In general the best practice is opposed to the highest tem- peratures, the minimum time of holding and the fastest cooling and favors a maximum temperature not far above 1600 degrees, a time not less than 40 hours near that temperature, and an aver- age cooling rate certainly not faster than 10 degrees per hour; preferably less, more particularly near the critical point. .When properly heat treated, malleable cast iron contains no combined carbon except just under the surface.- It is prac- tically impossible to entirely eliminate these last traces of pearlite from the casting, but this ingredient can and should be reduced to the point where it is equivalent to not more than 0.15 per cent of combined carbon as referred to the total weight of the casting. Approximately six years ago Thrasher published in graphic form the relation between carbon and silicon in while iron for constant tendencies to mottle. Based on the form of Thrasher's curves and known points near the middle of the range of composition for various classes of work, Fig. 27 has been prepared indicating the approximate relation between car- bon and silicon for various classes of work, based on the ten- dency to primary graphitization only. No attention has been given to the weakening effect of car- bon which sets limiting values on that element nor on pouring temperatures or other variables which may affect graphitization. The data presumably apply to ordinary air furnace practice and doubtless are subject to a certain amount of modification accord- ing to other variables. The more or less unavoidable oxidizing conditions in an- nealing remove some carbon from the surface. The extreme sur- face of malleable generally contains about 0.40 or 0.50 per cent combined carbon, while metal more than 0.1 -inch below the Metallography of Malleable Iron 69 surface is but little affected. Malleable castings sampled so as to include no material less than 1/8-inch below the surface will have nearly the ultimate composition of the original hard iron, except for the absence of combined and the presence of free carbon. If the sample is taken to include the entire cross sec- tion of metal the total carbon will vary with the thickness of Fig. 37 Changes of metallographic composition during the freezing and annealing of white iron the casting and will range from 0.40 per 'cent or even less up to the original carbon of the hard iron. In Fig. 37 the changes in carbon distribution during freezing of the hard iron and during its subsequent annealing are summarized in diagrammatic form. Time (estimated) is plotted as abscissae. At the top of the diagram the assumed temperature-time curve is plotted. At the bottom the relative weights of the various metallographic en- tities are recorded, the sum of course always being 100 per cent. Along the middle of the diagram the carbon concen- tration of the various homogeneous solutions (solid and liquid) is plotted for convenient reference. IV GENERAL MANUFACTURING AND PLANT TODAY all malleable foundries in the United States and Canada operate upon the same general principles although, of course, the manner of execution of the in- dividual operations varies with the ideas of the individual operator and the facilities at his disposal. Physically the foundries of the country differ widely both in size and type of buildings. The range in capacity of plants is probably from 50,000 tons per year down to 1000 tons or less. If plants making malleable only as a side issue are included, the minimum capacity is considerably less than 1000 tons. The plants range from antiquated structures of brick with low wood roofs to modern brick, concrete and steel buildings. A similar range exists in the facilities available in the form of mechanical equipment and, unfortunately, also in the personnel. It does not necessarily follow that the larg- est production is coupled with the best buildings, mechan- ism and talent although in this as in other industries, many things are possible for the large operator which are not a- vailable to the smaller. Large scale operations generally in- volve conditions better suited to the procurement of men and machinery of the highest order. In a previous chapter there have been outlined the prin- ciples upon which malleable cast iron depends for its proper- ties. It was there shown that the metal is the product of two distinct operations the making of castings of white iron and the malleablizing of the castings by a subsequent gra- phitizing or annealing process. This divides the process in- to two distinct stages and, generally, the plants into two separate parts the foundry and the annealing departments. Centered around each of these major departments are others of a contributing character such as the stockyard, mason's department, flask shop, patternshop, coreroom, melt- ing department, and chemical laboratory as foundry adjuncts 72 American Malleable Cast Iron bfi u> o bb General Manufacturing and Plant 73 and cleaning, trimming, inspection and shipping depart- ments, engineering and metallographic laboratories as ad- juncts to annealing. Plant maintenance also requires the operation of a power station, machine shop, electrical de- partment, etc. There are additional departments not direct- ly of a manufacturing character, including those pertaining to sales, purchase, accounting, labor, costs, first aid and others. The actual shop organization by which the departments are subdivided between groups of executives differs widely in different companies. Even the largest producer, operat- ing- seven malleable plants, finds it wise to use a somewhat different organization scheme in each of its foundries. Small plants usually are practically "one-man" shops. One executive, often the proprietor, exercises supervision over all works activities. The scheme is simple, but incapable of any very great growth. A common method is to divide the duties among three major foremen or superintendents. One has charge of the foundry and is responsible for everything up to the delivery of hard castings to the trimming room; another converts these into the finished product; and the third is in charge of power plant, carpenter, machine and pattern shops, etc. Sometimes the last two are co ordinated under one head, making only a foundry and finishing department. A much more highly organized and efficient system is represented in the organization chart shown in Fig. 38, which is applicable only to a fairly large organization and incidentally is not exactly followed in any plant of which the writer has knowledge. The raw material purchased by a malleable plant con- sists of pig iron and scrap as melting stock ; coal, coke and sometimes oil, gas and electric power as fuel ; molding and core sand for the foundry and refractories for the furnaces. In addition a wide variety of general supplies is used in more limited quantities. In almost all plants the melting operation is executed in air furnaces which generally make two heats a day. In 74 American Malleable Cast Iron K *i-i r General Manufacturing and Plant 75 some plants only one heat is made and in a very few two or three heats every other day and none on the intervening day. The latter practice is a survival of a practice in vogue 15 or 20 years ago. Heats vary considerably in size. On a two-heat a-day basis, they vary in different plants from seven tons to 24 tons each ; on a one-heat basis, from about 18 to 35 tons, and on a three-heat basis from five to 10 tons. In a few plants the cupola is employed 'for melting but this practice is not recommended for important work. Open- hearth melting has been tried by a number of producers and while not well adapted except to continuous operation and large tonnages is in successful use in a limited number of plants. A few small furnaces each having a capacity of about five tons are said to have been tried. The charge in most successful open-hearth installations averages from 14 to 20 tons. A single producer operates electric furnaces at two different plants. From 10 to 12 heats and even more when molds are available, are 'made in 24 hours, but the metal is delivered to different molders so that generally a given molder only pours off twice per shift. In these plants heats range from five to seven tons and from eight to fifteen tons in weight depending on furnace capacity. Six and twelve tons are the nominal furnace capacities. Molding still is done by hand in many shops as it was in all plants 15 years ago. The patterns being small, many are mounted on a single gate. The pattern is pro- vided with a match part and the mold made in a snap flask. Hand operated squeezers 'have been in use for many years, the air-operated devices apparently not having met with general favor, although used in some plants. Recently the trend has been strongly toward patterns mounted on plates and vibrated by air when the cope is being lifted or the pattern drawn. In many localities no labor now is available capable of commercially producing molds fro'tn other than plate pat- terns. Consequently this form of mounting which requires less skill of the molder than any other, is practically forced 76 Malleable Cast Iron be General Manufacturing and Plant 77 on the industry. Nearly all of the more complicated me- chanical devices have been tried, but so far they are not used extensively except for floor work, in which case various types of roll-over, roll-over drop and stripper plate ma- chines are successfully employed. As already stated the stna-le; molds are usually made in- snap flasks. Sometimes, when there is clanger of breaking out when pouring, the molds are strengthened with mold bands of strap ircn. 1 he use of jackets to prevent break- outs also is prevalent. The larger molds are made in box flasks, iron flasks being very common and desirable for use on machines. All malleable castings are made in green sand except for cored holes. Since only relatively unskilled help is avail- able, the use of three-part or other multiple-part flasks and loose pieces on patterns is practically impossible. Any pat- tern equipment which cannot be drawn straight out or rolled out on a flask hinge is incapable of quantity production under the conditions existing in most foundry centers. Cores generally are made of local sharp or lake sands using rosin, oil or some of the wood sugars as binders. As a rule, the work is of such character that large and complex cores are not required. A few foundries are beginning to prepare and deliver molding sand by mechanical means. One device for cutting sand on the floor is coming into fairly extended use, since human sand cutters are no longer available. Molds are commonly set on the floor by hand, although at least two semi-automatic devices for removing' molds have been tried, one of which offers prospects of successful oper- ation. No methods of molding, involving successive operations by a number of workers, have proved entirely successful thus far. Pouring is done either from hand ladles or from shank or "bull" ladles handled by two men, the former being more common. In cupola or air furnace practice molders catch directly from the stream as it flows from the furnace, the* tap hole being only infrequently closed by a clay stopper or iron bar. 78 American Malleable Cast Iron l 1'a o ^ I O rt en o c/3 ^ M f f course, very widely, depending on character of work, plant layout and so on, but the following table may be regarded as suggestive at least of the labor consumed in handling material for . production of one ton of castings : Table I MATERIAL HANDLED TO PRODUCE 1 TON OF CASTINGS No. of Tons of times Total tons material handled handled Melting stock 2.2 3 6.6 Molten metal 2.0 3 6.0 Sprue 1.0 3 3.0 Slag 1 6 .6 Castings 1.0 19 19.0 New molding sand 35 2 .7 Used molding sand 5.0 25.0 Core materials .25 10 2.5 Fuel 1.2 3 5.1 Cinders 175 2 .35 Annealing pots 1.5 6 9.0 Packing 5 2.5 Refractories 15 6 .9 ll925 81.25 Add 1/3 for handling supplies and equipment 27.08 108.33 The items in the above table are based entirely upon estimates. The writer knows of no attempt to actually de- termine the several items. Also, evidently the expense of handling a ton of material can have no unit cost assigned, for the term "handling" may mean picking up the material and transporting it by a crane; picking it up to inspect, piece by piece, or the laborious operation of firing a ton of 86 American Malleable Cast Iron bfl General Manufacturing and Plant 87 coal in an air furnace, or wheeling a ton of sand a consid- erable distance by hand. The table is here presented pri- marily .to show the importance of reducing the number of handlings each material undergoes and facilitating each by every available means. The 'history of labor in the malleable industry has been that of labor in all similar work. In the early days the workers were practically native Americans, supplemented by thoroughly Americanized English, Irish, Germans and Scan- dinavians. Later the two latter groups increased consider- ably, and still later toward the end of the last century the influx of Balkan immigration began. The native American and the original foreign 'groups meanwhile drifted almost entirely out of the labor and molding groups, though a few remain principally in the coremakers' trade. Most of these men and their sons headed toward the machinists, carpen- ters and patternmakers' trades, or toward other employment of similar character but requiring less skill. Type of Workmen Available Meanwhile the Hungarians, Bohemians, Poles and Aus- trian-Slavs began as laborers and gradually worked upward through the various grades of skill, being supplanted in the lower grades by Italians and later by Bulgarians, Greeks and Russians, and still later by Turks and Armenians. In some few plants the negro long has been employed in all but the highest skilled trade 1 - and the northward migration of the southern negro farm laborer is rapidly enlarging this condi- tion. Postwar developments meanwhile are making for the return of many former subjects of Austro-Hungary, Bulgaria and Russia to their native lands. He were a rash prophet who would attempt to discuss the net effect on the Amer- ican labor market of this emigration, the European tendency toward immigration to America, the discontent of those who returned to Europe, the industrial stagnation of Austria and Russia, all in the light of the American immigration laws and shipping facilities. Natural clannishness of foreign races has produced a segregation of nationalties in different parts 88 American Malleable Cast Iron of the country. The lines of course are not rigidly drawn but the Scandinavian still persists in (the northwest and to some degree in the St. Louis district. In the terri- tory extending from St. Louis to Terre Haute the Armenian is relatively prevalent ; the Russian and the Pole have set- tled in the Chicago district, as also the older class of Bo- hemians. The region around Indianapolis is manned by Greek, Bulgarian and Austrian-Slav foundry workers, while in northern Ohio Poles, Bohemians and Italians dominate. The latter element is very prevalent through the Pittsburgh district and through the Shenango and Mahoning valleys. New England and New York, being gateways to the in- terior, probably have a more mixed population than the Mid- dle West. Foundries of all kinds have been confronted with these conditions : First, a growing disinclination on the part of all labor to do foundry work; second, a trend toward less and less skilled and intelligent help; third, a more and more turbulent character of help from which the required force must be recruited. The trend toward negro labor repre- sents a turn in the tide at least in the latter respect. The industry is confronted with growing labor problems the solution) of which requires the best efforts of its ablest executives. These efforts will have to continue for a long time to come in order that the decreasing productivity of labor may be prevented from being reflected in the product in the form of prohibitive rates. The solution is in the utilization of mechanical aids to the utmost and in an enlightened labor policy. Metallurgy of Malleable Is Complicated Furthermore the malleable process is metallurgically more complicated' than that of either gray iron or steel foundry practice, and the chemical range consistent with good results is smaller than in the former. The most successful means of overcoming these handi- caps in manufacturing cost is to operate upon a sufficiently large scale and on more or less specialized products in order General Manufacturing and Plant 89 to take advantage of those manufacturing economies asso- ciated with such production methods. By inference the malleable industry is not well fitted for the manufacture of so-called short orders, that is, orders involving only a few pieces from a given pattern and small tonnages for a given consumer. It attains its greatest suc- cess when operating on orders of sufficient magnitude for each type of casting to warrant investment in the best possible pattern equipment and close study of each step in the manufacturing. V MELTING STOCK THE raw materials of the malleable industry may be classified as melting stock, fuel and refractories. The remaining materials are not peculiar to the malleable industry and therefore are not important in the present dis- cussion. Regardless of what melting process is employed in making malleable, the melting stock is selected from the same general classes of material. Sprue, which includes the feeders, runners and defective castings produced inci- dentally to the plant operation, is seldom if ever sold and never is bought by a malleable foundry. Being a product of the foundry-man's own plant, its composition and condi- tion are known to him and the material requires no "further description. Malleable scrap is a material derived in part from the work condemned at the plant after annealing. Also it is an article of commerce in the form of scrap material con- sisting of worn out malleable parts. The* scrap yard of a malleable foundry is shown in Fig. 46. Scrap has been some- what roughly divided into "railroad malleable" and "agri- cultural malleable." The distinction is actually one based on size of castings rather than on the former use. "Auto- mobile malleable" is regarded by some users as a legiti- mate subdivision but really does not differ materially from the railway malleable scrap from a metallurgical standpoint. Pipe fittings, often classed separately, could equally well be included with agricultural malleable scrap. The composition of purchased malleable of course is entirely conjectural and there is therefore a limit beyond which its use introduces serious uncertainties as to compo- sition of charge. It is safe to assume that railway and auto- mobile malleable, before annealing had a carbon content 92 American Malleable Cast Iron i Melting Stock 93 averaging about 2.50 per cent. No two pieces are alike in carbon, depending both on the original carbon and the de- gree of decarburization in the anneal, but the remaining car- bon in work of these heavier classes is likely to be around 2.00 per cent or a little under. The silicon is likely to av- erage around 0.70 per cent and in malleable scrap consist- ing of castings worn out in service the sulphur is from 0.06 to 0.10, the manganese 0.25 to 0.35 and the phosphorus from 0.16 to 0.20 per cent. In the case of agricultural and other light work, the initial carbon may have been considerably higher, but in view of the lightness of cross section this ele- ment may have been much reduced, possibly to 1 per cent and under. The silicon generally is somewhat higher than in the heavier materials, usually averaging about 85 per cent. The other elements are about as in railway malleable. Malleable scrap is open to the objection that when used as a considerable percentage of the mix in air furnace or open-hearth practice, serious errors may be introduced in the chemical composition of the charge. This condition is ag- gravated if the malleable scrap includes gray iron scrap rich in carbon, silicon and phosphorus. It is a most reprehen- sible practice of a number of junk dealers either to purposely mix or to not properly separate the two materials, thus practically destroying the value of the malleable scrap to the malleable founder. This separation can be readily made only 'at the point of origin as the user has no commercially effective method of inspection. Equally harmful in the op- posite direction is the admixture 'of steel. Another source of -trouble is the introduction of un- known amounts of rust into the charge when melting scrap that has been exposed to weather. Some scrap may con- tain 5 per cent or more of rust which of course is a dead loss in melting. It also forms a highly oxidizing slag which in turn strongly acts on the silicon and carbon causing unpre- dictable changes of composition in melting. The -effects of this evil can be minimized by the use of clean scrap, which unfortunately cannot be purchased and by the purchase of 94 American Malleable Cast Iron scrap of such form that it presents little surface to rusting. For this reason and because of the high labor cost of handling small scrap, agricultural material is not a satis- factory melting stock in air furnace or open-hearth mal- leable practice. Heavy malleable scrap stored out doors but not extremely heavily rusted usually behaves as though it contained about 1.75 per cent carbon and 0.47 per cent silicon. The presence of adulterations, except of high phos- phorus material, is of less consequence in electric furnace melting than with air furnaces or open 'hearths. Malleable scrap is used not because it is a means of cheapening the metal but for the definite purpose of regulating the carbon content of the mix. Successful air furnace practice requires a c'harge averaging around 3 per cent in carbon, hence some low carbon stock must be used to mix with pig iron which is always of much 'higher percentage of carbon content. Sprue is available in a quantity dependent on the found- ry practice but not usually sufficient to bring down the car- bon as far as necessary. Hence recourse is had to mal- leable or steel scrap. The use of scrap for the purpose of making up different 'amounts of sprue has been practiced for more than 30 years. The Chicago Malleable Iron Works has purchased scrap for air furnace charges on a commer- cial scale since 1885 and in 1888 the practice was well es- tablis'hed. Possibly others adopted it still earlier. Steel scrap is an article of commerce. What has been said of -malleable regarding freedom from rust and from ad- mixture of other forms of scrap applies equally well to steel. In addition there is a certain danger from the possible pres- ence of alloy steels which may introduce entirely unexpected elements. A case in point is the high manganese steel used in frogs, switch points and cross overs and containing up to about 13 per cent manganese. The carbon content of all steels is relatively low, rang- ing from around 0.90 to 1.00 per cent in some spring steels down to 0.25 or 0.30 per cent in castings. The silicon is always low and the manganese averages around 0.50 or 0.60 Melting Stock 95 per cent. The sulphur and phosphorus values are always lower than in any other ingredient in the charge. Consid- ering the fact that the material is always somewhat rusty it may be classed as pure iron in calculating a mix. Heavy steel scrap is preferable to the lighter material as is the case with malleable scrap. Thin sheet, small clip- pings, rods, pipe and light structural material are particu- larly objectionable when rusty or burned. Steel, as in the case of malleable scrap, is used to reduce the carbon content of the mix. Being lower in carbon a less percentage suffices for a given purpose ; therefore there is less danger of introducing large errors of calculation in , computing the mix or of large amounts of rust to compli- cate the reactions. Steel is rarely used in making cupola or electric furnace malleable. Its general use was adopted more recently than that of malleable scrap, but the old records of the Indian- apolis plant of the National Malleable Castings Co. show that for an extensive period, beginning in August, 1887, steel was regularly vised in the mix, and that the practice continued as circumstances warranted. The author has no facts to in- dicate whether this practice was original with the late James Goodlet, then in charge thei;e, or copied from some other plant. Wrought iron, which chemically is merely an extremely low carbon steel, was used at the inception of the industry, Boyden referring to it in his notes. At a later date it was regarded as harmful arid at present it is not available in sufficient quantity to possess interest. Pig iron is the raw material which makes up the bulk of the tonnage from which malleable cast iron is made. In the days of the fathers of the industry charcoal iron was generally if not universally used. Then, as now, it was made from relatively low phosphorus ores. In the early days, before the Civil war, the references are mostly to irons smelted in New Jersey arid Connecticut from eastern ores using charcoal from local forests. Bovden used such 96 American Malleable Cast Iron Melting Stock 97 irons. Alfred Hammer used New Jersey coke arid anthra- cite pig as early as 1878. In about 1885 there was 'a no- ticeable trend toward the use of coke-melted pig iron, first, as far as the author can judge, in the case of very soft pig iron. This was high in silicon, and was unusual in furnaces operating as cold as did the usual cold blast charcoal fur- naces of the period. The impression is quite general among the older found- rymen that, apart from differences of composition, there are differences in properties as between the products of different furnaces. Many also believe that it is preferable to use iron from several producers in each heat. It is not clear to the author upon what metallurgical considerations such differences could be based. Undoubtedly before the days of analyzed pig iron, these beliefs were based on sound reason; at present they would seem to be little more than prejudice as applying to malleable practice. A similar situation is encountered in a somewhat gen- eral feeling that the use of malleable scrap is in some way connected with the substitution of coke for charcoal pig. It has been only relatively recently that interest in the control of the product by limiting the total carbon content became at all general. Dr. Moldenke in his book, "The Production of Malleable Castings" (1911), recommends for instance that the carbon be not below 2.75 per cent and may range up to 3 per cent. Presumably this represents the best general understanding of the time. While since 1906 certain manufacturers realized the relation between carbon and strength and acted on this knowledge, it is not sur- prising that in the days when the substitution of coke for charcoal iron began the mixes used never were based on considerations of carbon content. With low silicon charcoal iron available it was easy to secure silicons low enough to produce a white fracture by the use of pig and sprue alone. Hot blast coke irons always contained enough silicon so that some material other than the available amount of sprue was required to reduce the 98 American Malleable Cast Iron Tf bb Melting Stock . 99 silicon content sufficiently to avoid "mottled" castings. The effect of this change of practice on carbon content was totally disregarded except by a very few observers. The general observation that charcoal iron malleable could and should be made lower in silicon than malleable for the same purpose made from coke iron probably was true. However, it originated merely from the reduction in carbon which unconsciously accompanied the changed prac- tice and not from the method of making the pig. Where malleable was made from charcoal .and coke iron of the same silicon content the former was somewhat the stronger, due to its somewhat lower carbon content, which in turn was due to lower furnace temperature. In view of such former experiences great caution should be used in regarding as cause and effect phenomena without apparent logical connection. The transition from charcoal to coke iron has extended over many years and is not yet complete. In the early ninety's coke iron was used very sparingly, but 10 years later the coke iron was far in the ascendant. At present comparatively few manufacturers continue the use of char- coal pig and they employ it only in limited quantity. To the writer it has seemed that this retention of char- coal iron results either from sentiment pure and simple or from a superstitious belief that for some unexplained rea- son a modicum of charcoal pig imparts a mysterious virtue of unknown character to the resulting product. Being smelted at a lower temperature, charcoal iron differs from coke iron in being generally lower in carbon. On account of the low sulphur fuel, it is always lower in sulphur. Also the range of silicon values commonly available run lower in charcoal than in coke iron. Again, this is the result of the furnace temperature. The lowest silicon charcoal pig irons commercially made contain less silicon than the lowest silicon grades of coke iron. Moreover, high silicon coke iron is more com- monly obtainable than charcoal iron with the same con- 100 American Malleable Cast Iron be c '-5. rt O Melting' S-otk : 101 tent, in spite of the fact that the "Scotch" grades of charcoal pig have a high silicon content. The writer has never been able to see any theoretical reason why charcoal iron should make a better product than coke iron, given a correct final composition. The late J. B. Johnson Jr., who dealt at length with the subject from the blast furnace viewpoint, ascribed the differences to the in- direct effect of oxygen. For the best available opinions in this subject, the interested reader is referred to the pub- lished reports on Johnson's pioneer w r ork on this subject in the Transactions of the American Institute of Mining and Metallurgical Engineers. In view of the radical alterations made in the raw material during the malleable process it is difficult to see how any differences, such as the form of crystallization of graphite in the pig iron, could survive the chemical and physical changes involved. The trade as a whole seems to look upon the matter in this light and from a tonnage viewpoint, charcoal iron is of little importance in the malleable industry. The production of malleable cast iron requires the use of relatively low phosphorus ores, those of the Lake Superior region being the most available for the purpose. Conse- quently, many of the blast furnaces producing malleable pig are situated along the lake ports. The proximity to the Pennsylvania coal fields producing coking coals, has formed another area extending from Pittsburgh down the Ohio river and up the Mahoning and Shenango valleys. The charcoal furnaces are located near the ore fields in heavily wooded districts. The ore fields of Minnesota, Wisconsin and north- ern Michigan are shown in the form of a shaded area in Fig. 47. Immediately adjacent to this section are the prin- cipal charcoal furnace plants, shown on the map by open circles. The coke furnace plants are shown as solid circles. Most blast furnaces do not make pig iron for one purpose only, but the map is intended to include all important pro- ducers of this class of metal in considerable quantities. An open pit mine on the Mesabi range "is shown in Fig. 48. 102 Matictiblc Cast Iron Melting Stock 103 The ores from which malleable pig iron is made 'have ap- proximately the following composition : Per cent Fe 51.5, present as Fe 3 O 3 . . 75.57 P .086, present as P 2 O 5 19 Mn. .40 to .70, present as MnO ' 77* Si0 2 9.50 A1 2 3 2.75 CaO 70 MgO bO H 2 O, CO 2 and undetermined 10.02 ^Average. Malleable pig iron is sold with ti guaranteed maximum of 0.05 per cent in sulphur, usually of either 0!19 or 0.20 per cent in phosphorus and is furnished with from about 1.00 to 2.00 per cent silicon, although 'higher values are sometimes required. The manganese varies from about 0.50 to .about 0.90 per cent, the lower and higher values being encountered frequently. The average carbon content for the country is now and lias been for at least 15 years close to 4.10 per cent, individual lots running normally from 3.85 to 4.40 per cent, 'i he carbon content practically is fixed by the blast furnace temperature. Pig iron may be either sand, chill or machine cast. The former 1 carries with it a certain amount of sand fused into the surface. The chill and machine cast irons are free from this foreign matter, which fact presents a certain advantage both because nothing but iron is paid for and because less dirt is carried into the furnaces. The two latter classes, being rapidly cooled, contain more combined and less free carbon than the former, other things being equal. The melting point and, presumably, the latent heat of fusion are thereby decreased. It is claimed that a material fuel econ- omy results. On all accounts the use of machine cast iron can present no disadvantages to compensate for the advan- tages outlined above and its greater uniformity of size and form. Recently there his been a decided tendency toward changes in chemical composition of commercial pig iron. Up to 1914 the sulphur content, while guaranteed as 0.05, was nearly invariably under 0.03 in the Ohio and Illinois 104 American Malleable Cast Iron Melting Stock 105 irons. Since then fuel conditions have so far deteriorated the quality of coke available that at present sulphur is usually only a little under 0.05 per cent and occasionally exceeds that figure. Ten or 15 years ago iron often was sold with a maximum phosphorus of 0.16 per cent, no extra price be- ing charged as compared with a 0.19 or 0.20 per cent maxi- mum specification. The gradual increase in the ratio of phosphorus to iron in the product of the Mesabi ore fields has, however, forced an increase to the latter figures as a phosphorus maximum. For about five or six years there has been a decided trend toward lower carbon malleable, brought about by the demand for increased quality of product. This results in lower percentages of pig iron in the mixes than formerly and therefore requires increasingly a higher silicon content to maintain the former silicon values in the product and in some cases raise them -slightly. Accordingly the metal con- taining under 1.25 per cent silicon is now almost useless and most -plants require some pig iron up to 2 per cent and possibly over in silicon. The average silicon content in all the pig iron consumed in the malleable industry is doubtless between 1.60 and 1.70 per cent. There seems to be increasing difficulty in getting any low manganese pig. However, this stringency has been somewhat counteracted by the decreased amount of pig re- quired and the increased sulphur content. Coke pig iron under 1 per cent in silicon and usually high in sulphur, is generally the product of an abnormal furnace condition, re- sulting in cold working and is not of a composition suitable to modern requirements. High, silicon pig, or blast furnace ferrosilicon is a metal usually running about 10 per cent in silicon. Its principal source is Jackson, O. The phosphorus, sulphur and carbon are kept low. The metal is used as a source of silicon when suitable pig is not available. In the electric furnace process, it may furnish most of the silicon of the cupola charge. Ferromanganese is a blast furnace product made from manganese ores. It usually contains from 70 to 85 man- ganese and nearly 6 per cent carbon. Silicon, sulphur and 106 American Malleable Cast Iron be Melting Stock 107 phosphorus are low, iron being the principal element, other than manganese and carbon. Ferromanganese is used gen- erally in -the form of an addition to the molten metal to supply a deficiency in manganese. Electric furnace ferrosilicon contains nominally 15, 50, 75 and 95 per cent silicon. The 50 per cent alloy, actually running from 48 to 54 per cent silicon, is most commonly used. In addition to silicon and iron the metal contains phosphorus, sulphur, aluminum and calcium. These elements are not usually present in important amounts. Ferrosilicon, being readily oxidized, is not suitable for cupola use. When charged into an air furnace with the melting stock it must be protected from contact with fur- nace gases as far as possible. It is generally used as addi- tions to the molten heat. VI FUEL AND REFRACTORIES THUS far we have dealt with the raw material actually entering the product. There remain two other classes of raw materials which, although they form no part of the finished product, are used in such quantities and so affect the shop operation as to be of decided industrial importance. The first of these groups is fuel. The fuels used in the malleable industry may be classi- fied as melting fuel, annealing fuel and power plant fuel. The latter, although it may be used in large quantites, as in electric furnace plants, should be considered from the viewpoint of power plant practice rather than from a metallurgical angle. Melting fuels not only furnish heat but also very distinctly affect the composition of the resulting product. On the other hand, annealing fuels need be consid- ered only from the standpoint of combustion. The original source of almost all the heat used in melting and annealing malleable is coal, although it may be convert- ed before use into coke, illuminating gas, water gas, or producers gas. Oil and natural gas are also industrially im- portant in some localities and for some purposes. Bituminous coal is very widely distributed throughout the country, as indicated in Fig. 53. Anthracite and lignite are not important metallurgical fuels and are therefore omit- ted from the map. Anthracite was formerly used as a cupola fuel and at an early date, possibly 1838 was used for anneal- ing by Belcher. It is still used in at least one plant for this purpose. Coal from practically any of the bituminous fields shown may be used for annealing, the choice generally being based on geographic and commercial considerations rather than on the properties of the fuel from any given field. Mine run fuel is generally used in annealing for hand firing. The crite- rion of quality is the absence of ash and water, these fac- 110 American Malleable Cast Iron bo _c 5 o 02 C/3 ex n5 I Fuel and Refractories 111 tors representing increased cost and operating trouble and not metallurgical suitability. -A low ash fuel is sometimes preferred for use with pulverized fuel annealing equipment in order to avoid trouble from the ash settling in the fur- naces and flues. The requirements of a crushing plant prac- tically necessitate a fuel either quite dry as received or dried artificially before crushing. - Since lump coal is of no advantage, pulverized fuel plants buy the smaller commercial sizes of fuels. However, the selection of fuel for crushing in annealing practice is not well standardized. The author knows of two large and ably managed plants within a few miles of each other, both an- nealing with pulverized fuel. One buys a high ash local coal and removes about 10 to 12 per cent of water by drying be- fore crushing, while the other obtains coal in the eastern fields hundred of miles distant which runs under 2 per cent in water. and around 3 to 4 per cent in ash. The subject of coal for 'annealing is therefore easily dismissed with the statement that practically any local fuel can be employed, economical conditions alone governing the selection. In the case of the melting coals conditions are quite different. Here, in addition to the purely economic prob- lems, there enter many other considerations which narrow down the choice. Coal burned in the air furnace is expected to furnish heat units as economically as may be practicable, and must have certain other definite characteristics. It must burn with a long luminous flame jof sufficient volume to en- tirely fill the air furnace. It must be so low in sulp'hur as not to prohibitively raise the content of- that element in the met- al. Its character must be such that none of the constituents will melt and run to a tarry mass at the temperature of the fire. Its ash must be fairly low in amount and of such char- acter as not to fuse together into clinker's at fire ibox tem- peratures. Its moisture content must -be reasonably low in order to maintain good flame conditions. These characteristics are found in coal from a very limit- ed geographical area which is shown in black in Fig. 53. 112 American Malleable Cast Iron bo 'Fuel and Refractories 113 In the writer's experience the fuel varies even within this district, 'being in general better in the southern portion of the area. No entirely satisfactory method of judging the quantity of a melting coal, except by actual test is available. This arises, in part, from the fact that the behavior of the fuel is dependent on the actual combustion conditions encountered which differ with different furnaces. The composition of a few good melting fuels is shown in Table II. Table II ANALYSES OF MELTING GOALS Origin Sulphur Moisture Pennsylvania 0.70 062 West Virginia 0.45 0.76 West Virginia 1.55 1.34 Kentucky 0.45 1.10 Vol. Comb. . 35.63 37.15 41.70 33.95 Fixed carbon . . . Ash , . . . 58.32 5.43 55.64 6.45 52.40 4.56 60.68 4.27 B.t.u. per pound.... 13,902 13,434 14,058 14,276 There is a general preference for coal under 1 per cent sulphur, although the sulphur which the melt takes up de- pends not only on the sulphur content of the fuel but also on the form in which it is present. Some fuels, moderately high in sulphur, produce metal lower in sulphur than other fuels, much lower in that element. Many coals exist, even some in the Illinois, Indiana and central Kentucky fields which based on composition should work admirably. The expectation, however, is not borne out in practice. What makes a long flame coal has never been definitely determined. The flaming coals are in general the coals best adapted to making illuminating gas. The flaming quality is associated with the distillation products of the fuel when heated in the fire box. The running of the coal is a phenome- non of the same character. The low moisture seems to be a necessary characteristic. Goals of this character artificial- ly wetted behave differently from the naturally wetter In- diana-Illinois fuels. The clinkering of the ash is largely a matter of chemical composition. Strictly speaking it depends on the com- 114 American Malleable Cast Iron Fig. 55 Picking table in a coal tipple, showing facilities for removing slate, sulphur, etc., by hand. Fig. 56 Adjustable loading boom which places coal in car without breakage Fuel and Refractories 115 pounds formed in the ash under the temperature and chem- ical conditions existing in the fuel bed. Therefore, analyses made on laboratory preparations of as-h are not correct state- ments of what may happen in the fuel bed, but are of some value as indicating what may be expected. An ash of a very satisfactory fuel had the following composition : SiO 2 Per Cent 44 52 A1 2 O, 43 75 Fe.O, 1 32 CaO 5 72 MgO . . . . 1 05 Na 2 O ) K ? O . ( 3.64 The analysis is of a laboratory preparation of the ash. On the grates the Fe,O 3 would be largely reduced to FeO. The fusing point of the ash of eastern coals is 2400 to 2850 degrees Fahr. Above 2600 degrees Fahr. is preferable. In general, the absence of iron oxide, alkalies and lime in the order given is considered a desideratum. The ash and sulphur contents of coal are considerably affected by the method of preparation and in recent years mining conditions have been such as to make for a steady deterioration along these lines. Air furnaces require a lump coal for their fuel 'but com- mercial practice varies as to the size of screen over which the coal should be passed before shipment. Some foundry- men desire coal not finer than that which will not pass a 4-inch mesh, while others tolerate all that will pass over a ^4-inch screen. The beslt practice probably is a little nearer the latter figure than the former say about 1^2-inch screened lump. When fuel is to be burned in pulverized form in melt- ing furnaces the quality of coal required is the sam'e as for direct combustion on the grates, except that the smaller sizes of coal can be utilized. A number of engineering concerns have developed highly specialized plants for grinding and pulverizing coal. The sequence of operations in all of them is substantially the 116 American Malleable Cast Iron d o be rt be S3 o be Fuel and Refractories 117 same. The coal, crus'hed to fairly small size or purchased after screening, passes through a device where it is dried by a current of warm air. A favorite method is to feed it in at one end of a rather long narrow cylinder rotating on its axis, whic'h is slightly inclined to the horizontal. As the cylinder revolves the coal rolls over and over and travels toward and finally out of the lower end of the cylinder. A current of warm air passes through the cylinder, usually in the direction opposite the flow of coal. From the end of the dryer the coal is automatically de- livered to a grinder, one type of Whic'h consists of an ar- rangement like the "fly balP' or centrifugal governor of a steam engine. The weights are in the form of rollers whic'h run against a surrounding ring when the mechanism is ro- tated. The fuel is ground to flour between these rollers 'and the ring, but if any hard lump such as a piece of scrap iron should fail to have been removed it merely crowds through between the roller and track and does no damage. Means are usually provided for screening or otherwise separating insufficiently ground material and returning it to be reground.- The product should be reground to pass a 100- mesh sieve and 75 per cent to pass a 200-mesh sieve^ When ground to size it is transported by belt or screw r <5onveyor to bins. A pulverizing plant is shown in Fig. 92. In -general it is well to store only limited qu#fitities of ground coal- -'owing to the fire hazards. Dried pulverized coal absorbs moisture readily, and sticks together and^feeds to the 'burner in a lump condition if it has an opportunity to take up water before being used. The transportation of coal dust by carrying it in a cur- rent of air is dangero ; us, the mixture being highly explosive. In the best installations the air and coal are mixed just as near the point of fen fry- into the furnace as possible to min- imize the danger. Gas as a fuel is only an indirect application of the com- bustion of coal, indeed it might well be maintained that any use of coal for this purpose involves its gasification even though that process may be carried out in the fire box in- stead of in a separate apparatus. 118 American Malleable Cast Iron be Fuel and Refractories 119 Gas fuels are classified as illuminating gas or producer gas. The former is either a distillation product of coal, or a mixture of hydrogen, carbon monoxide and hydrocarbons, called water gas and made by the action of steam on red hot coke. Producer gas is a mixture of carbon monoxide and hydrogen. Illuminating gas is too costly for extensive metallurgical operations. Its use is limited to crucible furnaces for brass melting, etc., and small core ovens. If the gas is a by-prod- uct in the manufacture of coke, it is commercially available and then only in the plant operating the coke ovens or in neighboring plants. If the gas is to 'be piped any distance it can generally be more profitably sold for public consump- tion for domestic requirements. The operation of a gas producer is simple in principle. A gas producer is merely a firebox in which a deep bed of fuel is burned with a limited supply of air, the intention being to burn the carbon of the fuel to carbon monoxide. Theoretically, the producer gas is air in which the oxy- gen has been converted to carbon monoxide and should con- tain about one-third carbon monoxide and two-thirds nitro- gen. In practice the water from the combustion of the hy- drogen of the fuel, the moisture of the fuel itself and the steam which is introduced with the air supply to avoid clinkering all react with carbom, liberating some hydrogen. Also the fuels rich in volatile matter distill off more or less hydrocarbon gases. Furthermore, if the fuel bed is allowed to get uneven permitting air to come through, some of the carbon monoxide is burned to dioxide. The latter constituent is more prevalent in producers blown with steam than in those blown with air alone. As a general statement of the composition of commer- cial producer gas, the following figures are quoted from Wyer: Table III COMPOSITION OF PRODUCER GAS H CH 4 C 2 H 4 N CO O C6 a Gas from hard coal 20.0 .. .. 49.5 25.0 0.5 5.0 Gas from soft coal 10.0 3.0 0.5 58.0 23.0 0.5 5.0 Gas from coke 10.0 .. .. 56.0 29.0 0.5 4.5 Gas air blast 4.43 .. . . 62.12 33.04 .. 0.41 Gas same as above with air and steam blast . . 14.00 53.3 27.2 5.5 120 American Malleable Cast Iron The CO 2 values are rather high, an attempt usually be- ing made to hold CO 2 to 3 per cent. It is obvious that the 'heat value of the gas from a pound of coal cannot be greater than the heat value of the original Fig. 59 Cross section of a modern gas producer pound of fuel. The combustion of carbon to CO liberates 4450 or roughly 30 per cent of the heat of combustion of carbon to CO 2 . This heat is transmitted to the incoming fuel and to the products of combustion as well as to the producer structure. It finally leaves the producer by radia- Fuel and Refractories 121 tion from the walls and also as the sensible heat of the gas. it therefore is of advantage, except in open hearth practice, to make the gas as near the furnace as possible to avoid the loss of 'heat units by coo. ing the gas stream in passing through long ducts. Where the gas is to be widely distributed or burned in small accurately controlled burners a cleaned gas from which tar and heavy hydrocarbons have been removed is desirable. As stated before, gasification adds nothing to the heat value of the fuel ; it may, however, result in heat economy due to the better control and more economical combustion conditions possible with gas fuel as compared with solid fuels. Producer gas being a fuel of rather low calorific power usually is burned with hot air. The use of cold air does not give sufficiently hot flames for melting operations ; in- deed the temperature m'ay not be high enough to maintain combustion unless a warm or hot air supply is provided or the gas itself be fairly hot. Producer gas usually is made from bituminous coal, al- though wood, peat, lignite, coke and anthracite can be used. The requirements for producer gas fuel in general are simi- lar to those for >air furnace fuel. The coal should be rea- sonably low in ash and the ash should not clinker. The coal must not soften or swell on heating and preferably should be low in moisture and high in volatile matter. Fur- ther, it should be fairly uniform in size and, for melting op- erations, low in sulphur. However, there are many bitumin- ous coals giving good results in producers which do not work satisfactorily in the air furnace. Coke is used as ; a metallurgical fuel in the malleable industry in cupola practice only. As everyone knows, it is gas 6r similar coal from which the volatile matter, includ- ing moisture, has been distilled in retorts, beehive ovens or by-product ovens. It contains all the ash in the coal from which it was made and is therefore from 50 to 100 per cent higher in ash than gas coals. The remainder of the coke is practically pure carbon. All coke contains sulphur and there is a general feeling 1 in favor O f foundry cokes con- taining less than 1 per cent of this element. Sulphur is 122 American Malleable Cast Iron Fuel and Refractories 123 taken up by the metal more readily in cupola practice than in the air furnace, owing to the fact that fuel and metal come into actual contact with each 'other. Moreover coke must not be too fine and must be fairly strong to make a suitable fuel. The ash should be as low as practicable and, if possible, siliceous in character, since it is easier to add basic materials to flux with the ash than to add acid 'materials. The as'h is similar in composition to that of coal and corresponds to low grade fire clay. Cupola fuel is not of great interest hi this discussion, owing to the general aban- donment of cupola malleable. In the case of electric fur- nace practice in which cupola metal is the raw material for the electric furnaces it is, of course, an important material. Oil is found rather widely distributed throughout the country. Fig. 53 shows the oil areas, exclusive of oil shales. Oil has many advantages as a fuel, including cleanliness, rel- ative freedom from sulphur, convenience of distribution and accuracy of control of combustion conditions. Twenty or 30 years ago it was customary to burn local crude oils just as they came from the ground. The need for gasoline and lubricating oils has caused the abandonment of this practice and today the fuel oil used consists of the residue remaining after the distillation of the commercially important products. Nearly all the petroleum products are hydrocarbons of the methane series having the general formula C r H 2I1 + 2 All have nearly the same 'heat value per pound, because n being a fairly large quantity, the atomic ratio of carbon to hydrogen is in all of them very nearly 1 to 2 corresponding to a ratio by weight of 6 to 1. The more volatile com- pounds such as gasoline, kerosene, etc., are the members of low molecular weight in which n is from 5 up. Fuel oil has been applied to 'annealing furnaces very conveniently. It is a useful fuel in open-hearth practice and has been successfully used in that connection in the malle- able industry. Under favorable circumstances it can some- times compete for this purpose with producer gas -and pul- verized coal. Furthermore, it is easy to arrange open-hearths 124 American Malleable Cast -Iron to permit the use of either oil or gas or oil or pulverized coal, which is a convenient arrangement. Attempts have been made to burn fuel oil in air fur- naces. No particular difficulty exists in actually doing the melting, but generally the process has not been either eco- nomically or metallurgically successful. J. P. Pero reports* what he regards as satisfactory results at an Illinois plant, but even there it is admitted that excessive oxidization losses were not overcome and 'the fuel cost was high. A plant in Michigan is said to 'have operated successfully with oil melt- ing, even at a high unit cost for fuel. The details are not available to the writer. Natural gas is actually the first member of the petro- leum series methane CH 4 , corresponding to n=l. It is found associated with petroleum. Its rapid exhaustion by wasteful use is one of the scandals of our economic system. It formerly was used for annealing. There remain for consideration '-the raw materials which are grouped under the heading of refractories. These mate- rials include molding sand, fire sand, fire clay, fire brick and, to 'a limited degree magnesite, magnesite 'brick, silica brick, dolomite, gannister and sands'tone. Molding sands are somewhat widely distributed in na- ture and -consequently each plant generally uses a local sand. Molding sands are generally derived from granite which has weathered and are frequently found in glaciated areas. Mold- ing sands differ among themselves and each purpose requires a sand of specific characteristics. In the malleable foundry a sand is desired consisting of well rounded quartz grains, of nearly- uniform and fairlv small size, coated evenly with a moderate amount on\y of fairly plastic but also reasonably refractory clay. The actual size of grain and amount of clay desired will vary with the character of the work. The heavier castings require coarser and clavier sands than the lighter. The uniformity of grain size and -roundness of grain are desired in order to give the greatest possible opportun- ity for the g-as to escape from the molds. If too much clay *Vol. XXVIIT. p. 316, Transactions. American Fonndrymen's asso- ciation. Fuel and Refractories 125 Fig. 61. Operations in a molding sand pit is present or if the 'Sand consists of grains differing largely in size the clay or small silica grains partly obstruct what should be openings between the grains. The clay is needed 'to hold the sand in place. The silica grain is very refractory, so that the refractoriness of the sand depends upon the property of the clay coating. If the clay contains lime or iron oxide the refractoriness is much decreased. Most sands contain vestiges of feldspar from Fig. 62. Hauling sand from a pit 126 American Malleable Cast Iron be Fuel and Refractories 127 the original granite and these sands are relatively easily fusible. The analysis and screen test of sand does not furnish a good guide to its usefulness, as they are difficult to in- terpret. The United States bureau of standards and the Ameri- can Foundrymen's association have gathered extensive data w'hich are available to the interested reader. Tests for porosity, strength of bond, imperviousness and fusibility are more valuable, but a discussion of these proper- ties and their relationships would be too technical to interest the general reader and in the present state of our knowledge would be largely speculative. Frequently sand free of clay is wanted in coremaking, the binder furnishing all the cohesion desired and preventing cores growing too hard, due to the burning of the clay. For such purpose wind-blown lake or sea sands, nearly pure quartz, are generally used. Fire sands are very pure silica sands usually in uniform rounded grains. They seldom contain over 2 per cent of im- purities and are used for the bottoms of acid open-hearth and air furnaces. The presence of a small amount of basic material is required to cause the sand to sinter properly. Sandstone is a naturally compacted mass of silica sand occasionally used in cupola and other furnace linings. Gan- nister is a siliceous sedimentary rock of highly refractory character used in furnace linings usually in crushed form. Fire clays are refractory silicates of aluminum occurring in nature. They contain as impurities oxides of iron, cal- cium and the alkalies as well as some of the rarer metals. The very pure and refractory flint clays possess little plasticity. Other varieties are more plastic and also more fusible. Fire clay is seldom used alone, being mixed with water and crushed fire brick or silica sand to form a mate- rial for patching furnace walls. Fig. 63 shows the location of the principal supplies of molding sand and high grade fire clay in the United States. Clay fire brick, made from fire clay usually at or near the source of clay, consist merely of mixtures of refractory and hard flint clays, ground fire 128 American Malleable Cast Iron jS^^v ^jH^S; $m Fuel and Refractories 129 brick, ground gannister and a plastic fire clay formed into shapes and burned at high temperatures. The manufacture of fire brick is one of the most im- portant ceramic industries and cannot be more than casually referred to here. Brick differs in the material used, the fineness or coarseness of grind, the density to which the ma- terial is compressed and the temperature at which it is burned. The material used -largely determines the refractoriness or melting point. Fine grained, fairly dense and not too hard brick possess great strength. Coarse, open, lightly burned brick resists rapid changes of temperature. Fine, dense, hard burned brick resist penetration of slags, hence every use has special requirements. A noteworthy feature is that all clay brick shrink when first heated. Fire Clay Refractories for Malleable Iron Works The chief deposits of high grade flint fire clays are lo- cated in Pennsylvania, Kentucky and Missouri. These clays are formed from the weathering of feldspar and feldspathic rocks which have the formula K 2 O, A1 2 O 3 6 SiO 2 . Pure kaolins should be A1 2 O 3 2SiO 2 , 2H 2 O, the potassium silicate having been dissolved. The flint fire clays approach this pure clay or kaolin in chemical composition except that they contain some iron oxide 'which gives the burnt product a yellow tint. They are a secondary or transported clay de- posited in still water and are found in the carboniferous areas or coal measures. Where the coal is thick the clay is generally thin, and when the coal 'thins out to almost nothing the clay thickens up to workable deposits eight to 20 feet in thickness. These flint clays usually are mined in the Pennsylvania and Kentucky districts, also occasionally in Missouri, but the Missouri flint clays often lie in pockets. In certain dis- tricts, such as at Mexico, Missouri, extensive deposits are worked 'by stripping the overburden and then mining in an open pit. The following chemical analysis of raw clay and burnt 130 American Malleable Cast Iron Fuel and Refractories 131 bricks will illustrate typical compositions for malleable fur- nace work: Table IV BURNT BRICK ANALYSES Pennsylvania Missouri Kentucky SiO 3 53.05 55.29 54.41 A1 2 O 3 41.16 40.18 . 36.20 Fe 2 O 3 2.65 2.44 2.10 TiO, 1.80 0.00 0.00 CaO 0.00 0.00 2.13 MgO 0.00 0.71 5.16 Alkalies' 1.34 0.76 9.39 Fluxes 5.79 3.91 Cones 32-33 34 31 Ram' Clay Analysis Missouri Flint Clays Per Cent Loss on ignition 12.66 Si0 2 49.08 Al,0, 35.67 Fe 3 O 3 1.28 CaO -. 0.00 MgO .:.... 0.63 Alkalies , ., 0.68 Fluxing parts , , -...,.;; 2.59 Free silica 7.6 The clay is ground and screened in a dry pan in some' plants while others put the raw clay in a wet pan and add excess water making the clay plastic and then introduce the correct per cent of coarse grog, chamotte or calcine. The] latter is simply burnt clay crushed to coarse .sizes 'help take care of strains occurring in brick in malleable iron practice. The clays are all pugged in a wet pan as this process develops the greatest placticity. This mud is carried to the molder who works up portions of it into long (soft mud) bricks and then throws them with great force into the molds which are bumped several times to cause the clay to fill the molds and give good sharp corners. These brick then are carefully dried on a steam-heated floor. In a num'ber of plants and for certain purposes brick instead of being molded as described are pressed hard be- fore drying giving increased density. When thoroughly dried the brick are trucked to kilns where they are set as shown, Fig. 67, leaving spaces for heat and draft. The kilns are down draft, fired with coal, 132 American Malleable Cast Iron Fuel and Refractories 133 natural or producer gas, the gas being used more on con- tinuous kilns. Silica brick, used for very high temperatures, as in the roofs of open-hearth and electric furnaces, is a brick made like a clay brick in which the material is nearly all silica, using only enough clay to permit the brick to be burned to hold together. They are very hard, very dense, and possess an enormous coefficient of thermal expansion. They are strong, almost infusible, but will not withstand sudden tem- perature changes. ft W Hi l| ! ::. ~ -^t, S-^W- ~ W^ W^*: ju^fr., M Fig. 67. Firebrick and special fireclay shapes in a kiln ready to be burned Magnesia consisting of MgO, obtained by heating the mineral magnesite, whic'h is MgCOa, to expel the car'bon dioxiide, is used both ground and as brick in basic furnace linings. In the malleable industry it is used only in electric furnaces. It is very refractory and resists basic slags. It conducts heat readily and must be backed up by a layer of clay brick if heat losses are to be made a minimum. It has relatively little strength. Dolomite, a double carbonate of calcium and mag- nesium, is used in electric furnace bottoms. It is burned before use, resulting in a mixture of CuO-f-MgO in the ratio of about 1.4 to 1.0. The commercial preparations may contain 134 American Malleable Cast Iron from 8 per cent to 25 per cent of other oxides, namely SiCX, A1 2 O 3 , and Fe 2 O 3 . Some producers purposely add iron oxides (or silicates) feeling that the material then deteriorates less in storage and sinters better. Chromite, zirkite, and bauxite, oxides of chromium, of zir- conium, and of aluminum respectively, possess no commercial significance in the malleable industry, although they are well known refractories. Carborundum, silicon carbide, is another refractory which has not found application. VII AIR FURNACE MELTING THE air furnace is the commonest device employed for melting malleable iron, having supplanted the cupola on the score of quality and the crucible furnaces of early days on the score of production and economy. The air furnace is of the reverberatory type in which the metal, in the form of a fairly shallow bath, is melted by the flame from fuel burning in a firebox at one end of the hearth. The flame is drawn over the hearth by a stack at the opposite end from the firebox. In the earliest type, the stack was at one side with a charging door at the end op- posite the firebox. The present arrangement is similar in character to that of a puddling furnace. The early air furnaces were very small; some of the first are said by Davis on the authority of George Belcher to have had capacities of 800 or 1000 pounds, a 1500-pound charge being viewed with alarm. Modern furnaces 'have been, continually growing in size, and now five-ton heats are unusual, capacities from 10 tons to 15 tons being most common in practice. Furnaces have been built and oper- ated with capacities beyond 30 tons, but there are relatively few in use with capacities far above 20 tons. Design Is Simple The construction of an air furnace is relatively simple. Fig. 68 shows an air furnace in side elevation and cross sec- tion. The furnace walls are of fire brick, usually 13 to 18 inches thick, supported and enclosed by cast iron side and end plates about 1 inch thick. The 'bottom or the hearth A is built of silica sand or more. rarely paved with fire brick. Coal is burned in the firebox B, the air being forced through the fire by a blower discharging into the ash pit C '; the ash pit doors D being kept closed. Air is also admitted through the tuyeres E to complete the combustion of the gas and flame com- ing over the front or fire bridge Avail F. The roof of the 136 American Malleable Cast Iron .;^y>'^l Air Furnace Melting 137 furnace consists of a series of removable fire brick arches, or bungs, supported in cast iron frames. A sufficient number of these are removed to permit the introduction of the melt- ing stock. When charging the furnace, the sprue to be melted is introduced first in the form of a. layer of fairly uniform thickness extending nearly the full length of the hearth. On this is placed malleable or steel scrap, the latter usually being kept well forward toward the front bridge wall. Pig iron is placed on top of this in two piles, one well forward, the other further back. Most well designed furnaces are of such dimensions as to be nearly full to the roof when a heat of normal size is charged. Care therefore must be taken to leave an oppor- tunity for the free passage of flame from F to the rear bridge wall. The bungs are then put on and firing commenced. The iron soon begins to heat, naturally first at the top and in front. The firing is so conducted as not to cause much melt- ing to occur until the lower part of the charge is well heated through to a gO'od red. Of the ingredients in the mix, sprue has the lowest melting point, pig iron next, then malleable scrap, and steel the highest. The melting points vary inversely as the com- bined carbon, although the conclusions are slightly compli- cated by the reabsorption or recombination of the carbon of malleable scrap below the melting point. Through the courtesy of H. W. Highriter, the author has been furnished data as to the recombination of carbon in pig iron when heated under circumstances comparable with melting conditions. The data 'have been shown graphi- cally in Fig. 69. Highriter observes a rapid increase in combined carbon at the expense of graphitic carbon above 2000 degrees Fahr. The author has calculated the temperature of the solidus for the observed combined carbon and plotted these temperatures in a dotted line. When this temperature falls below that of the specimen, incipient fusion has com- menced. Melting is complete when the temperature reaches the liquidus which is dependent on the total carbon and calculated by Highriter as 2372 degrees Fahr. The metal by 138 American Malleable Cast Iron observation fused at 2362 degrees Fahr. It will 'be observed that the melting point referred to by the author is that where melting is begun, above this temperature presumably the graphite is rapidly destroyed by solution. Moldenke many years ago published data as to the re- lation between combined carbon and melting point of cast iron and Dyer* refers to the same facts. In interpreting the author's statements, and presumably Moldenke's and Dyer's, confusion between the beginning and completion of melting must be avoided. If the firing is properly managed, it is not necessary to melt the steel, the molten pig iron dissolving the steel as it runs down before the steel actually melts. Some melters ad- vocate introducing the steel only after the rest of the charge is melted. The sprue melts fairly, readily even under all the other material due to its high combined carbon content. As the iron melts the surface oxidizes so that there results both liquid iron and liquid iron oxide, probably Fe 2 O 3 . The latter floats on top of the former and reacts with the carbon, silicon, and manganese of the metal, oxidizing those to CO 2 , CO, SiO 2 and MnO and being itself reduced to FeO almost or quite com- pletely. The oxides of manganese and iron combine with the silici to form an acid silicate which also dissolves some of the refractories in the furnace lining. The resulting slag .soon covers the surface of the molten metal protecting it from further action of the furnace gases. As pools of iron covered with slag form, a good melter will endeavor to roll unmelted pig iron and steel into these pools so as to bring the entire charge under the slag blanket as soon as possible, thus minimizing oxidation losses. The flame conditions also are carefully regulated by atten- tion to the dampers in the blast lines to the firebox and top blast tuyeres and by keeping the openings over the bridge walls and the channel or neck H to the stack of the right dimensions. When the charge is all melted it is well mixed by rabbling with a skimmer bar. The slag is then skimmed off by raking *lron Age, Nov. 17, 1921. Air Furnace Melting 139 it out through the skim holes, the skimmer bar consisting of a 1-inch iron bar having a flat plate, say % x 3 x 9 inches affixed by its center to the end of the round bar. The other end of the bar is bent into a ring to form a handle. Meanwhile the fire is being constantly worked with a poker to keep up active combustion. The heat has to be skim- med at intervals in order to make rapid heating possible Fig. 69. Graph showing recombination of carbon in pig iron and also to keep the final product fairly clean. One producer does not remove the slag, but drains it off after the metal has all been run -out of the furnace. The progress of the heat is judged as to temperatitre and composition by 'the inspection of a freshly broken surface of a not too rapidly cooled ' sample and of the molten metal in the ladle. For satisfactory work a knowledge of the composition of previous heats also is necessary. In a few plants more or less complete preliminary analyses are attempted before tapping. This chemical practice is attended with a certain amount of un- 140 American Malleable Cast Iron Air Furnace Melting 141 certainty as to further changes of composition between sampling and tapping and is therefore less effective than the correspond- ing practice in electric melting. When the metal is hot and of proper composition the clap stopper in the tap hole / is cut through and the metal runs out in a stream into the molders' ladles. In the early days of the art the profile of the furnace roof longitudinally was given very complex, almost fantastic curves. These usually had a sharp dip in the roof just beyond the front bridge, then a rise forming a sort of hump over the hearth, then a drop toward the rear bridge. wall and then a rise directed toward the stacks. Furnaces of the older type had sloping roofs but recently the tendency has been toward a nearly straight roof, lower at the rear bridge than at the front and sometimes rising again into the stack as a matter of convenience. A modern design is shown in Fig. 70. The flame in flowing through the furnace obeys laws similar to those governing the flow of water in channels. These laws 'have been completely investigated by Crum-Grzimai- lo of Petrograd, (Stahl und Eiscn, Dec. 7, and 11, 1911), who developed the mathematical formulae and coefficients applying to the problem in great detail. The discussion is much too technical in character to be even abstracted here beyond the statement that the laws are those which would apply to the flow of one fluid through another, if the two were not mixable and differed in density as does the hot flame and cold at- mosphere- This investigation coupled with a knowledge of combustion and temperature conditions to be expected forms the only logi- cal basis for furnace design. In practice actual furnace design is generally based on modifications of previous designs. This is in many respects sound policy as tending to avoid erratic prac- tices. On the other hand, there is a great tendency toward per- petuation of obsolete features inherent in such a process of ev- olution. An inspection of the designs of many furnaces shows a wide variation on some apparently vital points. These dif- ferences, however, are not always as little justified as may 142 American Malleable Cast Iron appear on the surface for the viewpoint of different designers may not be the same. Thus, for example, it is undoubtedly sound metallurgical practice to make but one heat a day on a furnace and make Fig. 71. A waste heat boiler connected to two air furnaces. Note that coal for auxiliary firing is on hand that a very large one, for ithe brger the capacity the greater is the melting economy, other things being equal. On the other hand, consideration must be given to the space re- quired for molds, to the physical ability of the men to pour, Air Furnace Melting 143 etc- Thus it is that this practice may not be feasible. If heats are required at given time intervals it may be more important to keep the time schedule correct than to get the maximum of economy, hence fuel consumption may be sacri- ficed to melting speed. Such a consideration also may limit the practicable size of heat. Also many furnaces are built in existing buildings, or under other conditions which handi- cap the designer by limiting him to certain dimensions from these causes. A general idea of the usual dimensions of air furnaces can be gained from the following: The volume of the hearth, (the volume of the basin below the level of the skim holes) is directly dependent on the amount of metal to be melted and is not subject to any discretion. One pound of melted cast iron, together with its accompanying slag occupies about 5 cubic inches; therefore 10,000 cubic' inches of hearth must be provided for each net ton of furnace capacity. There are certain practical limits to the depth of molten metal in the hearth which can be successfully worked. Shal- low baths presenting to the flame a large surface per unit weight of metal, heat easily and quickly but also oxidize easily and quickly. Extremely deep baths are difficult to heat, but the great weight per unit of surface favors the rapid transfer of heat from flame to metal per unit of hearth area. Moreover, large capacities coupled with shallow baths may involve impracticable dimensions. Again, the bottom of the furnace must have sufficient slope to assure. complete drainage to the tap hole. Even in unusually short furnaces this slope produces a difference in depth at the tap hole and rear bridge of perhaps 5 inches so that an average depth of less than 2 l / 2 inches is not workable in any event, because it neces- sitates a "feather edge" of metal next the bridge. In practice the average depth of metal ranges from about 5 to 9 inches, the greater depths usually occurring in furnaces of the greater capacities. The depth at the tap hole may be from 2 l / 2 to 5 or 6 inches greater than the average depth de- pending largely on the furnace length. These depths correspond to hearth areas running from about 13% square feet per ton down to less than 8 square feet per ton. 144 American Malleable Cast Iron The requirements of firing, skimming, etc., as well as the maintenance of roof arches sets a maximum inside width of between 5 and 6 feet for air furnaces of the usual design, a few large furnaces of special design have a clear width of 7 feet. When the maximum width is reached the capacity of the furnace can be increased only by increasing the hearth length. Extremely shallow baths are impracticable when large capac- ities are desired because they necessitate long furnaces- For example, 2Oto>n furnaces with a hearth area of IS 1 /-* square feet per ton would be about 45 feet long between the bridge walls. Hearths from 14 to 27 feet long are in common use, and in a few unusually large furnaces they are several feet longer. A certain length of hearth is desirable because it insures a better contact of flames and charge. Excessive lengths cannot be had with small capacities as the furnace would be too narrow. The practicable length also depends on the fuel and firing conditions since a length which does not allow the flame to reach to the rear bridge wall is unworkable. The firebox is almost f of necessity of the same widths as the hearth. The grate area required depends on the rate of combustion of fuel desired and this in turn depends on the furnace capacity and on the relative importance of quick as against economical heating. Air furnace grates burn from 43 to 77 pounds of coal per hour depending on firing' conditions. Values of from 67 to 77 pounds are more common than those near the lower limit. Reported tests indicate that air furnaces use from slightly under 500 to about 1200 pounds of coal per ton of charge. These are extreme ranges* the usual commercial range being from 750 to 900 pounds per ton, depending largely on the size of the furnace- These figures give some indication of grate areas required under various conditions, having in mind also the fact that an attempt to melt rapidly is often uneconomical. It seems to be usual practice to provide from 2 to 21/2 square feet of grate per ton of charge although a number of fur- naces exceed this rate. Many designers do not agree on the correct height of an air furnace roof. From 15 to 17 cubic feet per ton from Air Furnace Melting 145 hearth to roof are unavoidably necessary in order to accom- modate the unmelted charge. This sets a minimum of height for any given hearth area per ton of charge. Quantity of Air Varies Almost invariably the roof slopes downward toward the rear bridge. The old humpback furnaces had a somewhat great- er average height than the more modern straight-roofed furnaces. The average height of roof above the metal at the side walls is about 24 inches. A pound of ordinary melt- ing coal requires about 12j4 pounds of air for combustion under usual operating conditions. The relative amount of air entering the furnace through the top blast tuyeres and through the grates varies in practice, but the average ratio seems to be about 28 to 100. Therefore a pound of coal requires about 10 pounds of air through the grates and 2j4 pounds of air through the top blast in ordinary operating practice. The firebox is operated so that it produces a poor grade producer gas which is then burned with a sufficient amount of air for theoretical combustion. A typical gas leaving the firebox is composed of 1.2 per cent oxygen; 8.0 carbon dioxide; 12.1 carbon monoxide; and 78.7 per cent nitrogen. The gas leaving the stack contains 1.1 per cent oxygen; 12.7 carbon dioxide ; 3.6 carbon monoxide ; and 82.6 per cent nitrogen. The analyses take no account of the water from the combustion of the hydrogen of the fuel. The oxygen in this water and that used in the oxidation of silicon and manganese account for the relatively high value of the nitrogen. The flame gases also contain unburned hydrocarbons of unknown character and amount which escape sampling. Any attempt to further reduce the carbon monoxide content by adding additional oxygen, probably would result in a pro- hibitively high excess of oxygen in the gas, causing heavy oxidation during the melting process. The mechanism of this oxidation has already been re- ferred to as consisting of the oxidation of the iron to the Fe 2 O 3 followed by the subsequent reduction of the Fe 2 O 3 to FeO by the silicon, carbon and manganese of the bath. The amount of oxidation varies widely depending upon the furnace at- 146 American Malleable Cast Iron mosphere and similar conditions. Over an extended period, however, it seems nearly constant for any successfully operat- ing plant. The losses expressed in percentage of the total weight of original charge and in percentage of the amount of each element present are generally about as follows: Table V LOSSES OF ELEMENTS IN MELTING IN AIR FURNACE Total amount Total charge of element 100 -per cent 100 per cent Carbon 0.62 15.8 Silicon 0.33 31.4 Manganese 0.26 48.1 Phosphorus 0.00 0.00 Sulphur 0.01 22.2 Iron 1.14 1.2 2.37 The results of the figures in the second column form an in- teresting comparison of the "oxidizability" of the different ele- ments when melted in an acid furnace. Oxygen Absorbed During Melting A more interesting method of clearly showing the relative affinity for oxygen of the different metals is to calculate the oxy- gen combined with each one of the elements during melting. This calculation has been made using the preceding data and the results are shown in the table below. In the first column is shown the oxygen combined with each of the four oxidiza- ble elements in terms of the weight of original charge and in the second column in terms of the weight of the oxidized ele- ment present in the charge. Table VI OXYGEX ABSORBED BY EACH OF THE OXIDIZABLE ELEMENTS DURING AIR FURNACE MELTING Element present in Total charge original charge 100 per cent 100 per cent Carbon 1.60 50 Silicon 0.38 36 Manganese 0.06 11 Iron ..... 0.32 34 2.36 Air Furnace Melting 147 It will be seen that carbon combines much more greedily with oxygen than any other element, silicon coming next, man- ganese oxidizing much less readily and iron only slightly. Of course the results would differ with variations in gas com- position and furnace lining. It will be seen that the melting process oxidizes a total of 2.34 per cent of the original charge, and combines there with oxygen weighing 2.36 per cent of the original charge. There should thus result a weight of slag equal to 2.5 per cent of the metal charged and of gas equal to 2.2 per cent of the metal charged, were there no contamination from molten refractories. A typical sample of air furnace slag showed the following com- position : Analysis of Air Furnace Slag Per cent FeO 28.80 Fe 2 O 3 1.16 MnO 4.85 Si0 2 (etc) 50.42 A1 2 O 3 14.77 100.00 The metallic oxides aggregate 34.81 per cent of the weight of the slag. From the preceding tables, this corresponds to 13.8 per cent SiCX. Therefore the above slag consists of a mix- ture of 58.70 per cent oxidation products and 41.30 per cent molten refractories and since the weight of slag oxidation prod- ucts was computed to be 2.5 per cent of the weight of the charge the actual slag weight should be slightly more than 4.2 per cent of the original metal charged into the furnace. It is not assumed that these data are absolutely correct but they furnish a fair guide to what may be expected in practice. Refractories Destroyed by Melting Since every ton of iron melted destroys 34 pounds of re- fractories by melting, it is evident that frequent furnace repairs are necessary. The furnace parts most strongly exposed to heat usually are relined at intervals of from 10 to 20 heats. The roof over the hearth lasts usually from 16 to 24 heats and the sand bottom from 10 to 20 heats. In one instance the writer saw a furnace make 34 heats without relining, and 148 American Malleable Cast Iron Fig. 72. Gray sprue; characteristic of high carbon and silicon and sometimes of low pouring temperature (full size) Fig. 73. Gray sprue showing white patches; characteristic of less but still excessive carbon and silicon. Note "in- verted chill," i.e. greater grayness near the surface than at center (full size) Fig. 74. Moderately mottled sprue; characteristic of carbon, silicon and temperature suited to small work (fulj size) in another saw a bottom last 120 heats as a result of careful attention. However, this record is believed to be exceptional. The charge going into the furnace can be computed by adding to the final composition wanted the expected melting losses and then arranging a mixture from the available melting stock conforming to these requirements- The process is one Air Furnace Melting 149 Fig. 75. Normal sprue for metal of the higher carbon ranges of specification metal in average work. Note leaf-shaped bright crystal facets radiating from center (full size) Fig. 76. Similar to Fig. 74 but lower in carbon. Note decrease in leaf-shaped crystals (full size) Fig. 77. Similar to Fig. 76 but quite low carbon. Note finely gran- ular fracture from which the leaf-shaped crystal has almost - disappeared (full size) Fig. 78. "High" iron, i.e. metal low in carbon, silicon and manganese; fracture granular -throughout and edge showing blowholes (full size) of simple arithmetic and the great mystery made of the matter by the older melters was not justified. However, the -skill of ' the melter is important in main- 150 American Malleable Cast Iron taming furnace conditions so that the oxidation losses are uni- form and as small as practicable. The appearance of the flame in the furnace, the eddy currents in the bath and the appear- ance of the slag, whether viscous or liquid, indicate to the skillful melter what is going on in the furnace. Similarly the color and fluidity of the metal and the appearance of the frac- ture after cooling permit of close inferences regarding its composition. Interpreting Appearance of Fracture Among the more obvious indications of the fracture are the presence of graphitic areas or mottles indicative of too high a silicon or carbon or both, larger leafy crystals radiat- ing from the center indicating moderately high carbons de- creasing to very fine granular structures as the carbon falls to near 2 per cent. There also is the rim of fine blow holes and the spray of oxidizing iron arising from the surface of the metal in cases of "burnt" heats very low in silicon. The actual conditions are not even capable of illustration photographically since some of the fractures do not show up clearly except by looking at them in light falling in various directions. It can be shown that by far the largest part of the oxida- tion losses, occurring in practice, is complete, when the metal is melted down and ready to skim. From the time the iron is all melted, before skimming, un- til the moment of tapping no marked changes of composition occur as to carbon and manganese although the silicon will decrease perhaps 0.1 per cent during the removal of the first slag. This presupposes a properly operated furnace. Composition May Vary During Heat Samples taken from the last of a heat frequently show a considerably lower carbon, silicon and manganese content than those taken at the first of the heat. However, this is due, not to a progressive oxidation which would have affected the entire heat to that extent had it been left in the furnace, but to the effect of oxidation on the very thin layer of metal Air Furnace Melting 151 left in the furnace as the last metal is being withdrawn. Only a small weight of metal is of a composition different from the bulk of the heat. A feature that frequently is misunder- stood is the elimination of graphite. Often it is supposed that the fact that the longer the heat is left in the fur- Fig. 79. Changes of metal after tapping nace the lower the graphite is due to oxidation of carbon and silicon. As a matter of fact the elimination of graphite is largely a function of the pouring temperature and time, and metal will show progressively clearer fractures during the progress of the pouring of a heat without any accompany- ing change of ultimate chemical composition. Fig. 79 shows such a condition. In this figure the composition of the metal with respect to total carbon, graphitic carbon, silicon and manganese is 152 American Malleable Cast Iron shown for samples in the form of 1^-inch sand-cooled cyl- inders poured at intervals of three minutes each while the heat was running out of the furnace. It should be said in ex- planation that this was not a normal malleable iron heat but one for a special class of work requiring great perfection of surface on castings on thin sections, hence the high values of silicon and carbon. However, the curve shows strikingly the rapid decrease in combined carbon as the metal is exposed longer to high temperatures. Temperature of Furnace Temperature conditions in air furnaces are not accurately established. The metal flowing from the spout has a tem- perature from 2100 to 2500 degrees Fahr, as measured by radiation pyrometers. Such determinations involve a correction for coefficient of radiation since clean metal does not radiate heat as rapidly as would a theoretical black body. The use of optic- al pyrometers involves a similar correction for emissivity which is however of much smaller magnitude. Optical pyrometer measurements coupled with observations of metal at known temperatures suggest that true values are probably more nearly from 2500 to 2700 degrees Fahr. and that the radiation coefficient is not well established. The flame in the neck when the heat is melted has a temperature of about 2500 degrees. The furnace roof and the flame under it seem to . reach temper- atures up to 3000 degrees or somewhat higher, the average being about 2800 degrees. In the firebox the temperatures are about the same as in the neck, 2500 degrees, Fahr. The latter figures are probably more accurate than those on the flowing metal since black body conditions are more nearly approached. They are if anything somewhat low. The following heat balance gives a general idea of fuel consumption in an air furnace. Since there is considerable variation in furnace practice the correction of heat values for the actual temperature of fuel and air entering the fur- nace was believed an unnecessary refinement. While based only on estimates, this balance gives a fairly comprehensive idea of what becomes of the heat delivered Air Furnace Melting 153 Table VII HEAT BALANCE OF A TYPICAL AIR FURNACE B.t.u. per B.t.u. per ton ton charged charged Heat value coal burned 11,200,000 Heat from oxidation of charge 219,400 Heat of formation of basic silicates 30,000 Total 11,449,400 Latent and sensible heat of metal 878,940 Sensible heat of flue gas 6,112,000 Loss to incomplete combustion of C to CO only 1,232,000 Evaporation of water in coal 10,000 Heat value of unburned combustible in ash 37,335 Sensible heat of slag 42,000 Latent heat of slag (est.) 30,000 Sensible heat of furnace structure 600,000 Radiation conduction and unaccounted for 2,507,125 Totals 11,449,400 11,449,400 to the melting furnace. The values may be summarized on a percentage basis shown in Table. VIII. This indicates clearly that the larger part of the waste is in the sensible heat of the flue gas. This heat occasionallv is recovered by the use of waste heat boilers which gen- erate steam with the heat of the gases leaving the furnaces. The difficulties encountered are largely of a steam engineering character and arise from the intermittent supply of heat avail- able. Prof. Touceda in a paper before the American Foundry- men's Association in 1920, has given tentative suggestions for the utilization of waste heat from air furnaces. These sug- gestions are for various double hearth furnaces in which the waste heat from one hearth is used to preheat the charge in Table VIII HEAT BALANCE IN TERMS OF HEAT VALUE OF COAL FIRED Per cent Per cent Heat value of coal fired. . . 100 Heat in metal 7.81 Heat from reactions in Heat in flue gas . . 54.70 furnace 2.2 Heat in slag 0.64 Incomplete combustion.. 11.30 Heating furnace walls. . . . 5.35 Radiation and conduction 22.40 Total input 102.2 Total output 102.20 154 American Malleable Cast Iron the other. The mechanical means are somewhat complicated involving movable hearths and also somewhat continuous op- eration. From a thermal viewpoint, however, they are most interesting. Reference has been made to the use of forced draft- in air furnaces. The air supply is usually at low pressure, about 4 ounces per square inch, although a few plants use pressures of a pound. In such cases the furnaces must be equipped with doors at the fire 'hole and skim holes. At least one important producer operates on natural draft alone, using no blowers and consequently no top blast. This partic- ular plant depends on extremely high stacks. Many air furnace stacks are from 45 to 85 feet high, and have internal diameters from 24 to 48 inches. The lack of agreement is unaccountable except on the basis of poor design. Nearly all air furnace stacks have capacities far beyond their actual requirements. It has been stated in the general discussion of fuels that both oil and pulverized coal fuel have been tried in air fur- nace practice. As far as the author knows, the use of oil never has been generally satisfactory, owing to difficulties in maintaining the proper furnace atmosphere, free from excess of air or CO 2 . The chemical changes in melting depend entirely upon the temperature and composition of gas in contact with the metal. The use of producer gas entailed similar diffi- culties and was never commercially adopted, except of course in open-hearth practice. Similar difficulties have been encoun- tered in the use of pulverized coal but have been successfully overcome, at least by a few combustion engineers. A successful equipment of this character is shown in Fig. 80 and consists of a hopper containing the pulverized fuel pro- vided with screw conveyors for feeding a stream of coal into the current of air from the blower shown in the lower right hand corner of the picture. The ends of the con- veyor shafts are shown under the numbers 1-2-3-4-5 painted on the hopper. The current of air loaded with coal dust enters the furnace through three burners in the head wall of the fire Air Furnace Melting 155 box, which is blocked up; and through two burners through the roof at the point where the top blast usually enters. By proper manipulation of the relative supply of coal and air to these several burners, proper control may be main- tained and satisfactory working insured. The entire problem Fig. 80. A po'.vclercd coal attachment for an air furnace is merely the design of a burner capable of so feeding the fuel into the air as to maintain uniform combustion conditions with coal and air supply capable of regulation through a fairly wide range. Such a set of burners operated to duplicate furnace atmos- pheres corresponding to the best air furnace practice will produce results superior in control and economy to results under hand firing. The improvement results primarily from the constancy of ratio of coal to air throughout the heat, thus 156 American Malleable Cast Iron ^ avoiding the losses due to alternately incomplete combustion and excess air w'hich occur even with the best hand firing when the average condition is perfectly controlled. . Table IX CHEMICAL CHANGES IN AIR FURNACE Metal charged pounds Fe .1901.3 C 62.0 Si 21.6 Mn 11.0 S 0.9 P 3.2 Clay O N H Ash Water Refrac- Coal Air tories 14 Total [olten metal Slag Flue gas 1878.5 22.8 49.6 ... 617.0 15 66 Cin- ders 4.4 6090 671.0 21.6 11.0 5.9 3.2 34.0 1918.0 7243.0 39.0 21.0 14.0 "5.6 '.'.'.'.'. '.'.'. 5.8 5.2 1.1 1.2 2.0 32 V.6 . 34.0 .... 1918.0 ... .... 7243.0 ... 39.0 21.0 14.0 34.0 .... 15.0 1903.0 7243.0 2l'.0 39.0 '.'.','. '.'.'. ' 'l4.0 Total. 2000.0 Ibs. 688.0 9161.0 34.0 11,883.0 1953.2 84.8 9818.0 27.0 The author has seen the results of many tests on this type of equipment but it is doubtful whether data have yet been accumulated which warrant a definite conclusion as to economy of operation due to pulverized fuel. The tests which he has seen seem to indicate that the requirements as to furnace atmosphere are such that no direct saving on coal is practicable. The economies may rather be expected to result from decreased labor and refractory costs, and greater independence in using poor coal. The data at hand point also to a lower and much more constant loss by oxidation of the several metals than is nor- mal to ordinary air furnace practice, but insufficient experience is available to be sure whether this condition always exists. As a skeleton outline of the metallurgy involved in the operation of an air furnace the outline of the chemical changes shown in Table IX may be interesting. The summary is typical only and does not necessarily apply exactly to any given case. The summary is based on the weights of each material and each element entering into the reactions for one net ton of charge. Air furnaces usually are operated by a crew of either two or three men exclusive of those doing the charging. Air Furnace Af citing 157 bringing in fuel and stock, etc. The majority of air fur- naces make a heat in 20 to 30 minutes per ton plus about one half hour if the furnace is hot to begin with, or plus one and one half hours if the furnace is cold at the start of the melting operations. Large furnaces melt faster, per ton, than small ones, but large heats still take longer to make. It is said that one plant, using oil fuel made heats around 30 tons in three and one half or four hours, although the writer is not prepared to vouch for this statement. Another plant making heats of this size with coal runs 16 to 18 hours to a heat, it is said. In most plants skimming begins when the heat is well melted which will be from one and one half to two hours before the heat is ready. In a plant where instead of skim- ming the slag is tapped out after the iron is poured it is claimed that no loss of time or fuel is incurred due to this meth- od. The operation is on fairly large furnaces. In spite of the obvious desirability of this operation, if practicable, it has not been adopted elsewhere. The author does not know whether or not this conservatism is justified. The feeling seems to be one of suspicion as to the general economy and practica- bility of the operation. VIII ELECTRIC FURNACE MELTING PRACTICALLY the only radical change in melting prac- tice which has been introduced into the malleable indus- try in the last half century is the use of electric furnaces. So far only one producer operates under this method, which is protected by patents covering the conditions necessary to com- mercial success. In electric operation, increased accuracy of chemical con- trol is made possible and the success of the melting operation is largely independent of variations in quality of stock and fuel and of blast and similar conditions. The belief that electric melting is adopted because it permits the manufacture of al- loys of compositions unattainable in the air furnace is not founded on fact. While it is possible, for example, to make iron as low as .017 per cent in sulphur, if desired, there is no engineering advantage in such an operation. Electric melting as practiced today is conducted by the triplex process, developed by W. G. Kranz, which, as the name indicates, is conducted in three distinct stages. This process supplements the advantages of the electric furnace with the use of a cupola and a bessemer converter to assist the elec- tric furnace in operations to which it is not so well suited. The rationale of the process is as follows : The electric furnace alone is suitable for melting or heating metal with slight contact with air or any other substance ex- cept the furnace lining and slag. Therefore, it is suited rather to keep the composition of its contents unaltered than to make changes in composition. Chemical changes occur therein only as a consequence of the addition of various alloys of slag-making ingredients and the effect of such additions can be quantitatively controlled. The changes of chemical composition easiest of attainment in the electric furnace are increases in silicon, manganese, or phos- phorus and decreases in sulphur and oxygen. Carbon can be added or removed, or silicon removed with greater difficulty 160 American Malleable Cast Iron but the removal of phosphorus is not practicable under the usual operating conditions in malleable melting. 'Whereas the electric furnace is an expensive source of hear energy, the cupola is the cheapest known method for merely Fig. 81. Cupola producing molten iron The starting point of the Kranz triplex process melting cast iron, composition being no object. . Cupola melt- ing always removes at least part of the silicon and manganese and adds sulphur, leaving the phosphorus unaltered. The car- bon content is nearly independent of the mix used depending only on the condition of the fuel bed. The carbon content always is relatively high. Electric Furnace Melting 161 The bessemer converter furnishes an easy and economical way to remove all carbon silicon and manganese from iron but adds a great deal of oxygen. The three units form an ideal team, each possessing good qualities to supplement the weak points of its mates. The cupola furnishes cheaply a supply of liquid iron of high and approximately constant carbon content which readily can be controlled as to its maximum silicon, manganese and phos- phorus content, but may have high sulphur from the fuel. Carbon, silicon and manganese can be removed from this metal in the bessemer converter, although oxygen may be added. By taking the proper relative amounts of cupola and bessemer metal a mixture can be produced having a 'carbon content close to any desired value, and which also is below any desired fixed values in silicon, manganese and phosphorus. However, it contains an indefinite and relatively large amount of sulphur and oxygen. This molten mixture can be given its final heating in the electric furnace without too great expense, and"*.iby the use of suitable slags the sulphur and oxygen can be removed without any -effect on the silicon, manganese, or phosphorus. Guided by the analysis of the molten charge, silicon and manganese can be added to adjust these values as desired and a product- made without prohibitive cost, adjusted to chemical specifica- tions on each of the five common elements and freed from oxygen. . These are the steps in the Kranz process, which since passing through the experimental stage in 1913-1914 has pro- duced many thousands of tons of malleable cast iron in two plants of the largest producer of malleable in the world. The proc- ess as outlined comprises melting in the cupola ; decarburizing in the converter; heating, desulphurizing and deoxidizing and raising the manganese and silicon in the electric furnace; and, if desired, adding sulphur in the ladle. For still greater uniform- ity it was once suggested that the cupola and converter metal be stored in a mixer prior to its introduction into the electric furnace, but practice has proved that this step is not nec- essary. It has been found that a product varying from dead soft 162 American Malleable Cast Iron steel to" gray iron, and including alloy steels can be made by this, process at the will of the operator. If dephosphorization is desired, for example in steel-making, an extra step is re- quired in the electric furnace, involving the formation of a dephosphorizing slag and its removal before proceeding with the desulphurizing and deoxidizing. Fig. 82. Two-ton side-blow converter producing liquid steel from cupola metal in triplex process Metallurgy of Triplex Process It may be well to consider the individual steps involved in greater metallurgical detail. In general, the melting stock consists of sprue and malleable scrap and high silicon pig iron. The mix is calculated only to be close to the desired value in silicon content. The manganese automatically remains low and with a little care the phosphorus can be kept below about 0.19 per cent, which is all that is required. Electric Furnace Melting 163 It is intended that the cupola metal shall run slightly under 1 per cent silicon. Too low a value causes trouble from gum- ming' up the cupola taphole and spout and the ladle. The maxi- mum is determined by the metal to be made. The composition of the metal leaving the cupola under ordinary working condi- tions is approximately as follows: Carbon, 3.10; silicon, 0.80 to 0.95; manganese, 0.12 to 0.19; sulphur, 0.09 and up, and phos- phorus, 0.14 to 0.19 per cent. The dimensions of the cupola are such as to allow the unit to run continuously to produce the metal required by the electric furnaces. Interruptions and intermissions are undesirable because they affect the tempera- ture of the fuel bed and consequently the carbon content. The ratio of iron to coke in the cupola may average 7 to 1, varying somewhat with operating conditions. Two cupolas are provided and are used alternately to permit repairs. 'i he converter easily reduces the molten metal to a composi- tion about as follows : Carbon, 0.20 and under ; silicon, trace ; manganese, trace; sulphur, 0.12 and up; and phosphorus, 0.17 per cent and up. A considerable oxidation of iron also occurs, which together with the mechanical loss in the form of fine drops amounts to from 8 to 15 per cent of the converter charge. If a carbon content of say 2.60 per cent is desired, cupola and converter metal in the ratio of 240 to 50 will be required and the mixture will have a composition as follows: Carbon, 2.60; sili- con, 0.66 to 0.78; manganese, 0.10 to 0.16; sulphur, .095 and up; and phosphorus 0.14 to 0.19 per cent. Since each furnace heat is handled as a unit, it will be seen that the converter charge is dependent on the capacity of the electric furnace and the carbon content of the cupola metal. In the illustration chosen the converter must deliver 50/290 or about 17 per cent of the capacity of the electric for each blow. The metal introduced must exceed this amount by the expected oxidation and mechanical losses. The electric furnaces actually in use have a rated capacity of six and 15 tons, respectively so that when working at capac- ity the converter would have to deliver 1.02 and 2.35 tons respectively. The electric furnaces used are of the Heroult type, operat- ing on 3-phase, alternating current. The 6-ton units consume 164 American Malleable Cast Iron 800 kilovolt-amperes of power at 80 to 100 volts and the 15 -ton units from 18,000 to 22,000 kilovolt-amperes at from 90 to 110 volts. Handling Charge in Furnace The internal diameter of the larger units is approximately 10 feet. In all cases the bottoms are dolomite and the lining of the side walls magnesite. The molten metal is introduced into the furnace, the arc formed, and a lime slag made on the surface. The slag-making ingredients are lime, fluorspar and coke; in amounts determined by the working conditions and not by weight. About 150 pounds of lime and coke and 100 pounds of fluorspar may be used in a 12-ton heat, the active ingredient of the resulting* slag being calcium carbide, CaC 2 . The ac- tual amounts of slag-making ingredients are however not de- termined by weight but by the appearance of the slag and the "operating conditions of the furnace. This carbide reacts energetically with any metallic oxides present. For instance 3 FeO + CaC 2 = GaO -f 3 Fe + 2 CO No appreciable amounts of CaC 2 are formed until the oxy- gen is practically completely eliminated. At that stage the elimination of sulphur begins, the products being CaS and carbon, which dissolves in the metal. This process can not be conducted under any but a reducing condition for CaS is easily oxidized to CaO, the sulphur unfortunately not burn- ing to SO 2 but dissolving in the iron. This introduces certain difficulties in lowering the silicon content. For example silicon is easily and almost quantitatively oxidized by ore, the reaction presumably being Si + 2 Fe a O 3 = SiO 2 + 4 FeO Unfortunately the FeO of the resulting slag immediate- ly reacts as follows : FeO + CaS = FeS + CaO and the desulphurizing must be recommenced. The removal of silicon can be conducted in this way, but it is a cause of difficulty in the maintenance of the desired slag. Fortunately the high carbon alloys occurring in malle- able practice do not take up carbon from the carbide slags used to any appreciable extent, nor does the CaC 2 reduce a Electric Furnace Melting 165 considerable amount of silicon from any calcium silicates which may be present. A sample is taken for analysis after the metal is thor- oughly mixed in the furnace and should show a correct amount of carbon and phosphorus, and a deficiency in sili- con and manganese. These latter two elements are added as ferrosilicon, ferromanganese, Spiegel or similar alloys. Carbon can be added as pig, cupola iron, or in very 'hot heats as coke or can be reduced by steel additions. Silicon can be removed with ore as previously described but it is not intended that this Fig. 83. Transfer train consisting of electric motor car and trailer with crane ladle. This equipment is used in carrying cupola and converter metal to the electric furnaces be done in regular practice. The removal of phospohrus from malleable heats is so expensive that it is cheaper to scrap such heats than to attempt to correct them. Temperature Limited by Operations The temperature to which electric metal can be heated depends only on the refractories used and in commercial prac- tice is from 2600 to 3000 degrees Fahr. The figures are by radiation pyrometer and in the writer's judgment are likely to be lower than the correct values. More recent data by optical pyrometer show temperatures from 2900 degrees to 3000 de- grees Fahr. It appears therefore that the figures around 2600 166 American Malleable Cast Iron degrees arose from an improper correction for coefficient of ra- diation. The relative merits of the two systems of pyrometry have been discussed in connection with air furnace melting. It is customary to take a heat away in one or two large ladles and to proceed immediately with another heat. The advantages of the process already have been pointed out and all point back to ac- curacy of control. The most serious limitation is the expensive first cost of the melting installation, which places it beyond the reach of the small producer. Furthermore the process is not suited to intermittent operation involving the banking of cupolas and filling of electric furnaces with coke. To obtain success- ful results a 24-hour day during the working week is neces- sary. Counting iy 2 hours per heat or 16 heats per day and allowing for some loss of time for repairs between heats, and bearing in mind possible reductions in economy where very small units are used, a simple calculation will indicate that suc- cessful operation can be had only in plants of fair capacity. The two plants now in operation are equipped with three small and two large furnaces, respectively, and are intended to operate on large tonnages. Furthermore, the crane service required for the handling of hot metal, etc., almost precludes the introduction of hot melting into any but an especially built plant, thus further limiting its general introduction. All this is in addition to the limitations to the general use of the process due to its control through patent protection. Fur- nace repairs are relatively much less frequent in electric fur- naces than in air furnaces. The bottom is taken care of after each heat. The magnesia side walls and silica roof each last from 120 to 240 heats, while the basic bottom, being repaired after each heat, lasts indefinitely. The cost of heat in the electric furnace is high, but on the other hand the utilization of heat reaches an extremely high efficiency owing to the elimination of the losses in fuel-fired furnaces arising from the escape of the hot products of com- bustion. The current is on about one hour for each heat. Charging, tapping and patching consume up to 45 min- utes of time. Cupolas are intended to run a week on each lining but usually are repaired at 24 to 72-hour intervals. Electric Furnace Melting 167 The converters are of the side-blown type of a capacity suited to the Heroult furnace they serve and are lined with ganister. Converter bottoms last about a week, and the tops nearly indefinitely. It will be instructive to follow quantitatively the chemical changes occurring. The following analysis is typical of the slag produced by the cupola. Per cent SiO, -52.90 A1,O, 12.80 FeO 5.10 Fe a O 00 MnO 2.60 CaO 21.30 MgO 3.70 S 0.20 Undetermined and .error 1.40 100.00 This is practically a mixture of molten refractory and limestone, little oxidation of the metal having occurred under the strongly reducing conditions of the cupola. Assuming that the cupola charge consists of 10 per cent silicon pig, sprue and malleable scrap, the two latter averaging 0.70 silicon, in order to have a mixture at 1.10 silicon the mix will contain 4.3 per cent pig iron and, for example, 40 per cent sprue and 55.7 per cent malleable scrap. The average analysis of such a mixture figures out carbon, 2.68; silicon, 1.10; manganese, 0.27; sulphur, 055 and phosphorus, 0.177 per cent. This metal, when melted and leaving the cupola has a composition of carbon, 3.10; sili- con, 0.85; manganese, 0.15; sulphur, 0.09 and phosphorus, 0.177 per cent. This change of composition coupled with the pre- viously given slag analysis amounts to a net loss by oxidation of 0.166 per cent of the total weight charged. The oxidation of silicon, manganese and iron is nearly balanced by the gain in sulphur and carbon from the fuel. In practice there is a loss of noticeable magnitude due to me- chanical causes. By calculation the slag corresponds to 5.8 per cent of the weight of the charge; 1684 per cent is derived from oxidation of the metal ; 25 per cent from the lime- stone added as a flux ; and the balance from the fusion of the 168 American Malleable Cast Iron furnace lining, coke ash, impurities in stone, etc. Assuming the limestone to have been 90 per cent CaCO 3 , the weight of the limestone added was about 50 per cent of the slag weight or 2.9 per cent of the weight of metal charged. The limestone lost to the flue gas an amount of CO 2 equal to 11 per cent of the slag weight. When the cupola metal is blown in the converter it be- Fig. 84. Heroult electric furnace in which cupola and converter metal is charged for final step in triplex process comes a steel containing, for example: Carbon, 0.10; sulphur, 0.095; and phosphorus, 0.187 per cent. The slag formed has a composition of which the following is typical : Per cent SiO 3 57.50 A1 2 O 3 1.43 FeO 34.41 Fe,O, 1.45 M-nO 3.80 OaiO 0.25 M-gO 0.34 Error and undetermined 0.82 100.00 Electric Furnace Melting 169 A loss in weight of 5.36 per cent of the weight charged into the converter is indicated. In practice a larger loss is noted due to mechanical losses and to considerable amounts of iron oxide which escape as fume and are not taken into ac- count in the analysis. The slag is equivalent in weight to 4.86 per cent of the metal charged. Of this slag 50.08 per cent is an oxidation prod- uct of the metal and 49.92 per cent is fused refractory. In the electric furnace no oxidation takes place, the only elements affected being sulphur and oxygen which leave the metal to become calcium sulphide and carbon monoxide, respec- tively. The former remains in the slag, while the latter escapes as a gas. Therefore the slag in the electric furnace is not in any material degree derived from the elements in the iron, but depends for its quantity and largely for its composition on the slag forming additions used. These are lime (CaO) fluor- spar (CaF 2 ) and coke. The supposition is that the coke and lime form .calcium carbide which removes both sulphur and carbon. However the slags never are nearly pure mixtures of CaC 2 and CaF 2 . Typical slag obtained under conditions which would possibly have destroyed "any CaC 2 by the action of the atmos- pheric moisture had a composition as follows : Per cent SiO, 29.80 A1,O 3 2.85 FeO 0.50 Fe 2 O 3 nil M.nO 0.18 CaF 2 0.70 CaO 44.51 MigO 7.55 CaiS 7.20 Undetermined '6.71 100.00 From the behavior of the slag it seems reasonable that most of the lime is combined with silica and that there is but little free CaO as Ca(OH) 2 either normally present or derived from the decomposition of carbides. Possibly the CaO from these sources may run to 5 per cent or similar undetermined amounts. 170 American Malleable Cast Iron Possibly the MnO s'hown is MnS floating up from the met- al, in which case the CaS would be reduced and CaO increased to allow for the S combined with Mn. The fluorine apparently is largely eliminated in the furnace. Data as to slag quan- tities are uninteresting as having no connection with the metal- lurgical principles. The additions may amount in the aggre- gate to perhaps 1 or 1.5 per cent of the weight of the metal. Fig. 85. Heroult furnace tilted for pouring Metallurgy of the Slag Assuming a desulphurization of .07 per cent, the slag composition referred to and excluding sulphur from the coke amounts to around 44 pounds of slag per ton of metal. Of the slag the SiO 2 A1 2 O 3 and MgO are primarily derived from the furnace lining. Those comprise 40.2 per cent of the en- tire slag. Therefore for each ton of metal 17.6 pounds of re- fractory are melted and 26.4 pounds of slag is formed from lime, fluorspar and carbon and from the metal itself. Of the ingredients from the metal the principal item in weight is 1.4 Electric Furnace Melting 171 Table X BALANCE SHEET FOR DISTRIBUTION OF METALLOIDS IN ELECTRIC FURNACE PRACTICE In pounds per ton of cupola charge From From In cupola cupola converter charge coke air Total C . 53.6 8.3 61 Si 22.0 Mn 5.4 P 3.54 S 1.10 0.7 O ... .. 0.60* * 0.60 Fe 1914.36 1914.36 22.0 5.4 .3.54 1.80 In electric To electric In final In cupola In con- In con- furnace furnace product slag verter slag verter gas slag atmosphere C 51.90 ... ..Q> 10.0 0.00* Si 14.17 5.0 2.83 '. . . 0.00* Mn 2.46 2.4 0.50 . . . 0.04 P 3.54 ... ... ... 0.00* S 0.08 ... ... ... 1.00 O ...* ...* ...* ...* 0.60 Fe 1904.84 4.8 4.60 ... 0.12 *Includes only those amounts at some stage alloyed with the molten metal. pounds of sulphur, the MnO and FeO being only about 0.3 pounds per ton. Deducting these two, the slag has 24.7 pounds of material per ton of metal derived from the slag-forming additions. All of these figures men- tioned are to be considered as suggestive only. A balance sheet of the elements concerned in the triplex process is shown in Table X. It must be understood, however, that the process has not been quantitatively investigated to the point where all the reactions are clearly worked out. The figures in the balance sheet for oxygen are merely estimates. The sulphur data are not based on a complete series of tests, but are in accord with current practice. The table neglects oxygen in ori- ginal metal and final product. Ferromanganese and ferrosilicon are not supposed to be added. If charged into the electric, these alloy quantitatively with the charge. Heat Balance of Triplex Process A heat balance for the triplex process is extremely inter- esting as giving an insight into the character of heat losses re- 172 American Malleable Cast Iron Table XI GENERAL HEAT BALANCE OF TRIPLEX PROCESS Cupola B.t.u. B.t.u. Heat value fuel 3,718,000 Total heat, metal 1,692,000 Sensible heat, slag 63,800 Sensible heat, flue gas 180,000 Heat value of Fe, CO in flue gas 744,000 Radiation and unaccounted for 1,038,200 Total output 3,718,000 Converter Total heat of metal charged 282,000 Heat of combustion of C, Si, Mn 93,400 Total input 375,400 Total heat, metal 292,800 Sensible heat, slag , 10,900 Sensible heat, gas and undetermined 71,700 Total output 375,400 Electric Furnace Total heat metal charged 1,690,800 Heat equivalent of electric input 564,200 Total input 2,255,000 Total heat, metal 1,865,000 Sensible heat, slag 23,000 Radiation and undetermined 367,000 Total output 2,255,000 maining. Unfortunately the results of complete tests of the process including all the factors involved are not available. Also the heat of formation of some of the compounds entering into the reaction, more particularly in the electric furnace are not known. In the absence of this information the following bal- ance has been built up on estimates from other sources of the composition of gas leaving the cupola and converter, and of the temperature of the cupola gas, and of the metal at various stages. Also the heat of formation of the slag has not been considered a source of energy nor has allowance been made for the latent heat of fusion of slags and refractories. The presence of oxygen in the metal, at various stages has not been followed quantitatively so. that no account of the thermal effect of the formation and reduction of FeO can be taken. The latter items are included among the undetermined Electric Furnace Melting 173 Table XII HEAT BALANCE OF UNITS IN TRIPLEX PROCESS B.t.u. B.t.u. Heat value of coke 3,718,000 Heat value of current 564,200 Heat combination of Fe Si, Mn and C 93,400 Total heat input 4,375,600 Incomplete combustion in cupola. . . . . x 744,000 Sensible heat, slag : 63,800 Sensible heat, flue gas 180,000 Radiation and undetermined 1,038,200 Total cupola loss 2,026,000 Sensible heat, converter slag. 10,900 Sensible heat, gas and undetermined 71,700 Total converter loss 82,600 Ladle loss by radiation (preheated ladle) 12,000 Sensible heat, slag 23,000 Radiation and undetermined 367,000 402,000 Total heat, metal 1,865,000 Total output 4,375,600 losses at the various stages of the process. However, the balance in Table XI, based on one ton of metal charged into the cupola and on temperatures above atmospheric may be regarded as indicative of the major items. The cupola utilizes 45.5 per cent of the heat of the fuel. The converter delivers 77.7 per cent of the total heat sup-, plied, using 11.5 per cent of the heat of combus- tion of the 'elements burned in further heating the metal. The ladle loss in transferring the metal, not shown above, amounts to less than 1 per cent. The electric furnace delivers in the metal 82.7 per cent of all the heat furnished it, using 30.9 per cent of the thermal equivalent of the electric input in heating the metal. Heat Balance in Per -Cent A summary of the heat balance based on the process as a whole appears in Table XII. The tabulation may be condensed 174 American Malleable Cast Iron somewhat further and expressed in percentages of the total heat supplied by fuel and power as follows : Per cent Heat of combustion fuel 86.5 Heat equivalent of power 13.5 Heat of combustion of elements in converter 2.2 Heat loss in cupola 47.3 Heat loss in converter 1.9 Heat loss in ladle 0.3 Heat loss in electric furnace 9.1 Total heat metal 43.6 Totals 102.2 102.2 The figures show the relatively very great thermal effi- ciency of the process as compared with air furnace or open- hearth melting. A heat made from cold stock in the electric furnace would show a still hig'her thermal efficiency, approxi- mating that of the electric furnace alone. This would not, how- ever, correspond to a greater economic efficiency in view of the greater cost of a 'heat unit as electric energy than as coke. From the viewpoint of fuel consumption a vast consideration of the electric, furnace is not complete without pointing out that a consumption of 21/2 pounds of coal per kilowatt-hour is an ex- tremely economical figure, attainable only in unusually large turbine-driven plants. There would be superimposed on this further transformer and line' losses so that the electric furnace may get from 4 per cent to 8 per cent of the energy of the boiler fuel as electric energy. This consideration, coupled with the high overhead for the powder plant, accounts for the great cost of heat energy derived from electric power as compared with that of an equal amount of heat energy potentially present in the fuel. IX CUPOLA AND OPEN-HEARTH MELTING IX ADDITION to air and electric furnace melting, which was discussed in Chapters VII and VIII, there are two com- mercial methods of melting malleable. That which employs the cupola can he dismissed with a few words, since its use for producing specification metal has been prohibited by the specifica- tions of the American Society for Testing Materials since their first revision. The objections to cupola metal are based on lack of uni- formity of product and lack of control. Because of construc- tion of the cupola and its method of operation, no large amount of liquid iron is accumulated at one time; therefore there is no assurance that successive taps will be even nearly the same in composition unless the charge consists of only one material, which manifestly is impracticable. These variations are of no consequence in the general run of gray iron castings, but in malleable practice with its much reduced practicable range of composition they are prohibitive, especially for large work.'- Furthermore, since the cupola runs continuously for several hours there is no means of judging the fit- ness of the iron for its intended purpose either by analysis or fracture before it is poured. Control of Metal Limited Even when the best possible uniformity is secured the cu- pola process has limitations of control which render it unsuitable in the production of a general run of malleable castings. The molten iron runs down through a mass of' incandescent coke, meeting in the spaces between the coke a stream of gas, originally air, but converted by the fuel into a mixture of car- bon dioxide, carbon monoxide and nitrogen. Under any given operating condition, especially as to tem- perature, a definite equilibrium exists which determines the com- position of the products of combustion in contact with mean- 176 American Malleable Cast bo Cupola and Open Hearth Melting 177 descent carbon at that temperature. The descending liquid iron thus passes into a zone in which temperature and gas composi- tion are adapted to equilibrium with molten iron of only one specific carbon content and capable of adding or removing car- bon easily if the metal comes down lower or higher than this value in equilibrium with the gas phase. Therefore a cupola produces metal of a carbon content almost independent of that of the charge and dependent solely on the combustion conditions. The possible range of working conditions is such as to produce metal containing from about 2.70 to 3.25 per cent carbon a value too high for the production of a high class product except in small work. The sulphur content of cupola metal also is invariably high in view of the intimate contact of molten metal and fuel. Some cupola metal made for extremely small work thus is converted into white heart malleable, possibly without the full understanding of the operator, and the work is annealed by de- carbonization of the thin sections and not by graphitization. The surviving successful application of the cupola process to black heart malleable is in the manufacture of pipe fittings where the product usually does not have the greatest possible strength. The metallurgy of cupola melting has been considered in Chap- ter VII in connection with the triplex process. However, a higher fuel ratio is common in ordinary cupola melting than in the triplex process because the iron must leave the cupola at a higher temperature in order to run into molds than if it is to be, handled only by a crane ladle. A ratio of metal to fuel of between 4 to 1 and 6 to 1 may represent operating practice, and this represents the one great advantage of the cupola cheapness both of construction and operation, the utilization of heat being about two or two 'and one-half times as efficient as in the air furnace. Open-hearth melting, especially when large tonnages and continuous operation are involved, should be a desirable method of operation. That its practice is confined to relatively few 178 American Malleable Cast Iron o O 'o o c7) U) '5; v- U oo fcio Cupola and Open Hearth Melting 179 plants may be due to the conservatism of the industry and to the tonnage limitation. In general, the open-hearth furnaces used in the malleable" industry are similar in construction to those used in steel making and in size represent the lower limits of capacity used* in that industry. Some experimental heats have been made in basic furnaces but acid-lined furnaces apparently are used for com- mercial operation. The melting operation is similar in principle to aar furnace melting, except in the application of the heat. Furnaces ranging in capacity from 5 to 20 tons have been used, the larger units being preferred when practicable. The furnace roofs are of silica brick and the bottoms of silica sand. The regenerative system upon which the operation depends is so well known as hardly to require description. The products of combustion leaving the hearth pass through checkers of fire brick and impart their heat to these brick. When the brick is thoroughly heated the direction of gas passage is reversed, the air being drawn into the furnace through the previously heated checkers. The products of combustion pass out through check- ers at the opposite end of the furnace. When producer gas is used it also is preheated. The incoming air gradually cools the hot set of checker work while the products of combustion heat the checker at the outlet end when the latter grow hot the direction of passage is again reversed, this operation being continued. Using Heat of Flue Gases The period of reversal depends upon the heat capacity of the checker work and in ordinary design a reversal every 15 to 30 minutes may be contemplated. The object is to utilize the sensible heat of the flue gases. The gases leaving the iron can- not impart heat thereto unless their temperature is above that of the metal. However, their heat can be imparted to the furnace content by using it to preheat the air and sometimes the fuel used before the combustion begins. In this way a higher furnace temperature and lower heat loss are maintained. The heat loss depends upon the temperature of the out- ISO American Malleable Cast Iron f^&TSfa^W^ffifflfliXfff'. Cupola and Open Hearth Melting 181 flfrf 4) C <* O M ^ !3 cj jn< rt = G H jc -^ rt o ^ 0-5 S - ^Ilil 1 s SS'sIl s - S -a Z X 182 American Malleable Cast Iron going gases and this in turn upon the volume of the regenerator chambers and the period of reversals. In theory the outgoing temperature might be reduced to that of the incoming air and gas but this is practically impossible. Campbell states that open-hearth steel furnaces should be capable of operation without the stack gases attaining a red heat. However, this result is not often attained. Assuming this red heat to be 900 degrees and the gas composition to be the same as in air furnace melting the sensible heat of the out-going gases is only 9/25 of that of the air furnace, counting from degree Fahr. as a basis (which is not strictly correct). Therefore the heat value saved in the regenerators is 17/25 of that lost in the stack in air furnace practice. Using the heat loss in sensible heat of gases, leaving the air furnace as 7800 B.t.u. per pound of coal, and counting again from degrees Fahr., the heat saved per pound of coal would be 5304 B.t.u. or over one-third the heat value of the fuel. Quoting Campbell in Manufacture and Properties of Struc- tural Steel> for a given sized chamber the escaping gases are a certain number of degrees hotter than the gases that go into it. If this difference is 300 degrees, then if the entering gas is 400 degrees, the escaping gases will be 700 degrees, and if the en- tering gases are 700 degrees the outgoing gases will be 1000 degrees. It will be seen that this reasoning implies that no change of economy results from changes of temperature in pro- ducer gas passing from the producer to the furnace. If no heat is lost in the gas while passing from the producer to the regenerator a loss corresponding to this saving is incurred in the outlet gases. Since open-hearth furnaces are much less common in the malleable industry than air furnaces, correspondingly less is known of their design and operation. For general information on open-hearth operation the interested reader is referred to the literature of the subject regarding steel melting. By kindness of Messrs. Lanihan and Fulton; the writer has been given access to a certain amount of data accumulated in the successful operation of open-hearth furnaces by the Fort Cupola and Open Hearth Melting 183 Pitt Malleable Iron Co., Pittsburgh. Much of what follows is based on that practice supplemented where necessary by con- clusions drawn from other sources. Malleable melting in the open hearth differs metallurgically in one essential respect from steel melting. The steel maker operates to greatly reduce the carbon and silicon content of the bath by oxidation. In malleable practice this oxidation must be kept down as much as practicable to insure control and re- duce melting losses. Therefore the furnace atmosphere is sub- ject to the same limitations as to composition as in air furnace practice. In view of the fact that this oxidation is actually kept down to about the same limits as in air furnace practice it seems reasonable in the absence of direct figures to assume that the CO, CO 2 , O and N in the products iof combustion should be about the same as is given in Chapter VI. An essential difference, however, will be the presence of a greater proportion of steam or water, since these furnaces are operated on natural gas and oil. In the chapter on air furnace melting, the flue gas analysis was given as oxygen, 1.1; carbon dioxide, 12.7; carbon mon- oxide, 3.6; and nitrogen, 82.6 per cent. Assuming the gas in the present case to have this composition and assuming that the formula of the petroleum is C n H 2n + 2 the ratio of C to H in the fuel will vary from 3 to 1 to 6 to 1, depending on the molecular weight of the hydrocarbon being burned. We can calculate the flue gas per pounds of fuel closely. Assuming a ratio of C to H of 5 3/4 (which probably is a little high but will compensate for the inaccuracy introduced by neglecting the carbon burned from the metal) we may con- clude that one pound of fuel will require nearly 17.1 pounds of air for combustion, yielding 18.1 pounds of gas made up of the following amounts of the several constituents: Pounds difficult. The standard design of annealing oven contemplates the introduction of the pots at the front of the furnace. The opening is closed by doors, usually made in sections which are equivalent to a front wall. Annealing Practice 197 At an early date attempts were made to render the pro- cess approximately continuous. Seth Boyden built a "shov- ing" furnace of which G. H. Kings land of the Wilmington Malleable Iron Works writes as follows : "The furnace was torn down under my direction. The pots were 12 inches high and 10 inches wide each way, with Fig. 92. The interior of the powdered coal mill of a modern malleable plant. The horizontal cylinder at the left is the dryer a bottom just like a box without a cover. These were placed on rollers, pots being pushed in at one end and shoved out at the other. I believe the furnace held 30 of these boxes, five wide and six deep. One row of five was shoved out each working day and a row of five pushed in. The furnace was about 2 feet high at the crown of the arch, with flues under 198 American Malleable Cast Iron 3 O M- ,0 5* ^ till- ttti ' $11** t Mfl ^ it ^ >i r\ iM) N V? . ^^^$ m 218 American Malleable Cast Iron gas enters at the top, this circulation will tend to divert the descending hot gas from the hotter passages toward the colder spaces and thus heat the latter more rapidly. Were the heat admitted at the bottom the. circulation described would cause an in- crease in the difference of temperature between the hot and cold passages since the ascending current in the hotter space would draw the hot incoming gases with it. Time of Heating Varies Widely The rapidity with which an oven can be heated uniformly depends entirely upon its construction. A number of observers have recorded heating, cycles with pulverized fuel as short as 18 hours, whereas the author frequently has witnessed periods as long as 100 hours and over, usually under adverse fuel condi- tions. In some cases the increased time is due to the impos- sibility of burning the coal rapidly, while in others the rate of downward distribution of the heat in the furnace is the limiting factor. In the latter case it sometimes is necessary almost to cease firing and allow the heat to equalize by conduction and radiation in order to avoid overheating the top pots. This pro- cedure is sound metallurgically but necessarily involves a waste of time. The desired maximum temperature having been reached as uniformly and rapidly as possible, the next step is to main- tain this temperature until the reactions within the castings have attained a state of equilibrium. This time depends upon the temperature chosen and upon the chemical and structural characteristics of the metal. In experimental determinations the time to reach actual equilibrium is long. Under favorable con- ditions it may be 20 or 30 hours at 1900 degrees, 100 to 150 hours at 1500 degrees and several hundred hours at 1400 de- grees. In practice the times are materially shorter because a slight graphitization of cementite may be relied upon in cooling through the higher ranges of temperature and also because equilibrium is approached more rapidly during the earlier stages than when it is nearly attained. Indeed it might be said that actual equilibrium is attained only in infinite time at any tem- perature. Under fairly favorable conditions in well conducted plants the time to reach equilibrium within commercial limits Principles of Annealing 219 may be roughly as follows: 1700 degrees, 25 hours; 1500, 50 hours, and 1450, 80 hours or possibly 50 per cent more under less favorable conditions. These general relationships already have been indicated in graphic form in Fig. 36 in Chap. III. The time required is approximately inversely proportional to the temperature above A^ for alloys high in carbon or silicon the time required is less than for those lower in these elements. The presence of excessive manganese or sulphur, or of some of the more un- usual elements may prolong the time considerably. Also it is believed that the rate of freezing and possible other variables in the previous thermal history of the metal have an effect upon the rate of graphitization. The combined carbon content at equilibrium is greater the higher the temperature, therefore the iron is not completely annealed at the expiration of the required time at the maximum temperature chosen. The carbon content, or solubility of carbon, as dependent on temperature has been definitely determined for metal containing about 1 per cent silicon. The relation is shown in Fig. 24. Therefore the anneal will not be complete unless the reaction is allowed to progress to equilibrium at or just under A\\. The Ar-L point in commercial iron probably is between 1340 and 1375 degrees Fahr. Approach Temperature Slowly One way to accomplish the desired result would be to drop the temperature quickly from the maximum to just under Ar when the reaction at the former temperature is complete and to main- tain that temperature below Ar^ as long as may be required to re-establish equilibrium at the lower temperature. This opera- tion will readily yield perfectly annealed material but is difficult to execute in practice except possibly in tunnel furnaces. Under commercial conditions, equilibrium can be attained more readily just under Ar^ by approaching this slowly from above at a rate permitting the graphitization to just keep pace with the falling temperature than by a quick drop and a long wait to establish equilibrium. Rates of cooling between four and 10 degrees per hour usually are desired and most operators prefer to cool more and more slowly as the temperature drops. 220 American Malleable Cast Iron To make sure of attaining equilibrium a number of an- nealers wisely attempt to hold a constant temperature just under Ar for some time. Nothing is gained by additional slow cool- ing after the reaction at Ar is complete. In many plants the cooling rate is determined by the heat radiation of the furnace. In these cases the annealer merely seals the furnace at the high temperature and lets it take care of itself. Fortunately, since the rate of cooling decreases as the temperature of the oven falls, a well insulated furnace cooling naturally will fall in temperature at a steadily decreasing rate, as the metallurgical theory required. Therefore the results of this practice often are much better than might be expected. Difficulties begin to arise when the cooling is accelerated by some unforeseen or unknown cause and the illogical operator is no longer able to account for his results. It will be noticed that a complete annealing cycle may be subdivided into five distinct intervals as follows: Heating to maximum temperature, maintaining maximum temperature till equilibrium is attained in graphitization of cementite, cooling to critical point, holding just under the critical point, and further cooling to permit handling. The first and last periods have no metallurgical significance and can be accelerated as much as is convenient. However, the second and the combination of the third and fourth, are determined by the product being manufactured and cannot be reduced below definite minimum values. The minimum cycle is divided as follows: Heating to 1600 degrees, 30 hours; holding at 1600 degrees 45 hours; cooling to Ar^ and holding there, 35 hours; and cooling to handle, 5 hours. The total is 115 hours, which would make a six-day annealing cycle as an absolute minimum, the time above 115 hours being spent in charging and pulling. However, few plants are able to insure success in so short a cycle and seven days may be considered as the commercial minimum. Cycles of nine days and more are not uncommon with large furnaces in order to secure the best results. The minimum annealing time is fixed by natural laws which cannot be changed to suit the wishes of the manufacturer or Principles of Annealing 221 the consumer. Any attempt on the part of the user to hurry the producer is misguided. The response to such pressure will be in inverse ratio to the conscientiousness and intelligence of the particular manufacturer concerned. It would seem that self interest will drive the malleable founder to adopt the shortest workable annealing cycle in order to avoid the in- vestment in additional ovens and their fuel supply. Nevertheless the author has known many purchasers of malleable who seemed to regard the operation of a long cycle as an arbitrary wish of the manufacturer imposed upon his customer without any adequate reason. For many years the larger producing interests have been approached from time to time by frequently sincere but always poorly informed inventors claiming either to much reduce an- nealing time or sometimes to do away with annealing entirely. As a rule, those in the former class expect to accomplish results by changes either in furnace design, methods of heating, etc., or by some unusual and often secret packing. Being an atomic re- arrangement within the metal itself, the annealing reaction can- not be accelerated or retarded by the material surrounding the casting. The laws governing graphitization have been investigated by a number of entirely competent experimenters and depend on clearly known chemical fundamentals. The design of heat treating furnaces also is well understood. Changes in furnace design could only reduce the annealing time by accelerating the time of heating, since as already explained, the times and tem- peratures during the rest of the cycle are fixed by the metal be- ing annealed. All of these patented or secret annealing methods therefore are foredoomed to failure. It is conceivable, although improbable, that someone will discover an alloy with a carbon content, similar to that now used, of such a character that graphitization. will be suppressed at temperatures above 1600 degrees Fahr. but which will graphi- tize easily or even spontaneously at lower temperatures. Such an invention would accelerate or eliminate the present annealing process. Since the alloys of iron with most of the reasonably common elements are constantly being investigated and no indi- 222 American Malleable Cast Iron cations have been found of any elements with properties pro- ducing the complex effect here described in any degree, it seems most unlikely that any greatly accelerated annealing meth- od for producing black heart malleable will be found. Therefore producers and consumers should admit the necessity of adequate time for annealing and conduct their several operations in accordance. The author is still waiting to hear from a most enthusiastic engineer who, three months before this was written, offered to demonstrate the manufacturer's ignorance of annealing principles by taking home a sample of hard iron in the evening, annealing it over night and returning it completely annealed the next day. Other incidental changes are produced in the metal while graphitization is going an. The clearest evidence that these changes are only incidental is the fact that the process of graphitization can be carried on perfectly without any gain or loss of weight. To prove this, an accurately weighed speci- men of hard iron can be enclosed in a tube of difficultly fusible glass, the air displaced by hydrogen, the hydrogen pumped out to a fairly low pressure and the tube then sealed, so that the metal can be annealed surrounded by nothing but a trace of a reducing gas. Samples of 10 or 12 grams weight annealed in such a tube in accordance with the heat cycle of commercial practice, are unaltered in weight to 1/10 milligram. In other words, the weight remains constant to 1/1000 of 1 per cent. Migration of Carbon However, under commercial conditions the castings always are in an atmosphere having oxidizing possibilities. This at- mosphere may be the atmospheric air remaining in the spaces not otherwise occupied or it may be the products of combustion or gases arising from reactions with packing materials. There- fore there always is a tendency toward burning out the surface carbon. The mechanism of the removal is interesting. Only the carbon in the outer layer of molecules can combine directly with any oxygen in the surrounding gas. Therefore unless either the gas can penetrate the solid metal or the carbon can migrate to the surface, decarburization would be limited to the Principles of Annealing 223 infinitesimally small amount produced by burning out the car- bon one molecule deep. At one time it was generally believed that the gas penetrates but the migratory action certainly exists and is probably the .01 .oa .03 .04 .05- .06 .07 .06 .09 .10 .// Inches Be/ow Surface Fig. 99 Increase in carbon content at increasing depths below the sur- face of malleable cast iron major method by which carbon and oxygen are brought to- gether. Carbon exists in iron at any temperature above Ac z in part, as a solid solution of a definite saturation value at any given temperature. If the carbon concentration is locally low- ered below saturation, diffusion will enrich this area at the ex- pense of the more highly carburized areas. So long as ce- mentite, or undissolved iron carbide remains, the deficit will 224 American Malleable Cast Iron : be made up by solution of additional amounts of this element in such a quantity as to maintain the solid solution in a saturated state. This migration requires considerable time so that in gen- eral, carbon is oxidized at the surface much more rapidly than diffusion can equalize the carbon content. The result is a ma- terial poorer in carbon at the surface than in the center. As we go further toward the center, the increase in carbon content corresponds to a sort of gradient which is sufficient to feed the carbon to the surface as fast as it is removed. Fig. 99 shows the increase in carbon content at increasing depths below the surface. The graphs represent various de- grees of decarburization under commercial operating conditions. It will be noted that the graphs vary both as to carbon con- centration at the surface and as to the depth of penetration. The former depends somewhat on the oxidizing medium em- ployed, the latter on the length of time, the medium is applied, and on its activity. The effect of this decarburization on the physical properties of the product are relatively small. Fig. 100 shows graphically the results of careful tests made to determine the effect of the removal of 1/16 inch of carburized surface in specimens of various diameters on the tensile properties of the metal. The experiments were conducted by casting tensile specimens to a series of diameters, grinding one specimen of each size truly cylindrical, removing about 1/16 inch of stock. The ground specimens then were annealed with rough specimens from the same heat and turned to size after annealing. The graphs show the amount by which the properties of the specimen ground before annealing exceeds the corresponding properties of the turned specimens. The experiment was conducted in this form to eliminate variations due to cooling rate and original rough surface which variables are included in the data given in the chapter on tensile strength. The tests were conducted on one lot of metal, all annealed together. Therefore they correspond to one set of decarburiz- Principles of Annealing 225 ing conditions only. Since decarburization varies, as the an- nealing conditions vary, another series of investigations was made to determine the changes in properties in iron of initially similar composition by variable decarburization. D/omefer Of Specimen Inches Fig. 100 Graph showing effect of removing 1/16 inch decarburized surface in specimens of various diameters on the tensile properties of the metal - Results of 50 Tests In Fig. 101 have been plotted the results of some 50 such tests on iron having from 2.40 to 2.60 per cent carbon, 0.70 to 0.80 per cent silicon before anneal, which correlate the tensile properties with the carbon content after annealing. The graph 226 American Malleable Cast Iron is plotted from average values. Individual tests depart con- siderably from the average since small differences of carbon con- tent in the hard iron affect the results much more than much larger variations in this element due to decarburization. 1 id :l 9m . *> a^M^ '1 ~- )U ^ =S= = =: "* ^ssooo X j^SSOO s ^ ^ x, c>* y. V * f/000 ? V S s. O(JU(/(J ^ 49000 SfSS j Durir t Ann Tof a/ Carbon Be fore Ann ea/ .&O/^ on Oxidized /.5O U5 WO .7? ^gAnneo/ino *orbon After LO /.Z5 /.50 /,7S eal/ng Fig. 101 Graph showing effect of varying degrees of decarburization on tensile properties of malleable cast iron Figs. 100 and 101 serve to show that the final properties are relatively little affected by the decarburization process. Be- ing measured on surface metal the elongation probably depends only in the carbon content near the surface and but little on the depth of decarburization. The tensile properties are some- what more consistently affected by decarburization. Decarburization is controlled in practice by the character of the packing material. Perhaps it would be more accurate to say that the results in practice depend on the packing used, there Principles of Annealing 227 being but little available information with regard to the action of packing. The commercial packings depend for their activity chemically on the reduction of ferric oxide, Fe 2 O 3 to FeO, ferrous oxide. It is not to be understood that they actually liberate oxygen on heating as for instance potassium chlorate does. Four Possible Reactions The process is a chemical reaction in which the oxygen never appears as such but merely combines with carbon. Four re- actions are possible, depending upon the circumstances: 3 Fe 2 O 3 + Fe 3 C = 2 Fe 3 O 4 + CO + 3 Fe 6 Fe 2 O 3 + Fe 3 C = 4 Fe 3 O 4 -f CO 2 + 3 Fe Fe 3 O 4 + Fe 3 C = 3 Fe O + CO -f 3 Fe 2 Fe 3 O 4 + Fe 3 C = 6 Fe O + CO 2 + 3 Fe The two reactions 'FeO+Fe 3 C=Fe+CO+3Fe and 2 FeO -J-Fe 3 C=2Fe+CO 2 +3Fe are theoretically possible but occur only under unusual circumstances, if at all. The reaction 3FeO+50O Fe 3 C+4CO 2 can probably oc- cur under certain unusual conditions. The fact that the analysis of packings is expressed as a rule in terms of the Fe 2 O 3 FeO, SiO 2 and possibly A1 2 O 3 and other oxides has given rise to the unfortunate conception that they are mixtures of two oxides of iron with other inert oxides. As a matter of fact all packings in use, as distinguished from the raw packing, have become complex silicates. The practical annealer unconsciously acts on this knowledge where he limits his additions of roll scale, or other raw material to small amounts at any one time, for a packing containing any large amount of free oxides is not a workable material. The raw material from which packing is built up usually is roll scale or squeezer scale from rolling mills, pot scale (the oxide from the. outer surface of the annealing pots after they are drawn from the furnace) or air furnace slag. Iron ore was once used but probably is now obsolete. Table XVI shows the composition in the usual terms, of these several materials. It should be understood, however, that only the first three are actually oxides. Ore is nearly pure ferric oxide contam- 228 American Malleable Cast Iron Table XVI COMPOSITION OF TYPICAL PACKINGS FeO Fe 2 3 MnO SiO, A1,O, Ore 00 91.43 8.57 Pot scale 37.10 53.11 9.79 Roll Scale 61.47 31.99 6.54 Squeezer scale 69.74 9.34 .80 14.95 5.17 Slag 28.80 1.16 4.85 50.42 14.77 inated somewhat with silica minerals. Pot scale is a more or less impure magnetic oxide, Fe 3 O 4 contaminated by sand adher- ing to the pots. Roll scale is magnetic and ferrous oxide originally nearly pure, but contaminated in gathering it up and shipping. Squeezer scale is a mixture of basic silicates of iron and manganese with some iron oxides, mainly ferrous oxide dissolved in bibasic ferrous silicates. Slag is a neutral silicate contaminated with fused brick, etc. Some typical analyses of packings as actually used are shown in Table XVII both in terms of the usual proximate analysis and in terms of the compounds apparently present. It will be seen that the packings contain little free oxide and are mainly silicates. The ferrous silicates are incapable of reduction to metallic iron under the usual annealing conditions so that the oxygen for oxidizing the carbon is derived primarily from the reduction of Fe 2 O 3 to FeO although the ferrous oxide of pot and roll scale may enter into the reaction. The relative amounts of carbon monoxide and carbon di- oxide formed depend on the temperature and the packing used. With the materials and temperature of commercial practice the ratio is fairly constant; approximately 12J^ per cent of the car- bon being burned to CO 2 the remainder to CO. The principal reaction involved, assuming Fe 2 O 3 as the ac- tive medium, corresponds to the equation: 9 Fe 2 O $ -f 8 Fe 8 C = 18 FeO + 7 CO + CO a + 24 Fe The actual mechanism of the decarburizing reaction forms an interesting though complex problem in physical' chemistry. The oxidation of the carbon in the iron and reduction of the Principles of Annealing 229 packing are accomplished by the gas surrounding both. To be operative, a system must be chosen so that at the temperature and pressure in the .annealing pot the gas phase present is such that the reactions Fe 3 C + CO a = 2 CO + 3 Fe Fe 2 O 3 + CO + = 2 FeO + CO a FeO + CO =Fe + CO a can all proceed from left to right. In other words the system must be one in which a ratio of CO to CO 2 can be maintained which will at the same time oxidize Fe 3 C, reduce Fe 2 O 3 , and reduce FeO. If the relative concentration of CO and CO 2 be such that the first reaction ceases or reverses no decarburization will occur. If the reaction is initiated it would soon cease, due to the conversion of all available CO 2 to CO, unless the second re- action continuously reconverted CO to CO 2 . If the last re- action reversed, the iron of the casting would be oxidized in addition to the carbon in the consequent scaling . Only some of the more usual reactions have been considered there being a Table XVII ANALYSES OF PACKINGS Source Pot scale Roll scale Squeezer scale Slag Fe 4.04 6.88 FeO 54.36 57.33 58.49 38.25 Fe 2 s 9.04 5.97 3.14 1.03 MnO 1.50 3.03 SiO, 21.02 26.16 24.92 43.60 Al a O, and undetermined ..11.54 9.66 11.95 14.09 Proximate Composition of Above Per* cent Fe 4.04 6.88 FeO . 23.40 31.34 21.00 12.60 Fe 2 3 5.97 1.00 (FeO) 2 SiO, 40.90 56.90 (Fe 2 3 ) 2 (Si0 2 ) 3 ..14.10 5.40 (FeO), (Si0 2 ) a 27.91 ..... FeO Si0 2 45.20 Fe 2 0, (SiO,), .... 5.40 Various inert silicates by difference . ..17.56 27.90 16.70 35.80 230 American Malleable Cast Iron number of others possible between the components of such a system. Scientific investigations of the subject matter involved would be based on determination of the composition of the gas phase in equilibrium with the several oxides of iron and carbon con- cerned and a location as to temperature and concentration cor- Fig. 102 Equilibrium curves illustrating the reactions between carbon, iron and oxygen, after the data of Matsubara responding to the reactions proceeding in the desired directions. The subject has been but imperfectly studied, the available in- formation being mainly due to Schenks' summary "Physical Chemistry of the Metals." Matsubara, in a paper presented before the American Institute of Mining and Metallurgical Engineers, February, 1921, amplifies and checks Schenks' data, particularly with respect to the reactions into which the cementite enters in the presence of CO and CO 2 . Principles of Annealing 231 Fig. 102 is drawn from Matsubara's paper, based on his own results as well as those of Boucourd, Bauer, Schenk and others. It represents the percentage of CO in a mixture of CO and CO 2 for various temperatures at which the several reactions will proceed equally rapidly in both directions or at which they will cease and equilibrium will be established. The graphs are plotted for a pressure of one atmosphere as the sum of the partial pressures of CO and CO 2 . For other pressures the equilibria can be calculated from the equilibrium constants of the several reactions. Letting P be the pressure exerted by CO and CO 2 , X the amount of CO in the mixture of these gases and K lf K 2 and K 3 the equilibrium constants for equations 1, 2 and 3, respectively, then X* K,= P i 1 X X* K 2 = P \X X s K,- P dxy KI K 2 and K 3 can be calculated from Fig. 102 for any giv- en temperature and hence the change produced in X by changes of pressure at that temperature can be calculated and a dia- gram similar to Fig. 102 constructed for other pressures. Reaction (4) and (5) are independent of pressure. Un- fortunately nothing is known as to the locus of the curves cor- responding to (4) and (5) for the silicates forming commer- cial packings. The interpretation of the equilibrium diagram to determine what reactions occur is as follows: On areas below (3) cementite is oxidized to FeO and CO; in areas above (4) FeO is reduced to Fe with the formation of CO 2 , hence in any region below (3) and above (4), FeO will oxidize the carbon of cementite. Such regions exist only above 700 degrees Cent., therefore the reaction cannot be main- tained at lower temperatures. That the lines (1), (2), (3) and (4) should intersect at one point is curious, and indicates that at that temperature, pressure and composition, C, Fe, FeO and Fe 3 C or any two or more of these radicals can exist together 232 American Malleable Cast Iron in equilibrium. Almost any question as to the course of the annealing reaction or the behavior of packings could be answered .by the construction of such diagrams for the particular packing material. Many conclusions as to the reactions of the pure ele- ments and their oxides and carbides will present themselves on further study of the diagrams. XII PATTERNMAKING AND MOLDING IN MANY respects, patternmaking and molding practice in malleable plants does not differ from that in other branches of the foundry industry. The various devices adopted for repetitive work in gray iron or brass also are found in use in the malleable shop. Indeed, since the producers of malleable engage largely in the manufacture of small and moderate sized parts in large numbers the development perhaps is further ad- vanced than in gray iron practice. However, there are certain vital differences between patternmaking and molding for mal- leable cast iron as distinguished from the same operations in the gray iron trade. These differences arise from the metal- lurgical properties of the two materials. The two essential distinctions between white iron and gray iron lies in the melting point and shrinkage of the two metals. Gray iron castings of moderate size are made of metal con- taining, for example, 3.25 per cent carbon, 2.00 silicon and 0.50 phosphorus as compared with the composition of white cast iron which approximates 2.50 per cent carbon, 0.75 silicon and 0.19 phosphorus. The equilibrium diagram for the iron carbon alloys shows that all alloys above 2 per cent in carbon finish, freezing at the same temperature 1130 degrees Cent, or 2066 degrees Fahr. It shows further that the point where freezing begins varies with the carbon, decreasing nearly uniformly from 1550 to 1130 degrees Cent, as the carbon increases from nothing to 4.3 per cent. Leaving the other elements out of consideration, the white iron should begin to freeze at roughly 1310 degrees Cent, or 2390 degrees Fahr. and the gray iron at 1220 degrees Cent, or 2250 degrees Fahr. Thus gray iron will be completely liquid at a temperature 140 degrees Fahr. lower than that at which white cast iron has begun to solidify and .the range of partial solidification or pasti- 234 American Malleable Cast Iron ness is larger by that amount in white cast iron than in gray iron. The presence of silicon still further accentuates this point. According to Gontermann's data, metal of the composition as- sumed for gray iron should begin to freeze at about 1200 de- Fig. 103 (Above) Two gates of metal patterns in match part; (below) Pattern mounted on match plate and gated pattern mounted on vibrator frame grees Cent, or 2190 degrees Fahr. and be completely frozen al 1140 degrees Cent, or 2080 degrees Fahr., whereas white cast iron should begin to freeze at 1330 degrees Cent, or 2420 de- grees Fahr. and finish the process at 1170 degrees Cent, or 2140 degrees Fahr. The data are not exactly in accord with those based on car- Patternmaking and Molding 235 bon alone, due to minor differences in the observations on which the data were based. The point to be clearly brought 'out is the higher point of incipient freezing and longer partially frozen range for white cast iron than for gray iron. The presence of phosphorus in larger amount in the latter still further accentuates the difference, although the writer has no available data on the freezing conditions in the system Fe-Si-P-0. The data given show clearly that white cast iron must be poured at a much higher temperature than gray iron, since the latter will be liquid at a temperature perhaps 230 degrees Fahr. below that where the former has begun to set. Further- more, it is quite possible that the fluidity of white iron when at a temperature say 100 degrees Fahr. above its freezing point is materially less than that of gray iron at the same temperature above its freezing point. Within the author's knowledge data on this point are lacking. A further corollary of the difference in freezing conditions is that other things being equal there will be more shrinks or porous areas in white than in gray iron castings. This arises from the longer freezing range of the former corres- ponding to a larger fluid contraction of the still liquid alloy between the time and temperature of incipient and complete solidification. The consequence of this increased fluid, contrac- tion is that as the temperature of complete freezing is ap- proached there no longer remains a sufficient volume of liquid to fill the voids in the previously formed solid skeleton. Therefore, in the last freezing areas, voids remain between the dendritic crystals of the first frozen solid. The shorter the freezing range the less of this contraction can occur. It has been shown by Cesaro that liquid iron is a solution of cementite in iron and Wust and Peterson have demonstrated that all such alloys freeze as cementite and austenite. However, in the temperature interval just under freezing the higher silicon and carbon metals graphitize by the conversion of cementite into iron and carbon. 236 American Malleable Cast Iron Fig. 104 Hand operated squeezer-type molding machine and (below) mold and pattern equipment in position on machine. Heavier machines operated by air also are used in the industry Patternmaking and Molding 237 The iron resulting from this reaction occupies almost the same volume as the original cementite. The total volume therefore is increased almost by the volume of carbon liberated. As a consequence there is a tendency to expansion at these high- er temperatures. A number of observers especially Turner have recorded actual increase in linear dimensions while the metal was cooling and therefore contracting, just under the freezing point. The expansion due to graphitization is important in two respects. It causes the casting to be only about 1 per cent smaller in linear dimensions (3 per cent by volume) than the pattern instead of double these values for white iron, and also tends to fill up in part the voids left by fluid contraction. The difference in pattern equipment and molding methods in the malleable as compared with gray iron industries is due to the necessity for providing against the following differences in the properties of the two metals. 1. The higher melting point and lower fluidity of white iron. 2. Its greater tendency to internal shrinkage due to fluid contraction. 3. Its greater shrinkage from pattern size. It will be noticed that the noun "shrinkage" has two distinct but related meanings to foundrymen. One refers to the reduction in the overall dimensions of the casting as com- pared with the pattern and the other to the production of porosities due to voids left by the contraction of the fluid metal. A distinction based on the words "solid contraction" and "fluid contraction" seems desirable but has not- gained favor among foundrymen. Accordingly one must be constantly on the alert to avoid confusion due to the indiscriminate use of the term "shrinkage." Speaking first of this property in the sense of solid con- traction, the fact that the shrinkage of white iron is about ^4 -inch 238 American Malleable Cast Iron Fig. 105 Stripper and roll-over molding machines (Top) Plain stripper plate molding machine and equipment for cope and drag. (Center) Roll-over machine for drag. The cope is rammed up from a plain plate. (Bottom) Stripper plate machine for cope and a roll-over machine for the drag. Patternmaking and Molding 239 per foot instead of ^ -inch per foot as in gray iron does not cause any difficulty in patternmaking, except that a proper allowance must be made by using a "double" or 54 -inch shrink rule in laying out the work in case the casting is to be used hard. This shrink rule is merely a rule graduated in feet and inches and fractions of inches usually sixteenths in which the distance marked as one foot is 12.25 inches. A casting from this pattern will come from the mold about true to size. Experiment has shown that the solid contraction of white n Length in Per Cent of Length at 75 F. Co KJ O^ O / / . / ^sV/ / *4 \Y ^ ^ c SL Total Contraction Independent of Chemical Composition.Data on Samples of at>out 2^Tbtal Carbon No Graphite C7* 0.4 / / 7 00 1600 IZOO 800 400 Temperatures, Deq. Fahr. Fig. 106. Curve showing contraction in cooling from solidifica- tion to room temperature cast iron (metastable carbon iron alloys) is substantially the same irrespective of composition. The contraction in cooling from solidification to room temperature, is graphically shown in Fig. 106. On annealing the casting expands due to the fact that tem- per carbon and ferrite occupy a considerably greater volume than the cementite from which they are formed. The increase in volume and in linear dimensions, depends primarily on the original total carbon and to a less degree on the heat treatment 240 American Malleable Cast Iron by which the graphitization is attained and possibly on other more obscure circumstances. Some conclusion as to the changes of dimensions produced by graphitization can be formed from the following density data: Ferrite 7.90, cementite 7.438, carbon 2.30 to 2.70. Dimensions Determined by Trial The expansion in annealing is usually assumed to be one- half the original contraction making the net "shrinkage" allow- ance Y% inch per foot as for gray iron. This conclusion can be correct for only one particular carbon content. It was probably 3.10 500 0?QO \ ^X, \ \. >s \ "O |2BO c c270 _Q 5260 25C 3.40 X \ "S \ \ \, I 5 i ! f3 \ \ 1.0 1.10 1.20 130 1.40 1.50 1.60 Per Cent Contraction of Malleable Specimanfrom Pattern Size Fig. 107 Graph showing the per cent of contraction of malleable from pattern size fairly accurate in the days when high carbon iron was prevalent. W. L. Woody has given the writer data obtained in a study of over 1000 heats from which test specimens were cast from a pattern 12 inches long, the specimens being micrometered after annealing. The results are shown graphically in Fig. 107. The percentage of net shrinkage of unconstrained specimens can.be read from this graph. The author has determined the density of hard iron and malleable cast iron made therefrom for various carbon contents. Patternmaking and Molding 241 The data are shown in Fig. 108 calculations as to change of di- mensions in annealing from these changes in density yield re- sults apparently in error in the direction of too much expansion in anneal, i.e. to too small a shrinkage allowance. In determining pattern dimensions consideration must also be given to the fact that, due to rapping, the molds always are larger than the pattern, except on ''stripper plate" equipment. On vibrator plates this "rappage" will be small and uni- form, in bench and floor molding by hand it will be variable and may be large. Very small parts may actually require a negative "shrink- 7.7 d 75 y 74 CD Q. ^73 7.2 2.3 2.5 2.7 2.9 Per Cent Carbon In Hard Iron 3.IO Fig. 108 Graphs showing relation of annealing upon the density of the metal age allowance" "the rappage" exceeding the solid contraction. Further it may happen that in irregular and intricate cast- ings some parts constrain others when freezing and leave shrink- age strains. The relief of these strains during the annealing may cause unexpected changes of form. Therefore it often is necessary to arrive at the pattern size for important dimensions by actual trial and even then the castings will come true to size only so long as temperature of 242 American Malleable Cast Iron Proo>/e for/7?or/bn Of Crock Fig. 109 Casting with thin disk and thick hub, showing probable point of rupture pouring, chemical composition, and sometimes even the solidity cff sand and cores are maintained exactly constant. The heavy solid contraction of the white cast iron also im- poses a number of difficulties which would not be clear to the reader were he to consider the problem altogether from the standpoint of the net shrinkage of the finished product. It has been said that the total contraction of all white cast iron is constant. However, it is at least unusual that all parts of a given casting cool at the same rate. In other words, in prac- tically every casting some parts arrive at their final temperature, and therefore final size, ahead of others. This may develop ex- cessive stresses or even distort or disrupt the casting. Consider a casting having the form of a thin disk with a heavy hub at the center, as shown in Fig. 109. The hub will be hot and possibly almost fluid when the light disk has already set and cooled to nearly room temperature. The contraction of the disk during the cooling has met but little resistance from the hot plastic center. However, when the latter begins to cool its reduction in dimensions will be resisted by its attachment to Fig. 110 Type of casting with thin disk center and thick rim Pafternmaking and Molding 243 the solid thin flange. Sometimes this attachment will be so se- cure as to permanently stretch the pasty mass within. If this cannot occur the flange may be torn loose from the hub at one or more places or may even be entirely detached. In the reverse case of a thin plate surrounded by a thick rim, as shown in Fig. 110, the contraction of the rim would be opposed by the previously solidified center, either crushing the center or producing a radial tear in the rim. Generally the point of failure is at or near the hottest part of the -casting Where the strength is the least. Occasionally no external de- fect results due to the welding up of such defects by molten metal from the center. Then the consequence is a pipe or other void. The magnitude of the stresses from this source may be enormous, depending only on how rigidly the last cooling por- tions are held by their solid surroundings. In gray iron the difficulty is less pronounced due to the lower magnitude of the contraction and to the fact that the solid portions can be deformed slightly without breaking, whereas practically no dis- tortion is possible in the hard iron. Effect on Design of Castings The practical application of this reasoning is that, in the design of parts to be made of malleable cast iron great care must be used to avoid such forms and proportions as will rigidly connect parts of widely different cross section. All sections should merge uniformly into each other, avoiding abrupt changes of thickness. Fairly thin ribs intended to rigidly brace heavier sections, spoked wheels with hubs heavier than the rim and in general any design in which unequal rates of cooling can set up opposing stresses should be avoided. If such designs are suc- cessfully executed by the foundryman it is only by methods 'of gating or chilling calculated to accelerate the cooling of the heavier sections and retard that of the lighter. This calls for the exercise of great skill and judgment and may produce pro- hibitively higher losses with a corresponding increase in cost. We may now consider the shrinkage produced by fluid con- traction and resulting in porous material in the areas freezing 244 American Malleable Cast Iron last. It is impossible to suppress these so called shrinks in any casting. Their formation is inseparably connected with selective freezing over a temperature interval and hence always occur in every casting. Depending on particular conditions, these shrinks may be widely distributed in insignificant amount at any one place, or they may be concentrated in one spo't, aggregating a consid- erable volume. A casting freezing at a nearly uniform rate throughout, due to equality of section, etc., and freezing almost as rapidly as the iron enters the mold may have the porosity so uniformly distributed and so nearly filled up from the ladle during pouring as to be practically sound. On the other hand, a casting having a heavy cross section in some one place which Fig. Ill Dendrite (about half size) from shrink in hard iron ingot 8 inches in diameter by 20 inches high which was poured without feeding is fluid long after pouring ceases will show a great shrink, especially if the heavy section is high up in the mold. Two remedies are employed for this trouble. The older is the application of iron chills, which are pieces of cast iron buried in the mold so that they form its inner surface at the points where shrinkage is prevalent. By accelerating freezing they suppress the shrink in their immediate vicinity. However, since the reduction in volume still exists an equal volume of shrinkage will develop elsewhere. This practice is good if the shrink in the new location does no harm, or if in that lo- cation it can be suppressed by feeding ; otherwise it is merely camouflage. Continuously supplying molten iron until the en- tire casting is frozen is the only actual preventive of shrinks. Patternmaking and Molding 245 The shrink always is found in the slowest freezing locality. Therefore, if to the pattern there is attached a feeder of still slower cooling rate so located that metal can flow from it to the location in which the shrink was found, then the shrink will be transferred to this feeder and be of no consequence, since the feeder is not a part of the finished product. The actual design of feeders, to meet a given set of conditions may require much skill and experience, but the operating principle is simple. t Feeders are expensive, not only from the molding view- point but also because they involve the melting of much ad- ditional iron. Nevertheless their use is the safest possible found- ry practice to insure sound castings. The high freezing point of white cast iron necessitates much greater care in gating than is requisite for gray iron. The relatively thin gates commonly used for that metal do not admit of a sufficiently rapid flow to prevent freezing before the mold is filled. Most castings must have metal admitted at a number of points in order to permit the mold to fill sufficiently rapidly. Because of the large gates, it is necessary to use special means to exclude slag or sand floating with the current of metal. The thin knife gates of the gray iron industry will choke the stream enough to permit these impurities to rise to the surface and be trapped in the runners. The same principle is used in malleable foundry but greater care is necessary in making the runners large and providing places for the ascending slag to be trapped on account of the rapid flow of iron required. Frequently the iron is poured through a strainer core placed at the bottom of the riser, which is intended to cause the latter to remain full of metal and allow the slag to accumulate and float up. (Because of the quick filling of the mold, necessitated by the quick freezing of the iron, great care must be used in se- lecting molding sands, and in venting the mold. The air and gas must be able to escape rapidly enough to allow the iron to enter at the rate required to keep it from freezing before the mold is filled. The selection of molding and core sands of core binders, 246 American Malleable Cast Iron as well as the actual ramming of the sand are further influ- enced by the high solid contraction of white cast iron. The mold and cores must be made so as to give readily under the heavy contraction of the casting in freezing. If for instance, a core be so hard as not to disintegrate before the metal begins Fig. 112 Typical gate for malleable castings showing strainer, core and skimmer gates for furnishing clean metal for feeders and producing sound castings to shrink it may set up such a strain in the casting as to actu- ally cause rupture. The patternmaker can frequently save the customer money by a judicious selection of the number of pieces made in one mold. A reasonable increase in the castings per mold is good economy. f Any attempt to increase the weight per mold by putting in so many pieces as to cause pouring difficulties or to prohibitively increase the dimensions of the mold it not justifiable. Patternmaking and Molding 247 In general the steps in the improvement of molding meth- ods have been as follows: Starting with a plain pattern as the simplest equipment, the first step was to permanently attach thereto models or pat- terns of the gates, feeders, etc., in order that these need not be the subject of separate operations. In the case of small parts this leads to the mounting of several patterns on one gate. To avoid the labor of producing a parting by hand for each mold, match parts were introduced, which are merely a semi- permanent duplicate of one half of the mold (generally the cope). In the interests of greater stability plate patterns were developed, consisting of fairly thin flat plates, usually of alumi- num with the patterns mounted on one or both sides to- gether with the gates, etc. The plate being at least as large as the exterior of the flask separates the cope and drag by its own thickness. Each half of the mold being rammed up off its own side of the plate, the mold when closed corresponds in form to the parts mounted on the plate. To do away with hand-rapping the pattern to withdraw it from the mold; air or electric vibrators often are attached. In some cases, especially for heavy work, the pattern is with- drawn, usually by a lever motion, without rapping, through a stripper plate. The stripper plate is merely a plate represent- ing the parting of the mold having an opening exactly fitting the contour of the pattern at the parting. When drawing the pattern downward through this plate the latter supports the sand and prevents its following the pattern. Unless the cope and drag are duplicates, two machines are requisite for each job as the construction is evidently such as to be applicable to one-half the mold only for each unit. Extremely heavy work is frequently handled on a roll-over machine which is especially available for making the drag. After the drag is rammed up, necessarily parting downward, the ma- chine facilitates turning it over to its proper position by sustain- ing and counter balancing most of the weight of the mold and pattern by springs. The pattern is sometimes withdrawn 248 American Malleable Cast Iron through a stripper plate and sometimes by letting the mold sink away from under the pattern by a suitable lever motion. The sand is compacted by hand ramming, by the use of hand or air operated squeezers, and by jolt ramming. The latter operation consists of mechanically raising the mold repeatedly and allow- ing it to come down on a solid support which uses the inertin of the sand itself for compressing it. On floor work pneumatic rammers sometimes are used. XIII CLEANING AND FINISHING OPERATIONS of cleaning and finishing malleable iron castings are conducted in part by the manufacturer, but frequently also by the consumer. Some of the simpler operations may be dismissed almost with a word but certain others such as machining, welding, galvanizing, etc., which are performed usually after the castings are delivered to the buyer merit more extended discussion. Castings generally are cleaned of sand as the first step on leaving the foundry. An exception to this is found in some cases of large muffle annealed castings where the finish is relatively unimportant. Such castings are often annealed with out cleaning. In most cases, the hard iron castings are cleaned in tumbling barrels, using any of the standard equipment. The operation is in no sense distinctive, the only peculiarity being the brittleness of the castings. To avoid breakage greater care must be used in handling the material and packing the barrels than would be needed in gray iron practice. Castings of a very fragile character can not be cleaned in this manner without breakage. Therefore, it is usual to pickle or sand blast them, usually the former. Pickling may be in dilute sulphuric acid which loosens the sand largely by the ac- tion of the hydrogen gas formed on the surface of the metal or less commonly in hydrofluoric acid which dissolves the silica sand with but little action on the iron. If the latter acid is to be used, economy will dictate the mechanical removal of as much sand as possible before pickling to avoid -the needless exhaustion of the acid through the dissolving of loose sand. Castings Must Be Cleaned Large castings are sometimes sand blasted one at a time by hand more easily and safely than they could be cleaned by rolling. A second cleaning is practically always necessary after annealing and this may be by rolling, often using scraps of leather, old shoes, etc. to impart a polish. If clean cut edges 250 American Malleable Cast Iron Fig. 113 Tumbling barrels are used for cleaning castings Cleaning and Finishing 251 are required, sand blasting is often resorted to either in barrels or by hand. Pickling is not common except as a preliminary to plating. Sulphuric acid, hydrochloric acid, and a hot solu- tion of acid sodium sulphate may be used to remove the oxide scale left by annealing. Since the castings are very likely to become warped during the anneal a straightening operation is often necessary if the castings are at all complex in shape. In many cases, especially on complex and thin work, no better method can be used than the hand method. When pos- sible a drop hammer fitted with suitable dies may be employed. Since the development of arc and acetylene welding, the practice of reclaiming defective material by this process has received at- tention both by the producer and the consumer. The operation of welding has two entirely different aspects, the repair of me- chanically unimportant faults of surface and finish in the pro- ducer's plant and the repair of castings broken in service. Reference will be made later to the latter process, that is welding by or for the ultimate consumer. Limiting ourselves for the moment to welding as practiced in the malleable foundry, we may start with the premise that the founder should deliver to the buyer no casting which is not high-grade malleable iron through- out. In welding, the material of the weld is melted and the cast- ing, in part at least, is brought to this same temperature. Thus in welding with iron, regardless of whether the filler is wrought iron, soft gray iron or any other material, the casting will be heated to a point far above the critical point and hence on cooling will revert to the condition of white iron.. No in- genuity in the selection of a filler therefore will overcome the presence of a glass hard spot at the weld. This condition can be obviated only by using for a filler either white cast iron or malleable, more conveniently the former, although both will be white after remelting. If the welded casting is then annealed, or re-annealed precisely as in the regular practice the material in the weld will be the same as that throughout the casting. The temperature of the arc is so high that a thin layer of metal can be melted and the operation completed before the un- 252 American Malleable Cast Iron Fig. 114 Sand blast equipment is used for removing sand from castings Fig. 115 Sorting and inspecting small castings are important operations in many plants Cleaning and Finishing 253 derlying metal is much heated. The author once had the op- portunity to observe the work of an expert arc welder. Work- ing on castings retaining their original ferrite surface, this oper- ator was able to weld so rapidly using Swedish iron wire, that the heat was confined to the ferrite layer and hence a perfectly soft weld resulted. Such a result presupposes two conditions not usually existing; the first, the use of an extremely skillful artisan and the second, a character of repair which does not re- quire welding to" a part of the casting below the decarburized skin; The latter condition, depending as it does on the char- acter of defect to be repaired, is entirely beyond control. All Faults Not Cured by Welding Whether or not the casting is annealed before welding has no effect on the final product and may be left to the welder's discretion. Welds made in the above manner by a skilled ar- tisan will render the product equal in quality to an initially perfect casting. Since the element of skill enters, however, it may be a measure of safety to exclude from repair by welding, faults which if not perfectly repaired would be the cause of serious failures. Generally, snagging or the grinding away of gates, fins, etc., is the duty of the producer. The operation is performed either with the casting in the hard state or after annealing. Usually most of these imperfections can be broken off with a light ham- mer before annealing and the final finish produced by grinding. Grinding before annealing is slower and more expensive than if performed on the finished product. But since the former method produces somewhat better looking castings, especially on sand blasted work, it is sometimes specified when the con- sumer feels that this feature is worth the extra cost. Hard iron is ground on a very hard and rather fine grained emery wheel; malleable is ground on a soft and coarse wheel. The size of casting and finish required influence the selection of the exact grade of wheel. For malleable grinding wheels of artificial alumina, 14 and 16 grit, in a hard grade are used ex- tensively. The preceding discussion covers the usual finishing opera- tions which the malleable foundry performs for its customers, 254 American Malleable Cast Iron however, the customer may perform a number of additional oper- ations. Disk grinding, machining, straightening, welding, tin- ning, galvanizing, electro-plating, occasionally local hardening and possibly other operations come into this category. Since the customer's requirements and method are likely to be peculiar to his individual conditions, he is better informed as to his processes than is the manufacturer of the castings. It will be well, therefore, to confine the present discussion to considera- tions of the producer's attitude toward these several operations. Of the technique of disk grinding little need be said, the one essential point to be observed being that in this as in all other forms of grinding the operation be not crowded to the point where the temperature of the surface metal reaches Ac^ Many grinding operations will readily raise the metal in con- tact with the wheel to a red heat. A portion of a malleable cast- ing which has risen to such a temperature has had some of its carbon recombined and has been locally hardened to a degree which may render it brittle or unmachinable. Should Allow for Finish Theoretically, tool life should be long and cutting speeds high for malleable cast iron, since the material be- ing cut is a dead soft steel which is one of the easiest ma- terials to machine. Moreover, the presence of temper carbon should favor machining both by breaking up the chip and by acting as a lubricant for the chip and tool. That this conclusion is correct is indicated by the con- ditions under which malleable is machined in practice. In ma- chining malleable cast iron not much over 1/16-inch of stock is removed at one cut. Only in rare cases are cuts of %-inch to 5/32-inch necessary in practice. The commercial speeds in lathe operation seem to run from 70 up to 160 or 170 feet per minute. The heavier cuts usually are run at the lower speeds. Fine feeds are commonly used, ranging from .01 to .02 inches per revolution. Although generally these conditions are suc- cessfully met in operation, machining troubles sometimes are encountered. Therefore there is definite reason to believe, either that there exists a fairly wide range of machinability in nor- mal malleable or that in individual cases an abnormal product Cleaning and Finishing 255 is unexpectedly encountered in a small amount mixed in with a large mass of normal material. In the absence of systematic study on the point, no recom- mendations are possible by the producer. It is well, however, to point out some special features influencing machining. If any pearl- ite remains in the finished casting, it is generally very near the sur- Pearl/te Norma/Structure Center of Rotation in lathe -Finish ed Diameter Original Diameter- Fig. 116 When machine center and casting center are not concentric, apparent hard spots may be found face. It is therefore well to design malleable parts with a con- siderable amount of "finish" for it is usually easier to remove 1/16 to 3/32 inches of metal by turning or planing than to take a very light cut which may be almost entirely in this slightly pearlitic area. At the same time this allowance is a 256 American Malleable Cast Iron necessity to take care of the variations of expansion in annealing which are not yet entirely under control of the metallurgist. The film of pearlite just referred to sometimes gives the misleading impression of hard spots in an otherwise sound casting. If the finished surface is not concentric with the sur- face of the rough casting is may be that in only a few places the lathe tool cut traverses the pearlitic areas which then act as hard spots. The fact is that this same area of pearlite exists over the entire surface and had it not been that the eccentricity in machining threw the cut alternately into ferrite and pearlite, no trouble would have been encountered. Fig. 116 illustrates this condition on an exaggerated scale. Such metal as this, of course, is not of the best quality; the manufacturer should and does usually remove this pearlitic lay- er. Howeve'r, attention is called to it here to explain the cause of complaints sometimes made and to suggest means of using such metal which is identical internally with a normally an- nealed product when the pearlite is removed by a cutting tool. Hard spots in malleable, in the sense of microscopic areas containing ungraphitized carbon, and scattered irregularly through the mass of a perfect casting are rare indeed. So rare is the occurrence that complaints of this fault are found to be almost always based on erroneous observation. The symmetri- cal pearlite rim just discussed is the most common cause and represents not a hard spot at one or two points but a tough area of little more than microscopic thickness parallel to the surface throughout. Shrunken Areas Cause Trouble Occasionally, also, a defective casting which for- some rea- son has failed of complete graphitization is soft enough to machine, though with difficulty. If after most of the machining is complete, a tool fails on the casting, the machinist is apt to feel that a hard area has just been encountered. In addition it occasionally happens that in castings made without suitable feeder heads, a machining operation may penetrate a shrink. Such areas always show a bright cut and are mistaken for hard spots. Cementite in fine granules frequently is present in the Cleaning and Finishing 257 shrunken areas and dulls the cutting tool if much of the cut is in the shrink. If the turning operation which penetrates the shrink is thread cutting, the threads will crumble away and the metal may be regarded as defective when the fault is with the feeding of the individual casting. (Both items are to be controlled by the foundry but frequently the character of the complaint is misleading as to the cause of failure. In the case of threading and reaming operations, it is not uncommon to encounter diffi- Fig. 117 (left) Cementite psrsisting near a shrink. The metal in porous areas is somewhat oxidized. Fig. 118 (right) Hard slag inclusions just below the surface which may dull cutting tools rapidly culties with perfectly normal metal. A metal which has been decarbonized considerably may have the entire thread, especially if of fine pitch, cut into the pure ferrite rim. Ferrite cuts freely, but in rather long chips, hence the flutes in dies, taps or reamers may become clogged and prevent a clean cut. In work of this character too deep a decarbonization is objection- able. An interesting operation other than machine tooling occa- sionally may be practiced on malleable. This consists of press fitting and is accomplished by applying sufficient pressure to a casting to bring it to the desired dimensions and perfection of surface. To produce reasonable perfect finishes a pressure of 100,000 pounds per square inch is required. The method is 258 American Malleable Cast Iron Fig. 119 Malleable casting effectively arc welded with Swedish iron. The changes A is soft iron but very slightly recarburized from the malleable; B is an carbon due to Fig. 120 Hard iron casting successfully acetylene welded with hard iron and slag. A is the original casting, B the slag, C the material of weld as noted of a little pearlitc Fig. 121 Ineffective hard weld of malleable casting using ingot iron wire and filler converted into hard iron by migration of carbon from the malleable. bitic due to recombination of carbon at Cleaning and Finishing 259 visible microscopically were insufficient to make notable difference in metal. Area oxide or slag film, and C is the malleable showing but little resolution of close confinement then annealed. Note metallurgical homogeneity of casting except for presence of by larger grain size, and D the material of weld as noted by persistence due to decarburization acetylene method. Neither material has its original structure. A is the soft iron B is the original malleable iron, the background of which has become sor- temperature the metal reached in welding 260 American Malleable Cast Iron particularly applicable where relatively small objects have to be brought to an exact thickness. It is also possible to form small objects, for example, radiator nipples in press dies. The method is sometimes preferred where it is desired to retain a ferrite surface. Welding Is Limited Welding of broken or defective castings by the user is of course subject to the limitations which apply to this operation when carried on by the producer with the additional difficulty that reannealing is impracticable. Had the consumer facilities for the long accurately controlled heat treatments required, he could of course weld in the same manner as does the malleable founder. During annealing finished surfaces would suffer and warping might possibly occur. Under ordinary conditions, therefore, welding with iron is not to be regarded as practicable as a repair operation. Thus no repair can be made, irrespective of the welder's skill, which will restore the original strength of the casting. The only resource is to braze, that is, to use bronze as the welding material. The melting point of bronze is low enough to permit operating below the critical point for iron hence if care is used a weld can be made without heating the metal to a dangerous degree. This, however, involves great skill and care on the part of the welder. Ordinary brass, Tobin bronze and Parsons' manganese bronze has been suggested as suitable for this work. Of course, welds made with nonferrous metals do not permit of the complete merging into one another of the metal used as filler with the material being repaired. They apparently fail invariably by tearing apart between the iron and bronze, thus the entire strength of either material is not developed. The strongest welds of this type ever tested by the writer were made by an expert operator using Parsons' bronze. These welds developed an adhesion between iron and bronze of substantially 45,000 pounds per square inch thus producing a tensile strength of the welded part approximately equal to the American Society for Testing Materials, specifica- tions for malleable iron. The failure occuring entirely along the plane of contact 'Cleaning and Finishing 261 between bronze . and iron produced a failure with only a negligible elongation, as might be expected. If the circumstances are such as to permit making a joint similar in form to the wiped lead joint of the plumber, running the bronze up on the side of the iron part some distance each way, welds occasionally can be made with this metal which develop the full strength of the original metal, elongation excepted. Such welds are seldom made. A manufacturer of alternating current arc welding equipment claims that with his apparatus and a nickel filler small machineable welds can be made in malleable cast iron. The writer has not yet personally investigated this procedure. Work of this character can be intrusted only to very skillful artisans. Unusual care and ability are required to produce me- chanically perfect welds without even momentary overheating of the surrounding metal. Theoretically, there should be no rea- son for preferring electric to acetylene welding or vice versa, T)ut the writer's observation has been that better work is obtained with the gas torch. Possibly this observation may be due to the relative skill of the operators whose work has been observed. Of straightening operations little can be said here, since these operations are in general entirely mechanical. Occasionally there comes to the malleable manufacturer's attention heavy castings which have been bent in service and straightened in a blacksmith's fire. Such castings originate more particularly in the repair shops of railroads. Hot straightening is an extremely dangerous operation and in general should be avoided by the consumer since even severe punishment under a heavy hammer will do the castings less permanent harm than an instantaneous heating above the lower critical point. The best practice is to straighten in a screw or hydraulic press. Must Use Accurate Temperatures Next to this the use of the lightest hammer blows which will accomplish the result is to be recommended. Some castings are of such shape that nothing short of a steam hammer will do any good. In the absence of properly fitting dies such a hammer may so mar the casting as to destroy its utility. Un- der these circumstances hot straightening is an advantage but -can be executed only under conditions permitting of the use of 262 American Malleable Cast Iron accurately known and controlled temperatures. Such straight- ening should be done at temperatures between 1000 and 1100 degrees Fahr. At temperatures below 900 degrees Fahr. the metal is not sufficiently more ductile than when cold to justify the heating operation and at temperatures over 1200 degrees, the danger of accidentally overstepping the critical point is so great as to be unwarranted. In the absence of pyrometer con- trol, hot straightening of castings whose failure would cause loss of life or heavy loss of property is almost criminal. Application of protective coatings to malleable iron to in- crease its rust resistance yet remains for consideration. Pro- jection is obtained by a coating of metallic zinc, applied molten Fig. 122 Photomicrograph showing heavy pearlitic rim which may cause machining difficulties as in hot dip galvanizing; by a peculiar form of penetration at temperatures below the melting point of zinc, as in sherardizing ; and by electroplating as in so-called electrogalvanizing. The relative merits of the three systems is so much in controversy that it is hardly within the province of the article to attempt any decision as between them. It is of course essential to apply such a coating as will furnish the maximum protection under service conditions. The prevalent opinion seems to be that the results of hot dipping are in this respect superior to the two competitive processes. On the other hand, the author is informed that a large consumer of malleable in the form of trolley parts after exhaustive tests determined to his own satisfaction the superi- ority of the sherardized coating. Another extensive user of Cleaning and Finishing 263 malleable, who applied his own coating decided upon the electro- plating method as being equally satisfactory in service and the least liable to injure the product to be coated. Hot galvanizing can and should be done without heating the metal to be coated above 900 degrees Fahr. Under such circumstances there is no reason to fear any recombination of the carbon. Unfortunately, however, there are on record a number of well established instances in which originally perfect malleable castings were seriously impaired by galvanizing. Fig. 123 (Left) An effective acetylene weld, malleable becoming sor- bitic due to resolution of carbon. A is gray iron converted into white cast iron by remelting. B is malleable. .Fig. 124 (Right) Tobin bronze weld in malleable. Note absence of oxides and slag in weld and absence of recom- bination of carbon due to relatively low melting point of bronze. A is bronze, B is malleable Such castings are white in fracture and quite brittle. The fault is believed to arise from careless galvanizing resulting in overheat- ing of the iron to the point of recombination of the carbon. It seems questionable whether a zinc bath could be heated commer- cially to above A x . W. R. Bean, as a result of extensive in- vestigation, believes that such . recombination of carbon never occurs in practice. The writer, and apparently some galvanizers, feel that although rare, it can not be said that such a recom- bination is commercially impossible. However, a very similar variation in quality has been ob- 264 American Malleable Cast Iron served where it was positively determinable that no such over- heating has occurred. Indeed, it is sometimes though rarely observed in tinning where the temperature is never too high. The cause of this well established fact is still obscure. Attempts to correlate it with the absorption of hydrogen during pickling, with heat treatment alone and with the action of the zinc in alloying with iron have all been inconclusive. One malleable metallurgist in a preliminary private communication to the au- thor expressed the belief that similar deterioration was caused in steel and pure iron but escaped notice since the difference in the accompanying fractures is less visibly marked than in malleable castings. Some experiments with various heat treatments at tempera- tures far below the critical point would indicate the possibility that the phenomenon is associated with the grain structure of the material. How these structural changes are produced or overcome is still entirely too little understood to permit useful conclusions as to operating practice. It appears, however, that these faults are rare in sherardized material and have not been observed in the electric galvanized product. On the other hand, hot galvanizing is so generally successful that it may be con- cluded this operation is not necessarily harmful to the physical properties of the iron. In the absence of all definite knowledge, the malleable founder as yet is unable to do anything to assure the success of the operation nor can it be said that any one grade of malleable is better adapted to hot galvanizing than another. The difference in results is more likely to arise from vari- ations in the coating process than from the metallurgical char- acteristics of the castings. Most manufacturers take the ground that they can assume no responsibility for galvanized material beyond the delivery of acceptable castings to the galvanizer. From time to time also tinners and galvanizers think that they observe differences in the way different lots of castings take the coating. Occasionally the claim has been made that entire ship- ments could not be galvanized or tinned, that is, that the coating Cleaning and Finishing 265 could not be made to adhere. No logical reason for such a phe- nomenon seemed evident. All malleable castings consist of the same metallographic ingredients, indeed the surface metal is in all cases practically pure iron which can be tinned or galvan- ized successfully. Careful following up of material complained of for this reason has disclosed that in no case was the fault with the metal itself. Cases occur where the castings have not been cleaned properly and hence do not present suitable surface conditions for coating. This is at times the fault of improper cleaning after annealing and also occasionally due to the formation of a rust or grease coating while the castings are handled in the consumer's plant. In some cases also the fault has been found due to oxidized and dirty zinc or tin baths and to the use of tinning alloys too impure to give good coatings. A manufac- turer for "many years producing malleable castings which he tinned himself in large quantities has assured the writer that no cases have ever been found where castings would not take the coating perfectly if proper tinning practice is maintained. In all that has gone before in this chapter, great stress has been laid on the necessity of avoiding even momentary heating of malleable castings above the critical point. If such heating does occur the carbon instantaneously recombines with iron and can be caused to separate again only by a slow cooling equiva- lent to that at completion of the annealing process. In some few cases advantage is taken of this process to reharden malleable purposely. The combined carbon content after reheating is a function of the temperature attained; the hardness depends on the cooling rate adopted. The result of course is a metal of entirely different character from malleable iron, the malleability and ductility being entirely lost and a new product obtained having some of the general characteris- tics of hardened tool steel. Unless conditions are accurately controlled, the properties of the resulting metal may be quite erratic. To the writer's knowledge, the process has not been applied to any important work. Case hardening is said to have 266 American Malleable Cast Iron been applied to malleable, particularly when used for wood working tools, but the author is unfamiliar with any such practice. However, he has been assured by a consumer that quite recently at least two producers still furnished castings for edged tools. XIV INSPECTING AND TESTING INSPECTION and testing of the finished product falls some- what naturally into two subdivisions, the examination of the material as to its metallurgical properties, and the inspection of the individual castings for perfection of form, etc. The first examination is made generally on the' basis of a system- atic control of the works operations without reference to any particular castings. Insofar as this inspection is conducted by the manufacturer for his own information, but one satisfactory system is used. This system consists as a minimum in the chemical analysis of every heat, either before or at any rate promptly after casting and the breaking in tension of at least one test specimen from each heat. The chemical analyses are of no interest to the consumer. The permanent recording of a test from each heat is required by specification A47-19, section lib, of the American Society for Testing Materials. The maintenance of a systematic record of chemical analyses is an almost unavoidable necessity to insure the found- ryman against making heats which will not pass the specifica- tions. Since test specimens will not come through the an- nealing process for 10 to 14 days after casting, they would not give warning of bad furnace practice in time to prevent the manufacture of a considerable quantity of bad iron. Some dif- ference of opinion may exist as to just what constitutes ade- quate chemical control of the product, but the greatest weight must be laid of course on the control of those elements most likely to be subject to dangerous fluctuations. Color Method Unreliable Carbon and silicon certainly should be determined in every heat. The determination of manganese seems urgent in view of the fact that in air furnace practice this element is oxidized in considerable amounts. The determination of these three ele- ments will furnish a check on the mix, or charge, being fed into 268 American Malleable Cast Iron the furnaces. The fact that in hard iron all the carbon should be in the combined state has lead some chemists to the poor practice of determining total carbon by color. Since the ad- vent of the cheap and rapid direct-combustion methods there re- mains no excuse for such a practice. The color method cannot be relied upon to give correct values on high carbon metal and now survives mainly in consulting laboratories doing cheap Fig. 125 Anatytical laboratory in malleable plant contract work. While occasional expert operators can consist- ently check the correct values to perhaps less than 0.05 per cent the author has seen results emanating from supposedly reputable laboratories as much as 0.50 per cent in error. An expert observer can guess more closely by inspection of a broken sprue. Carbon values to be. useful must be within 0.05 per cent of correct and should be better. This is only possible by combustion methods. Results by color should be disregarded as inaccurate. With good coal and melting stock, sulphur does not vary much from one heat to the next; with poor fuel, however, a close control must be kept. It must be remembered that while Inspecting and Testing 269 considerations of speed usually necessitate sulphur being deter- mined by evolution, the results on white cast iron seldom are exact due to the formation of compounds of carbon, hydrogen and sulphur. Oxidation methods also may fail due to the evolu- tion of gaseous sulphur compounds. Chrome, in the Aug. 10, 1921 issue of Chemical and Metallurgical Engineering, presented data on this point. The writer's experience is that evolution methods seldom give accurate results and may be short 25 per Fig. 126 Apparatus for determining carbon cent of the total sulphur. Oxidation methods executed carefully give the total sulphur but only at the expense of much time. The phosphorus content of the metal, in a commercial sense, can be predicted exactly from the analyses of the stock, there- fore the attention to this element as required by the finished product varies inversely as the supe'rvision given the raw ma- terial. Prudence will dictate the determination of silicon and usually also that of manganese at least in every carload of pig iron. The carbon content of pig iron is fairly constant but must not be neglected entirely. Sulphur and phosphorus being 270 American Malleable Cast Iron subject to specifications should be watched closely. It seems hardly necessary to describe in detail the methods of iron analysis which are applicable to hard and malleable iron. The procedure of iron analysis is becoming so well standardized that mere reference to accepted methods will doubtless give the chemist reader the information he requires without burden- ing the nonchemical reader with uninteresting data. Carbon should always be determined by direct combustion in oxygen, determining the CO 2 formed either by direct weigh- ing in soda lime or preferably by absorption in standard Ba(OH) 2 solution and titration of the excess alkali with stand- ard HC1. Solutions in which 1 cubic centimeter =^0.10 per cent on a 1.0000 gram sample are convenient. It is sometimes an ad- vantage to add to the sample about 1 gram of carbon-free iron before burning to secure better combustion. The use of CuO or of platinum black to complete the oxidation is superfluous. Silicon is invariably determined by a modification of Brown's method substituting a mixture of HNO 3 , HC1 and H 2 SO 4 for Drown's method of solution. The major precaution is to bake well till SO 3 no long comes off to render SiO 2 insoluble. For manganese the persulphate method of Walters is com- mon, finishing the determination either by color or arsenite titra- tion. It is well to destroy "combined carbon," that is, the colored nitro compounds produced in the reaction of cementite, with HNO 3 by oxidation with persulphate before adding any silver solution. Phosphorus may best be determined by solution in HNO 3 ; oxidation, in solution, with KMnO 4 ; precipitation as "phospho- molybdate"; and finishing by alkali titration, all in the usual manner. Where very few determinations are to be made direct weighing of the "yellow precipitate" in Gooch crucibles is con- venient. Evolution sulphurs are made in the usual way. Rapid solution in rather concentrated acid tends toward complete conversion of S into H 2 S. It is also a valuable precaution to heat the weighed sample for one hour under graphite and Inspecting and Testing 271 allow to cool slowly before dissolving. The graphite must be sulphur free. The writer prefers KIo 3 to iodine as a titrating solution. If the oxidation method is used, concentrated acid and slow .solution in a capacious and well covered vessel are desirable. This should be followed by evaporation and subsequent bak- ing for one hour at not over 400 degrees Fahr. Precipitation is made in a cold solution not exceeding 100 cubic centimeters in volume containing besides the 5-gram sample 6 cubic centi- meters of concentrated HC1 using 10 per cent BaQ 2 solution. The solution and filtrate should stand one or two days to allow the latter to crystallize. In view of the length of the process care must be used to avoid contamination by the laboratory atmos- phere. Supervise Sulphur Content Aside from economic considerations sound metallurgical practice would dictate a supervision over the sulphur content of the fuel. Taking into consideration the commercial variations in fuel, stock and furnace operations, a minimum standard for good laboratory control will include the determination of car- bon, silicon and manganese in each heat, silicon and manganese in each car of pig iron, sulphur in all fuel taking an average sample from each group of 5 to 15 cars where coal is delivered in large shipments, and occasional determinations of sulphur and phosphorus in the product. Extending the work to include sulphur in each heat and carload of iron, and phosphorus and carbon in each car of pig iron sometimes may be well repaid. The analysis of scrap material usually is not of value since no means exists for ob- taining a true sample. Analytical investigation of steel scrap suspected of containing unusual elements is sometimes justified when buying scrap direct from the producer. Determination of the tensile properties of one bar from each heat already has been referred to. The best type of works control to insure uniformity of metallurgical quality will in- clude a permanent automatic record of all annealing oven temperatures. The progressive manufacturer will further avail himself of microscopic methods in seeking the cause for defec- 272 American Malleable Cast Iron Fig. 127 Inverted types of metallographic microscope Fig. 128 Detail of inverted type of metallographic microscope (Bausch & Lomb) Inspecting and Testing 27 Z tive material. Methods of metallography yield much valuable information relative to the cause of any failures when these are due to mischances in heat treatment. The metallographic characteristics of hard and malleable iron already have been discussed in connection with the metal- lurgy of the product. Extended discussion here would amount to little more than needless repetition. Messrs. Bean, Highright- er and Davenport presented in a paper before the American Foundrymen's association in 1920 an extended description of "Fractures of Microstructures of American Malleable Cast Iron," showing some 40 illustrations mainly of typical micro- structures. The interested metallographer may well consult the original publication. The technique of the microscopy of these materials is in no respect unusual. Hard iron is rough ground on an emery wheel polished further upon fine emery cloth and finished upon broad- cloth charged with rouge. Some operators conduct the inter- mediate stages of polishing upon broadcloth charged with F. F. F. emery flour and then upon broadcloth and tripoli. The etch- ing medium is almost invariably alcoholic picric acid. Method of Polishing In polishing malleable care is necessary to prevent undue deformation of the soft material and the "smudging" of the temper carbon. Polishing speeds above 600 feet per minute seem undesirable. The specimen is best flattened by milling or planing followed by filing and finished as previously indicated. Suspended alumina has occasionally been used as the polish- ing medium. The etching may be with picric acid if pearlite is to be examined or usually better, especially if grain boundaries are important, with 10 per cent alcoholic nitric acid. A solution of nitric acid in amyl alcohol sometimes overcomes a tendency to stain. Special reagents such as alkaline picrate or Stead's are occasionally required for particular investigations. It is well to begin the examination of malleable at 50 or 100 diameters, to obtain an idea of the form and distribution of temper carbon pearlite, etc. At 200 diameters grain size can 274 American Malleable Cast Iron conveniently be studied. The identification of solid solutions may require 500 to 1000 diameters and the finer details such' as the boundary structures, minute residues of cementite, crystals of titanium cyanonitride or nitride can be seen only at 1000 to 2000 diameters. From the manufacturer's viewpoint, inspection and control of his product in a metallurgical sense involves chemical analyses of raw materials and finished- castings to insure uni- formity of product, autographic pyrometer records to insure uni- formity of heat treatment, systematic testing of tensile specimens to determine the quality attained and metallographic work to seek the cause of otherwise unexplainable faulty material. Inspection for physical properties of the product when conducted by or for the -consumer best can be made in accordance with the Amer- ican Society for Testing Materials, specification A47-19, adopted I -12- Fig. 129 A. S. T. M. Tension test specimen Sept. 1, 1919. For completeness these specifications are quoted in full as follows: 1 These specifications cover malleable castings for railroad, motor vehicle, agricultural implement, and general machinery purposes. I MANUFACTURE 2 The castings shall be produced by either the air-furnace, open- hearth or electric-furnace process. II PHYSICAL PROPERTIES AND TESTS 3 The tension test specimens in Section 5 shall conform to the following minimum requirements as to tensile properties : Tensile strength, pound per square inch 45,000 Elongation in 2-inch, per cent 7.5 A (a) All castings, if of sufficient size, shall have cast thereon test lugs of a size proportional to the thickness of the casting, but not ex- ceeding $/& x %-inch in cross-section. On castings which are 24 inches or over in length, a test lug shall be cast near each end. These test lugs shall be attached to the casting at such a point that they will not interfere with the assembling of the castings, and may be broken off by the in- spector. (b) If the purchaser or his reperesentative so desires, a casting may be tested to destruction. Such a casting shall show good, tough malleable iron. Inspecting and Testing 275 5 (a) Tension test specimens shall be of the form and dimensions shown in Fig. 129. Specimens whose mean diameter at the smallest section is less than 19/32-inch, will not be accepted for test. (b) A set of three tension test specimens shall be cast from each melt, without chills, using heavy risers of sufficient height to secure sound bars. The specimens shall be suitably marked for identification with the melt. Each set of specimens so cast shall be placed in some one oven containing castings to be annealed. 6 (a) t After annealing, three tension test specimens shall be selected by the inspector as representing the castings in the oven from which these specimens are taken. (b) If the first specimen conforms to the. specified requirements, or if, in the event of failure of. the first specimen, the second and third specimens conform to the requirements, the castings in that oven shall be accepted, except that any casting may be rejected if its test lug shows that it has not been properly annealed. If either the second or third specimen fails to conform to the requirements the contents of that oven shall be rejected. 7 Any castings rejected for insufficient annealing may be rean- nealed at once. The reannealed castings shall be inspected and if the remaining test lugs or castings broken as specimens, show the castings to be thoroughly annealed, they shall be accepted; if not, they shall be finally rejected. Ill WORKMANSHIP AND FINISH 8 The castings shall conform substantially to the patterns or draw- ings furnished by the purchaser, and also to gages which may be specified in individual cases. The castings shall be made in a workmanlike man- ner. A variation of ^-inch per foot will be permitted. 9 The castings shall be free from injurious defects. IV MARKING 10 The manufacturer's identification mark and the pattern numbers assigned by the purchaser shall be cast on all of sufficient size, in such positions that they will not interfere with the service of the castings. V INSPECTION AND REJECTION 11 (a) The inspector representing the purchaser shall have free entry, at all times while work on the contract of the purchaser is being performed, to all parts of the manufacture's works which concern the manufacture of the castings ordered. The manufacturer shall afford the inspector, free of cost, all reasonable facilities to satisfy him that the castings are being furnished in accordance with these specifications. All tests and inspection shall be made at the place of manufacture prior to shipment, unless otherwise specified, and shall be so conducted as not to interfere unnecessarily with the operation of the works. (b) The manufacturer shall be required to keep a record of each melt from which castings are produced, showing tensile strength and elongation of test specimens cast from such melts. These records shall be available and shown to the inspector whenever required. 12 Castings which show injurious defects subsequent to their accept- ance at the manufacturer's works may be rejected, and, if rejected, shall be replaced by the manufacturer free of cost to the purchaser. These specifications contain a number of points which perhaps may be subject to criticism, nevertheless representing 276 American Malleable Cast Iron as they do the consensus of opinion of a committee acting for all interested parties and having the approval of a large body of able engineering specialists, the specifications* may be con- sidered the best practicable solution of the problem of inspec- tion of malleable. The specifications further have the approval of the Ameri- can Foundrymen's association and of the American Malleable Castings association. Therefore, it would seem to the best in- terests of all that this specification, together with its further authorized versions, should be adopted by all producers and consumers as a universal guide to quality. Any attempt to modi- fy or adapt it to supposed special conditions as a rule will be ? ' " o i " * & ? ' " 3 *-t 1 3 J 1 1 |D.a. ft 2 i" Fig. 130 Dimensions of proposed tension test bar productive of intolerable confusion and secure no compensating advantage. The benefits of standardization will be lost and the resulting specification, not having the foundation of mature consideration by many minds is likely to be less satisfactory than the standard. If in any special case it is agreed by buyer and seller that it is to their mutual interest to waive the specifica- tions, of course no objections can be made to that course provided the understanding is clear to both parties. It will be seen that inspection by means of test lugs is prov- ided for in the specification. This is a valuable check on the ^Revisions in the specifications quoted on pages 274 and 275 were adopted as tentative at the 1922 meeting of the A. S. T. M. Section 3, is tentatively changed to read : "The tension test specimens specified in section 5 shall conform to the following minimum requirements as to tensile properties : Tensile strength, pounds per square inch 50,000 Elongation in 2-inch, per cent 10.0 In Section 6 (b}. the following sentence is added: "In case one of the retest specimens contains a flaw which results in the failure of the bar to meet the specifications, at the discretion of the inspector additional test specimens from the same oven may be tested, or test specimens may be cut from castings." It is further recommended that the standard test specimen be modified to conform to the dimensions shown in Fig. 130. Inspecting and Testing 277 quality of individual castings. Test lugs are projections in the form of a frustum, of a rectangular pyramid, or of a cone which are broken off by the works inspector or by the consumer to determine the quality of the metal in the casting. The size of these test lugs depends upon the size and thickness of the cast- ings to which they are attached. Thus it is impracticable to lay down definite rules for their size, form and location. In general, lugs should be applied to all castings where quality is important. Pieces weighing less than 3 pounds or heavier of thin cross section are usually too small to permit of putting on a lug and breaking it off without damage to the casting. The round test lug is much affected in appearance by shrinks and is quite deceptive at times. The author's preference is for rectangular test lugs in which the smaller dimension at the point of fracture is ^4-inch less than the layer. Generally the height of a test lug should be about equal to the larger dimen- sion at the point of fracture, and the taper about 1/32 to 1/16-inch per 1 inch on each side. Useful sizes of lugs are specified as follows : Dimensions at Dimensions breaking point at top Height in inches in inches in Class of work Length Width Length Width inches Very heavy sections 1^4 -inch thick and over -K ^ H Jz -K Intermediate 5/g ^ -ft & y & Light castings up to ^-inch thick.. TS ~fs Yz Y$> . A Test lugs, to represent the metal properly, must be free from shrinks; hence in general should be located in the drag of the mold. In inspecting castings by test lugs, care should be used that the lug is not bent in opposite directions to break it off. The practice of nicking lugs with a chisel before break- ing also interferes with a correct interpretation of the result. Under such circumstances the lug breaks off "shorter," that is, shows less toughness than it should. Three factors must be given consideration in determining the quality of a casting from test lug inspection. These items are the effort required to break off the lug, the distortion it sustains before breaking, and the appearance of the resulting 278 American Malleable Cast Iron fracture. While the effort cannot be measured and recorded in figures, after a time it becomes simple to compare different results fairly accurately. In general the hammer should not be so heavy as to break off a good lug with one or two blows. A fair idea of the energy consumed can be formed from the number of blows required to produce fracture. The amount of distortion in breaking usually increases with the blows required to do the breaking. Test lugs should Fig. 131 A 200,000-pound Olsen universal testing machine bend out of line materially before fracture. All conditions be- ing equal, small test lugs will bend further than large ones. On small work where small lugs may be unavoidable, they will often hammer over flat before breaking. On heavy lugs a displace- ment of 30 degrees will indicate very good material. The inter- mediate and smaller sizes listed in the table may bend some- what more, even up to 60 degrees. Distortion is greater when the break is made by frequent light blows than by a few heavy Inspecting and Testing 279 blows. Striking the lug alternately on opposite sides of course will produce no distortion and hence is valueless. The fracture of normal malleable iron, in the absence of much compression, is of a velvety black appearance, having a mouse gray rim of fair depth. Occasionally two bands are ob- served, the outer one being somewhat lighter than the inner. The outer rim in such cases, however, is never steely in ap- pearance. In bending the lug over, the concave side is of course considerably compressed and this compression so distorts the crystal structure of the ferrite as to materially alter its appear- Pig. 132 Ewing-type extensometer for determining elongation under load ance. Toward the concave side of such a lug the fracture will be silver white in color and rather fine in grain, that is, not coarsely crystalline. This structure may occupy half or even more of the entire fracture. However, a band free from any steely rim and of normal appearance will always be found toward the convex side. .When the so-called "compression fracture" is but slightly developed, danger exists for mistaking it for a rim unless it is 280 American Malleable Cast Iron observed that the white edge is along one boundary of the frac- ture only instead. of uniformly around it. Lugs broken by being struck on opposite sides may show this compression edge on the two opposite boundaries and may be difficult of interpretation. .They may even be clear white. '- Occasionally fractures are encountered which have a so- called "picture frame" rim or ''shuck." This is a rim, usually of crystalline appearance, completely surrounding the fracture as Fig. 133 Olsen-type. torsion testing machine a band of uniform width. If the rim is narrow, the material may be strong and will bend fairly well. Such rims usually con- tain pearlite and the resulting metal is not readily machinable. Where machining is no object, a reasonably narrow edge of this character need not condemn the product if the lug withstood punishment well. Where machining is involved, the inspector should use discretion in taking any material with edges in order to exclude this condition. Entirely white fractures somewhat rarely occur. These may be due to an anneal so incomplete that the original hard iron Inspecting and Testing 281 Fie;. 134 Leeds & Northrup Co. apparatus for determining critical points by Roberts Austens method structure is but slightly altered, in which case the castings should be returned for reannealing. Occasionally the fracture is com- posed entirely of steely brilliant facets surrounded by a narrow rim of a more gray color. Such iron is useless from the Ameri- can viewpoint, being that normal to white heart malleable. It is due to radical faults of chemical composition and cannot be saved by any ordinary reannealing. A further type of white fracture sometimes met with Fig. 135 Apparatus for measuring magnetic properties of metal 282 American Malleable Cast Iron resembles in color and texture the compression fracture men- tioned before but extends over the entire fracture. Such lugs usually bend but litlte though they are decidedly tougher than those defective on account of an incomplete anneal. This ma- terial is normal under the microscope and contains no combined carbon. The fault lies with the crystalline structure of the fer- rite and can be remedied by suitable further heat treatment. A coarse black center surrounded by a slate-colored rim accompanies weak lugs and is characteristic of poor, high carbon material. Considerable experience is necessary to interpret ab- normal fractures properly. Indeed, those who pretend off hand and from inspection alone to solve all problems as to quality of material and causes of failure, usually overestimate their own abilities. In many cases all the resources of a chemical and metal - lographic laboratory are required to diagnose troubles. Since the consumer's inspector is not interested in the cause of trou- Fig. 136 Farmer fatigue testing machine Inspecting and Testing 283 Fig. 1.37 Charpy hammer for impact tests bles he may be guided in the acceptance of material by the following considerations : 1. Deformation of the lugs must be up to standard. 2. Bending should require a fair degree of effort. 3. Irrespective of the fracture, reject all material in which the lugs snap off sharply. 4. Irrespective of a fracture, accept any material in which the lug has sustained much more than average punish- ment as a result of which indications are rendered worthless due to the heavy distortion. 5. Where machining is to be done, reject any castings which have more than a paper thin rim. 6. Where machining is no object, accept castings with a wide steely edge only if the performance of 'the lug under pun- ishment is unquestionable. 7. Reject all castings having a coarse structure and the slate colored rim. Such lugs generally are defective with respect to the first three tests also. 8. Return for annealing all condemned castings in which the fracture is partially or entirely silver or steely in color. 284 American Malleable Cast Iron It may be well also for the inspector to assure himself of the absence of injurious shrinkage by breaking hard or annealed castings from time to time and by watching the fracture of heavy unannealed castings for the presence of primary graphite f 1 Fig. 138 Brinell hardness tester "mottles." Both shrinks and mottles are found preferentially in the last cooling sections. Occasionally questions are raised as to inspection for vari- ous purposes after arrival of the product at the consumer's plant. Except in very exceptional cases, inspection and condemnation of entire lots on the basis of faults observed on individual pieces cannot be resorted to fairly. As the average malleable found- ry is operated, it is quite possible that no two castings in a Inspecting and Testing 285 given sack, or barrel, are representative of the same heat in the melting department and oven in annealing. The fact that in an impartially drawn sample a certain small number of defective pieces are, or are not found proves nothing as to the remaining pieces. Therefore, only an inspection piece by piece is equitable after the castings can no longer be identified with certain specific lots made in the foundry. The test lug inspection was devised for this very purpose. Upon occasion the problem has arisen of selecting from a large and indiscrim- inate mass of castings those too hard to machine. Brinell and Shore tests are useless for the purpose unless the material is practically unannealed. Some inspectors feel that the behavior under a preliminary drilling operation is suit- able as a means of weeding out hard castings. Others have used the ring of the casting, that is, the pitch of its musical note when struck. However, none of these methods are as cheap, as simple, or as conclusive as the breaking of a test lug. Occasionally it is desired to inspect the finished or semifin- ished article to make sure of its fitness for the intended loads. Where the maximum loads do not require a proof load beyond the yield point of the article, the application of such a load is an ideal test. Thus a link belt can be loaded in tension to about the yield point and defects which would result in failure under- service conditions can be discovered. Castings which are straightened after a material deforma- tion receive of course a test similar in principle to such a proof test. It is therefore hardly necessary to deal with them here in any detail. Inspection of castings as to their being true to size and form, etc., has not been discussed but this is done by the usual methods of gaging and is not different from similar inspec- tion on any other product. XV TENSILE PROPERTIES TENSION is the simplest stress which can be applied to a material. The ease of execution ,of this form of loading has made tensile tests a favorite means of judging the quality of a metal even though relatively few structural details are sub- jected to pure tension in service. When an elastic material is stretched it first lengthens in exact proportion to the applied load, in other words, it follows Hook's law of the proportionalit/ of stress to strain. Beyond a certain definite loading the stretch increases more rapidly than the applied load. The point where this occurs, beyond which the material no longer obeys Hook's law, is always referred to as the proportional limit. In many materials the increasing rate of 'Stretch is at first so slight as to escape detection by any but the most sensitive of measuring instruments. As more and more load is applied a point is usually readied, however, where the material begins to elongate very rapidly with practically no increase in the load applied. This load is called the yield point and is more easily recognized than defined. The term elastic limit, frequently used and also frequently misused, signifies that stress up to which the material is not permanently deformed. In other words, a material may be loaded to any amount up to, its elastic limit and when the load is removed will return to exactly its original length. This test is seldom employed. Like the proportional limit, the elastic limit depends largely on the sensitiveness of the available means of measurement. Explanation of Terms Frequently the three points are confused and used as if they were identical- The proportional limit is necessarily below the yield point ; how much below depends on the material 288 American Malleable Cast Iron being tested and the accuracy of the measurements. The elastic and proportional limits may be considered identical in principle, but up to the yield point the permanent set, or elongation, might be so small as to escape recognition. The reader should remember that by the very definition of the proportional and elastic limits the apparent location of these points will vary with the available methods of measurement, the proportional limit being the largest load the material will sustain without visible departure from Hook's Law, and the elastic limit the largest load it will sustain without taking a permanent set. The greater the precision of the measurement, the lower will be the stress corresponding to these definitions. Doubt is fre- quently expressed whether cast metals actually have any propor- tional limit larger than zero, the thought being that with suf- ficiently delicate extensometer measurements, the graph would be a curve from the origin. In view of these facts an attempt to find the elastic limit by watching the drop of the testing machine beam will give apparently higher values than determining this point by the divider method and the divider method will give materially higher results than the extensometer. The engineer will therefore require to know how these points have been determined in making intelligent use of the information. Action of Metals in Tension Most ductile materials when loaded in tension beyond the yield point do not stretch uniformly at all points of their length. The larger part of the deformation usually occurs quite close to the point of failure. The specimens accordingly neck in and finally break at the smallest portion of the necks. The per cent of elonga- tion is therefore less the longer the gage length in which it is meas- ured. The difference between the area at the point of fracture and the original cross-sectional area, expressed in per cent, is called the reduction in area. A high reduction in area is even more indi- cative of a very ductile material than a high elongation. The ratio of stress to strain, below the proportional limit, is known as the modulus of elasticity- These various constants, proportional limit, yield point, elongation, reduction in area, modulus of elasticity, and ultimate strength and the relationships between them give a very good picture of the behavior of any material Tensile Properties 289 under static loads. These constants also yield some information regarding its behavior under dynamic stresses. The application of each constant is fairly evident. The proportional limit is useful when the deflection must be temporary and predictable. The yield point limits the stresses which may be applied with- 50000 40000 b 30000 c. 20000 10000 Curve A' Curve B- Specimen; Diameter = 0. Gage length* 5 004 0.0004 0.06 0.0006 0.08 0.0006 0.10 00010 O.IZ o.oo ie Unit" Elongation Fig. 139 Stress-strain diagram of malleable cast iron in tension out producing visible 'permanent changes of shape in the mater- ial. Elongation and reduction of area are indicative of the amount of distortion a material can stand without fracture. The ultimate strength measures the load that can be sustained without failure, although with permanent deformation. The modulus of elasticity serves to determine the elastic deflection under relatively small loads. The behavior of a material under tension is most con- veniently expressed by means of a stress-strain diagram, in which the elongation in per cent in some definite gage length is plotted against the increasing load in pounds per square inch. Fig. 139 shows a graph of this kind somewhat typical of malle- 290 American Malleable Cast Iron able cast iron. The various constants are marked in the graph in the appropriate places. The curve is made from a malleable casting about the tensile strength .prescribed by the A. S. T. M. specifications. Malleable iron of higher tensile strength would have the proportional limit, elastic limit, and yield point raised very closely in the same proportion as the tensile strength increased. In other words, fhe proportional limit would always be about one- third of the ultimate strength and the yield point as measured by extensometer about six-tenths the ultimate strength. The yield point determined by the divider method will be about two- thirds of the ultimate strength. The tensile strength of malleable cast iron, as measured in a test specimen of specified form and dimensions s'hould be 45,000 pounds per square inch and its elongation in 2 inches 7 l /2 per cent according to the 1919 specincatioins of the American Society for Testing Materials. The specimen is to be of the form and dimensions shown in Fig. 129. The apparent tensile strength of this, as of any other cast product, is affected by the gating of the castings forming the test specimen. This is not due to any effect on the properties of the metal as such, but on the degree of soundness wlhich is secured in the casting. Obviously, to give representative results it is necessary to take such precautions as may insure the freedom of the specimen from shrinkage. Specimens Must Be Representative The point seems worthy of discussion in this chapter be- cause criticism and confusion often arise when specimens cut from castings or parts of castings do not conform in properties to the American Society for Testing Materials specimens from the same heat. The discrepancy frequently is due mainly to inter- nal defects of the castings from which specimens are taken. Failure of 'such specimens to pass the test indicates im- properly fed castings ratiher than weak metal. The tensile strength and elongation of malleable as made today by the lead- ing manufacturers exceed the American Society for Test- ing Materials specification by a safe margin, the metal now Tensile Properties 291 sold by reputable makers rarely being under 48,000 pounds per square inch' in tensile strength and 10 per cent in elongation. The product probably averages about 51,000 pounds ultimate strength and 12 per cent elongation. The tensile strength and elongation of daily specimens submitted by all of the more than 60 members of the American Malleable Castings association have been averaged by months and the results plotted as shown in Fig. 140. The recent data average better than the author's personal estimate. Occasional record performances have been noted. The highest grade malleable known to the writer was a single piece having a strength of 58,000 pounds per square inch, and an elongation of 34 per cent. A strength of 64,000 pounds coupled with an elongation of 18 per cent was once noted. These were single isolated cases and in no sense typical of a routine product. One plant produced castings over 57,000 pounds ultimate strength and 20 per cent elongation continuously for about a month. As might be expected from its microstructure, the tensile strength of malleable cast iron is largely dependent upon its carbon content, since the more carbon the greater the interrup- tion to the mechanical continuity of the casting. This applies rather to the original carbon content than to that after anneal. Carbon once liberated has accomplished its destruction of con- tinuity and even if it can be removed after formation, it leaves behind the hole it occupied. Furthermore, the other elements present besides carbon may affect the physical properties of the ferrite just as they affect Fig. 140 Tensile strength and elongation plotted from specimens submitted by members of American Malleable Castings Association 292 American Malleable Cast Iron the properties of a dead soft steel. This, however, is of less practical importance than the variations due to carbon, since within 'the limits capable of commercial annealing none of the other dements are likely to have an effect of the order of magnitude of those due to the latter element. The writer in the past has had occasion to make comparisons of the tensile proper- ties of many thousands of 'heats with their chemical composi- tions. As a rule investigations of this character are influenced by so many variables that a summary which is strictly accurate as well as fairly simple is hardly possible, save at the expense of space for detailed technical explanation which could be spared only in a monograph upon that one subject. Increased Carbon Lowers Strength In general it may be said that an increase in carbon always carries with it a decrease in strength and elongation. The de- crease in strength per unit increase in carbon is greater the greater the total amount of carbon and the higher the silicon- Manganese and sulphur when present in correct relative proportion and within anything resembling commercial limits have relatively little effect. Phosphorus up to about 0.20 or 0.25 per cent strengthens the metal without decreasing its ductility. The considerations just /outlined would seem to furnish a basis for a graphical or tabular summary of the relation be- tween tensile strength and chemical composition. The great dif- ficulty is that even though the effect of each element may be well established, there remain variables due to the form of test specimen, the soundness of the specimen and the effect of the previous thermal history on the physical and grain structure of the ferrite. Accordingly the presentation of such a summary might be misleading to the interested user of malleable and would serve :no useful purpose as a guide to specifications or to successful practice, unless the other variables could also be successfully defined and prescribed. As a guide to the general order of magnitude of the effect of carbon and silicon on normal malleable iron, Fig. 141 shows the average tensile strength of malleable of varying carbon content but of constant silicon as Tensile Properties 293 averaged from a large number of heats. An increase of 0.01 per cent silicon decreases the tensile strength about 20 pounds per square inch for low-carbon iron (about 2.25 per cent), and about 75 pounds per square inch high-carbon iron (about 3.25 per cent). From these data it would seem that a simple arith- metical calculation should show what the strength of malleable 54000 c ^nnn Per Cent C 235 .40 45 .50 .55 I. arbc DO U >n Before Annea >5 70 .75 j50 21 I J5 .90 .95 3. 30 * B= 5=^ ^ ^ ^stooo o 51000 50000 0-49000 tf> i_ A o nr\r\ X x ^s ^ \ v \ ^, \ V Q)40WV CL \ V TJ 46000 ^45000 AAC\f]C\ 5il icoi i = ,75 PerCer t \ \ \ 43000 Fig. 141 Effect of carbon on tensile properties of malleable iron cast iron in pounds per square inch measured in the American Society for Testing Materials test specimen should be for any given composition. Any attempt, however, to apply these figures literally is not likely to be productive of results, since the formula is purely an empirical one and since no account is taken of some of the other variables, notably of the effect of heat treatment in the properties of ferrite. Malleable iron, when completely annealed, stands alone among the ferrous materials in that variations of composition 294 American Malleable Cast Iron affect the elongation in the same direction as the strength. That is, malleable cast iron has a higher elongation the greater its strength. The reader should not lose sight of the fact that what has just been said concerning the proportionality of tensile strength and elongation is only true of completely graphitized 'products. For many years and up to relatively recently misguided efforts were made by ill-informed or careless manufacturers to produce a metal of great strength by using a chemical com- position or heat treatment calculated to produce incomplete de- composition of the combined carbon. The resulting metal is, of course, stronger than good malleable cast iron, since the matrix is more or less pearlitic instead of pure ferrite; and also since less temper carbon is formed by the amount remaining combined in the matrix. However, the relative lack of ductility of the pearlite, interrupted as it is in addition by temper carbon, ac- counts for the lack of elongation shown by material of this character. The elongation may fall as low as 2 per cent in such cases- High Strength May Be Deceptive Material in which a strength approaching or exceeding 60,- 000 pounds per square inch is observed, without a correspond- ingly good elongation (at least up to the average or preferably as high as 12 per cent or 15 per cent) should be looked on with grave suspicion as not being the product of well controlled malleable practice. Each circle in Fig. 142 shows a group of heats of a given analysis, the different circles representing different analysis. They are located according to the strength and elongation of the resulting product. It is plainly evident that increasing strength is accompanied by higher elongation. This graph fur- nishes some basis for conclusions as to the effect of chemical compositions on elongations by demonstrating the approximate proportionality of f the two properties. An exception has been noted in that while silicon slightly decreases tensile strength and hence should decrease elongation, the reverse is true for very low silicons, especially in the presence of low carbons. The Tensile Properties 295. departure may perhaps be explained in the light of minor inter- ferences with complete graphitization. The tensile strength of malleable iron further varies with the cross-^sectional area of the piece under ; consideration. This phenomenon is not due to the long-exploded thought that the; strength of malleable iron its only in the skin. This thought persisted from the days when malleable iron was made; JWUU . E 50000 ST & _> lUsCOO I t/) j> 40000 35000 / s 8 ,J f*. < y A o ^ / _>^ y ,/ / \/\ ^ / / > 7 6 9 10 II \Z Elprrcjation in 2 in T percent Fig. 142 Relation between tensile strength and elongation of mal- leable cast iron "malleable" by decarburization only, as is the case with the so- called "white heart" product of Europe. The skin of normal American or black heart nfalleable dif- fers only in degree from the center. W. R. Bean* gave figures indicating that specimens from the same heats tested in their condition as cast and after machining off at least 1-16 inch, and sometimes % inch of the surface, had practically the same strength. Tests made by the writer indicate that on 'sections 'up to one inch in diameter, after machining, the ultimate strength *Piaper -presented at the annuad meeting of American Society for Testing Materials, 1919. 296 American Malleable Cast Iron 8000 7000 "I 6000 ISOOG f- 4000 ^ 3000 c Jzooo 1000 ^ o - ro ^ ^ Percent j Not ?. Data Plotted are the Difference - v - 2; ^ -- --/> ^ \l ^ Corrects. Apparent '0^ /^OA- rlisprinl. Relation Between L oad on One Inch D/e and rensi/e Properties. / \ - --T- ^ -I ^/ ^ N VM >J f > \ y " / fte/atf on Between Dri/ /ing Quo/ityond \ Tensr/e \ m r!5A ^v. s 1 \ 1 ^ s^ G \ ^ <., .00& 'er/?e>, \N p j \\ \ \ \\\ \ \ ^ \ Dl"P >r A 7 ^v \\! \ \ x - ^ x ^^~^ ^s te \ X \ \ ^^^ > K^^ / o/- ^ 1 1 ? i * *+. *< \ - . - \ \ "^^ ~^ ~^^ I & \ s '^~ 2ss .O2"Pe/'/3ev. \ w K -. "^~- ^ ^_ -~-4 ^ ^ ~fi~~i-J ^ ^ ^ "~ <^ > ^ __ h^'Fer /?&v. ~^_ r~~ itH 1 ? 200 300 400 $0 600 $25.005 Dl .02 .04 Sfiee&/s? /&?M. Feed Per tfe\/o/vf/ on J/?d- feeds Fig. 160. Graph showing values of a in drilling formula per minute. Both loads decrease slightly with increasing speed and are more nearly constant with variations in speed at high speeds than at low. Representing by T v the torque and by T t the thrust of a drill of diameter d f running at s revolutions per minute with a feed, /, in a certain uniform iron and T t = bfd Fatigue, Impact Hardness and Wear 333 in which a and b are constants depending on s f t, and d. The values of a and b can 'be interpolated from the graphs in Fig. 160 and 161 respectively. The drilling properties are further affected by variations in the character of the metal- The investigation disclosed /00,000 92/XX>\ dspoo 86OO& 84,00, 82000 80&00 780OO 76,000 100 X \\ \\ \ I I I 13m yoo&W W&ffi. ?00 400 3W 600 |' a J" M Fig. 161. Graph showing values of b in drilling formula that machining stresses were not related to either Brinell num- ber or strength as effect to cause. However, there is a gen- eral coincidence between the three properties in completely annealed malleable. The data of Figs. 160 and 161 were obtained from malle- able equivalent to a tensile strength of 52,000 pounds per square inch and a Brinell number of 120. In Figs. 162 and 163 the effect of Brinell number and strength on T v and T t are plotted in the form of coefficients for reducing, the previously calcu- lated values to suit other tensile or hardness properties. 334 American Malleable Cast Iron The observations in a measure substantiate Smith and Barr's ideas as to the increased machining difficulty of stronger metal. Apparently there is a variation of from 25 to 30 pet- cent in the stresses developed as between the weakest and strongest malleable. These variations are not nearly sufficient to bridge the gap between malleable and even the softest steel. 90 to LX Fig. 162. Correction factor for drill torque and thrust in terms of ultimate strength It is again to be emphasized that neither the author's data nor that of Smith and Barr can be interpreted in terms of tool life. Furthermore, it must be clearly remembered that the data were all obtained on completely graphitized material and that nothing heretofore sajid has any relation whatever to white edged or white material resulting from mischances in annealing. The subject has already been referred to in connection with Fatigue, Impact Hardness and Wear 335 the discussion of Brinell numbers where it was shown that such mischances do not necessarily influence the hardness test. They do, however, greatly affect machineability both with re- spect to tool life and stresses. The machining difficulties occasionally encountered might be explained on either of three grounds. First, the material /.SO {.25 93 .90 .83 .80 Fig. 163. Correction factor for drill torque and thrust in terms of Brinell number may be so tough 'that the heat developed per unit of time causes the tool temperature to increase rapidly. The tool fails for per- fectly normal causes but under much accelerated conditions. Second, the material may contain particles sufficiently hard to work as an abrasive and so destroy the cutting edge. Third, 336 American Malleable Cast Iron the material may set up so heavy a tangential load in the tool point as to cause it to break off irrespective of the failure of the cutting edge. A study of abnormally early tool failures seems to indi- cate the occurrence of failures of all three types- Since all malleable cast iron consists only of ferrite and temper carbon it is difficult to see how any great difference could exist between different products varying only in the percentage of temper carbon present. This is all the more true since in general the cutting is in a region where relatively little carbon remains due to decarburization in anneal- In the case of imperfectly 'annealed iron a condition ac- counting fof any or all these causes of failure may exist. White cast iron is known to exert very heavy unit stress on the tool point, hence a metal so imperfectly annealed as to retain much of its original pearlite-cementite dendritic structure would set up abnormal tool loads and cause a failure of the third class. Material in which cooling 'has been so slow that all pearlite is graphitized but in which some cementite persists would pro- duce failures of the second class. Cementite is an exceedingly hard ingredient, the hardest of any carbon-iron alloy. Its hard- ness on the mineralogical scale is between 8 and 9, since it is harder than the hardest steels. In imperfect malleable of this kind it would be found scattered as granules through the ferrite. Being present In very small amount only, it could hardly exert any very great effect on the ferrite mass in which it is imbedded and therefore is not likely to either in- crease the tool temperature or the load thereon. The tool edge however, will encounter those granules lying in the finished surface and these grains will rapidly wear away the cutting edge which rubs against them. Failures of the first class are very largely due to so called "picture frame" iron in which there remains a consider- able pearlite layer just under the surface. This layer is identical in composition and properties with annealed tool steel. As such the cutting speed will not be great before sufficient heat is generated to rapidly destroy the tool. Unusually bad cases Fatigue, Impact Hardness and Wear 337 of this character may also produce failures of the third class. It should 'be remembered that all normally made malleable is easily machined, there being minor differences only between the machineability of malleable of varying composition. When machining difficulties are encountered the explanation general- ly is due to failures of execution in individual cases rather than to the character of the product as a whole. Resistance to Friction To all intents and purposes, resistance to frictional wear obviously is the converse of machineability. Experience seems to indicate that the most successful bearing metals are those consisting of fairly soft matrix in which a relative hard con- stituent is imbedded. The hard constituent takes the wear and is supported by the soft. Further, the soft constituent wearing down a little, furnishes the certainty of a supply of lubricant to the bearing. The soft ingredient is further desirable since if a grain of abrasive enters the bearing and lodges tightly in the bearing metal it will >soon cut away the rotating mem- ber where the latter rubs against it. With a soft bearing metal the grit will at most cut a groove in the easily replaced bearing without damage to the shaft. Since malleable does not contain the hard skeleton or grain required to promote long life it cannot be regarded as suitable metal to resist wear. By 'analogy also with gray iron this conclusion seems warranted. Extremely soft gray irons, which resemble malleable more closely than those containing more combined carbon, are inferior to the harder irons as bearing materials. Malleable is not a suitable material of construction, where the major requirement is resistance to wear, as for instance in journal bearing's. It will of course resist minor friction incident to other service. . Under such circumstances, the conditions as to hard- ness and smoothness of the material rubbing against it is of prime importance in determining the service to be expected. Determinations of the coefficient of friction for the metal are not available. While they would be highly interesting, they 338 American Malleable Cast Iron are not of great practical application because of the general unsuitability of malleable for friction service and because under normal lubricating conditions in machine parts there is not metallic contact between shaft and bearing. Therefore the friction losses depend mainly upon the lubricant and not upon the material of the shaft and its support. XVIII PLASTIC DEFORMATION IN CONSIDERING the behavior of malleable cast iron under mechanical stress we have noted that like most other materials its deformation, or strain, under load is of two entirely distinct characters, depending upon the intensity of the stress. Under light loads the deformation is elastic; that is, it is pro- portional to the applied stress and is not permanent, the metal returning to its original dimensions upon the removal of the load. At higher stresses the strain increases very rapidly and the spe- cimen becomes permanently deformed. This change of form is termed "plastic" deformation as dis- tinguished from "elastic," and especially characterizes ductile metals. Much interesting work has been done in the investiga- tion of the phenomenon of plastic flow, the property by virtue of which a material is malleable and ductile. The property is usually measured in terms of yield point, reduction in area and elongation. Nutting has developed the thesis that plastic strain may be expressed as the product of constant and exponential functions of the stress and time. In other words, the strain is measured by the expression AS x t y where S is stress, t time and A, x and -v characteristics of the material. Hook's law is a special case of this formula when x=\ and y==0. The author has no desire, in the present connection, to at- tempt any exposition of the theoretical aspects of plastic flow. However, since malleable cast iron is in quite a marked degree capable of plastic deformation, and in fact, owes many of its most valuable properties to this property, it seems well to sum- marize the effect of plastic deformation on the metal. Summary of Theory As has been shown, malleable is in effect a mass of ferrite made up of individual grains. Each grain is made up of many crystals all oriented in space in the same direction. The several 340 American Malleable Cast Iron grains are held together, according to the now generally accept- ed view, by a thin layer of amorphous (non-crystalline) iron acting as a cement. This amorphous iron is supposed to be stronger than the crystalline variety and is supposed to behave like a very viscous liquid. It is also supposed that crystalline iron will go over into amorphous iron under heavy stresses. The behavior of a metal under even the simplest stresses is as a rule complex. Even when a stress is applied in only one direction the behavior of the material indicates that com- plex systems of forces result. While we speak of the elonga- tion or compression of a metal these terms are in a sense mis- nomers, since solids are but slightly compressible in the sense of a decrease in volume or mutatis mutandis capable of elonga- tion. Metals compressed or lengthened by plastic deforma- tion do not materially gain or lose bulk. For example a speci- men compressed until it was only one-fourth its original height had its density reduced from 7.206 to 7.196 in the process, a change in the opposite direction to what might be expected. Behavior of Specimens The increase or decrease of dimensions parallel to the di- rection of applied stress is made up by decreases or increases of cross section in a plane normal to the axis of stress; the tensile specimen necks in, the compression specimen becomes barrel shaped. We note also that plastic materials do not fail in tension or compression in a plane normal to the stress. The tension speci- men shows a cup shaped fracture, at least on one side of the break. The compression specimen tears apart either in a plane approximately at 45 degrees to the direction of stress, or more rarely on a conical surface whose axis of symetry coincides with the direction of load. From these observations it is evident that there is a consid- erable motion of translation within the stressed material in di- rections at right angles to the direction of the applied stress. This rearrangement may conceivably be of two kinds in a ma- terial composed of crystalline grains either a deformation of the individual grain (intragranular) or a separation and Plastic Deformation 341 rearrangement of the grains at their boundaries (intergranular). Both phenomena are easily recognized. A deformation of the grain itself is accomplished by a shearing of the grain along Fig. 164. Slip bands in ferrite of malleable iron Nitric acid etch 1000 diameters Note that there is but little evidence of any separation at grain boundaries mtragranular crystal boundaries. Such a slip, if occurring in a grain in a polished surface, shows a series of parallel lines on the polished surfate which are fine grooves and ridges in the originally plane surface. Fig. 164 shows a micrograph at 1000 diameters of such slip bands in a ferrite grain in malleable cast iron. Such a deforma- 342 American Malleable Cast Iron tion, increased in magnitude, may result in the rupture of the grain itself at right angles to the slip bands, as shown in Fig. 165, or by producing such a distortion of the grains that it can no longer articulate with the surrounding grains closely enough to be held to them by the cement of amorphous iron at the bound- aries. Change of Structure When Deformed On the other hand examination of the originally polished plane surface of a specimen parallel to the direction of stress which had failed by primary intergranular fracture would show Fig. 165. Intragranular fracture of a ferrite grain in malleable Nitric acid etch 1000 diameters Note that the path of rupture has advanced about two-thirds through the grain at right angles to the slip bands no slip bands but a considerable displacement of the polished surfaces of the individual grains from their initial location in the polished plane provided the failure was due to shear at the grain boundaries. On the other hand, if failure was due to forces having a tensile component normal to the grain boundary, the separation of originally adjacent grains would be shown. Where the conditions have been such as to produce fairly great plastic deformation it may even be possible to note the effect of intragranular flow in the changed orientation of the polygons marking the individual grains. Plastic Deformation 343 In unworked ductile metal there is no preference as to the direction of the longer diameters of the grains in any given surface nor are the diameters in various directions widely different. After plastic deformation however the originally equi-axed grains may be flattened into sheets, drawn out into threads, etc., etc., depending upon the character of the stress and the direction of the stress with reference to the polished Fig. 166. Intergranular failure of malleable Nitric acid etch 400 diameters Note that the surfaces of the different grains no longer seem to be in the same plane surface under examination. Of course it is obviously necessary that such changes of form can be detected best in a plane parallel to the direction of load and are visible only as changes of grain size in a plane normal to the deforming stress. Microscopic examination of the path of rupture through a metal, of the deformation of grains under load or, when applied to surfaces prepared before the application of the stresses, of intragranular slip and intergranular displacements is capable of 344 American Malleable Cast Iron interesting disclosures as to the mechanism of plastic deforma- tion or ultimate failure under various types of stress. Shows Two Systems of Slip Bands A very cursory summary of the changes in malleable is at- tempted in the accompanying photomicrographs. .Fig. 166 shows an unusual failure of intergranular type. It will be seen that at several points the grains have the appearance of being above or below the general surface. These grains have slipped not Fig. 167. Ferrite grains in malleable, showing slip in two planes at right angles Nitric acid etch 400 diameters Note the cohesion at grain boundaries even after severe plastic deformation on the crystal faces within the grains but at the surface of contact of adjacent grains. The field of view is near the com- pression side of a piece distorted by cross bending and it is pos- sible that this slip at grain boundaries produces the white so called compression fracture. The comparative absence of slip bands is interesting. Fig. 167 is reproduced from the tension side of the same piece and shows well developed bands. In some grains two sys- tems of bands are seen due to slip along two directions. The fact that the adjacent grains are not separated even under heavy strain shows the strength and ductility of the amor- phous boundary. Fig. 168, taken from a piece loaded in pure compression, shows that the structure of Fig. 166 is not always characteristic Plastic Deformation 345 of this type of loading and also shows plainly two systems of slip bands in practically every grain. In all of these photo- mi'crographs, note that the direction of slip is constant in any given grain, but is not usually the same in adjacent grains. The direction of slip has no direct relation to the direction of the stress but is determined by the direction of the crystallographic axis of the ultra microscopic crystals making up the individual grains. Figs. 169 and 170 show the distortion of grains in compres- Fig. 168. Slip bands due to plastic compression in malleable iron Nitric acid etch 500 diameters sion as seen on a polished section parallel to the direction of stress prepared after the distortion has occurred. The grains are much flattened as are the nodules of temper carbon. The grain boundaries are nearly obliterated but there is no separation of the adjacent grains. The effect is more strongly marked at the axis of the specimen than near the surface due to the fact that the barreling out of the specimen has permitted part of the reduction in height to be made by bending the outer fibers instead of upsetting them. The specimen from which these illustrations were made was compressed to a little less, than one-half of its original height. The effects of plastic deformation upon the grain structure can be destroyed by somewhat prolonged heat treatment below the critical point. By such treatment a new series of equi-axed 346 American Malleable Cast Iron Fig. 169. Plastic deformation of malleable in compression Nitric acid etch 100 diameters Field near axis of the specimen in a plane parallel to the stress. Note the flattening of ferrite grains, faint grain boundaries and distortion of temper carbon grains is formed, whose size depends upon the degree of the previous plastic deformation and the heat treatment adopted. Fig. 172 shows an axial section of a specimen similar to that shown in Fig. 169, after about five hours at 650 degrees Cent. While the ferrite becomes equi-axed and fine grained the deformation of the temper carbon still persists. Path of Rupture Shown Fig. 171 shows the path of rupture of malleable broken in cross bending. It was prepared by breaking a wedge-shaped piece by bending it over until fracture occurred. The fracture was then plated with copper, the specimen sawed in two at right angles to the ruptured surface and parallel to the cross bending stress and the exposed surface polished. Plastic Deformation 347 Fig. 170. Same specimen as shown in Fig. 169 Nitric acid etch 100 diameters Field near surface of specimen in plane parallel to stress. Note the difference from Fig. 169 in lessened intensity of all changes It is particularly interesting to note how the path of rupture goes far out of its way to include temper carbon nodules. This makes many deep depressions in the broken surface and due to the shadows in the bottom of these depressions produces the characteristic black fracture of the product. It is not often recognized that the presence of temper carbon is not a suffi- cient explanation of the black fracture -for this material, rep- resenting about 6 per cent of an average cross section, would not be nearly sufficient to darken the surrounding silver white metal. It is only due to the fact that the plane of rupture takes in many more nodules of carbon than would be found in an average section and in so doing produces a ,sort of "nap" that 348 American Malleable Cast Iron o c ^, 3 train D/aqromOrnalle obleC rast *~ *** Iron In Repeated Tension Under^. ^^ *>** 5000 Increasing Lood5 / X 1 / t 4500 4000 X] */ ' ]/ 7 r t / / ZO 000 c/) / f f T3 f . f f t i 17.500 > c 15 000 Q. o 12500 ^ Or, ,Q D 7500 ^ 5000 S- 2.500 A J f 50OO f / f t / & O 2500 / / / 1 A y f / I 000 yj f 7 i / / / 1500 / // . I / y/ ^ ,' // / J / / /J t > /^ f f ' f f '/ j / ^ /> / # / / . 001 .002 003 .004 005 .006 .007 006 .009 010 011 012 .015 ClongoTion In fight Inches Fig. 176. Stress strain diagram of malleable iron in repeated tension under increasing loads sorption of energy. Similar loops can be observed in malleable if the magnification be sufficient. They are barely visible in the diagram. The simplest case of stress reversing in algebraic sign is that alternating between tension and compression of equal intensity although alternate torsional shear is also of considerable im- portance. We have seen in the earlier chapters of this series that the behavior, at least within the elastic limit, of malleable 356 American Malleable Cast Iron in tension and compression, is similar; the proportional limit, being about 15,000 pounds per square inch and the modulus of elasticity about 25 x 10 6 pounds per square inch. Thus there is an elastic range of about 30,000 pounds per square inch, one half on each side of the neutral or unloaded condition through which the intensity of stress can be varied without plastic deformation. Applying Alternate Tension and Compression One of the simplest experimental methods of applying alternate tension and compression to a specimen is that of bending a beam to and fro in opposite directions. The be- havior of malleable under cross bending stresses has already been fully considered, notably the fact that ultimate strength and elastic limit determined in this manner bear no direct re- lation to these constants as determined in pure tension and com- pression. The explanation of this observation has also been detailed. The graph in Fig. 176 indicates the response of a malleable beam nominally J/ -inch' wide and 1-inch deep on supports 10- inches apart to alternations of stress. The deflections are plot- ted against apparent maximum fibre stress, as calculated from the known dimensions of the specimen and the applied load. When as the apparent proportional limit is not exceeded, the stress-strain diagram under this cyclic cross bending is merely a straight line through the origin at an angle depending upon the modulus of elasticity o the metal. However,, when the load in either direction exceeds the proportional limit the stress strain diagram becomes a curve, plastic deformation taking iplace. As the specimen is unloaded the elastic deformation alone is removed and at zero a certain permanent set equal to the plastic deformation remains. Elastic Limit Increases The effect of this plastic deformation is represented not only by the measurable permanent set but also by the increased elastic limit in the direction of the previously applied load. On reversing the direction of stress the elastic limit is en- countered sooner than it should be and the plastic deformation Plastic Deformation 357 begins at a lower stress than was the case in the unstrained metal. When an intensity of stress equal. to the previous maximum but of opposite sign is attained in a perfectly homogeneous specimen, an equal and opposite strain would ensue although in the present case the negative deflections all seem somewhat ?25 50 37.5 W l5 12.5 5 57.9 50 62.5" 6tre55ln Thousand Pounds Per SOuare Inch Fig. 177. Behavior of malleable under cyclic bending under increas- ing loads less than the corresponding positive ones. Action of Specimen On unloading the specimen it straightens out first elas- tically, retaining a negative set at zero load. Under reversed loads it finally .deforms plastically until at the stress corres- ponding to the first (positive maximum) it has the original deflection. 358 American Malleable Cast Iron Thus the cyclic cross bending stress-strain diagram is a spindle shaped loop whose area represent? the work done 1 in plastic deformation. Plastic deformation in a given direction raises the elastic limit in that direction and decreases the abso- lute value of this constant in the opposite direction, the elastic range remaining approximately constant. With successively increasing intensities of stress the area of this mechanical hystersis loop grows larger and larger as shown in Fig. 177. If instead of applying cyclic cross bending in a manner so that each cycle oscillates through a wider range of stress than the preceding one we merely repeat a given cycle indefinitely, it is found that the hysteresis loop decreases in area with suc- cessive cycles. Fig. 178 shows the first and tenth loops of such c.03 bl.5 50 37.5 25 125 12-5 5 315 50 625 Stress In Thousand Pounds Per Square IncH Fig. 178. Behavior of malleable under cyclic cross bending at constant maximum stress Plastic Deformation 359 ,160 Maximum Deflection 2,345 6 789 No. of Applications Fig. 179. Maximum deflection and permanent set under cyclic cross bend- ing at constant maximum stress a series. The decrease in work per cycle is due to the smaller plastic deformation in each successive cycle due to the hardening of the metal from the cumulative effect of all the slip produced. The decrease in deflection and permanent set is not at constant rate but decreases with each successive loading as shown in Fig. 179 and approaches a fixed minimum of finite size. The deflections and sets are shown to be different according to which half the specimen is in tension. This is presumably due to lack of com- plete symmetry about the neutral axis. The work done by a great number of such alternations will finally rupture the speci- men. This constitutes the phenomenon of fatigue. The phe- nomenon of fatigue of metals so far as it is known has been discussed in another chapter. The experiments just re- corded having shown the approximate extent to which tensile or compressive loads strengthen the material for subsequent loads 360 American Malleable Cast Iron in the same direction and weaken it for loads of opposite sign. From these experiments we can gain at least a qualitative insight upon the effect of a previous cross bending upon subsequent tension or compression in a direction parallel to the length of the specimen and vice, versa. The quantitative interpretation is impractical perhaps impossible owing to the difficulty of ac- counting for the distribution of stress in a plastically strained material. Behavior of Specimen Consequently under the subsequently applied longitudinal stress the elastic limit will be first exceeded on that edge of the specimen which is experiencing a reversal of stress. As the applied longitudinal load is increased a greater and greater por- tion of the area experiences plastic deformation until finally the elastic limit also is reached at the opposite edge. At intermediate intensities of stress in a portion of the spe- cimen elastic strain exists, in another portion plastic strain. From the nature of the case the ratio of strain to stress is greater for plastic than elastic deformation. The side experiencing a re- versal of stress will stretch or compress more rapidly and an eccentricity of loading will result from the unequal strain dis- tribution. Such an eccentricity in the case of compression will result in the superposition of a bending moment on the longi- tudinal stress, as in the case of columns which are eccentrically loaded and a given load will produce far greater unit stresses than might be expected. In the case of tension the eccentricity of loading will re- sult in the transfer of a disproportionate amount of load to a few of the stiffer fibers with an accompanying high unit stress. Conversely the effect of a previous longitudinal stress upon subsequent cross bending loads is to shift the neutral axis to- ward that surface of the specimen which is being stressed in the same sense as the first load. This shift goes on until the moment of resistance of the portion of the specimens in opposite sides of the axis about the axis are equal. The sum of the two moments, constituting the moment of resistance is thereby decreased. In either event, although we may not be able to solve numerically the complex mechanics we Plastic Deformation 361 may draw the conclusion that cross bending weakens the ma- terial for subsequent tension or compression and vice versa. The practical application of this conclusion is that a detail which in fabrication has been subjected to severe cold work cannot be expected to be as strong under loadings involving a reversal of the stress previously encountered as unworked metal would be. This conclusion applies equally to all ductile materials and should serve as a warning against needlessly energetic straightening or beading operations. Many malleable castings are cast to a simpler form than intended and then bent to the more complex shapes demanded. Air brake hose clamps are examples of this practice. Such parts will never develop the full strength of the original metal. In all the preceding cases the loadings have been such as to set up strains parallel to the subsequent stresses. A variety of circumstances are possible in which the final load has no component parallel to that producing the plastic deformation. Two typical cases are torsional shear followed by tension and compression followed by tension or compression in a di- rection normal to the first compression. Compression followed by a cross bending load parallel to the direction of compression is, of course, a special case of the preceding involving both tension and compression. The combination of compression followed by tension, com- pression, or both, normal to the original strain is the condition which may arise where a piece is reduced to the desired dimen- sions by compression in a press rather than by machining. In Fig. 180 are shown two stress-strain diagrams on specimens nominally ^>-inch square subjected to cross bending load on supports 10 inches apart. One specimen, A, is of normal metal in its original condition, while the other, C, was produced from a thicker bar by compressing it to a final depth of J/ inch. The compressed dimension is vertical, that is, parallel to the direction of the load in the final test. The effect of relatively heavy compression under these circumstances can be learned by a comparison of the two graphs. A few scattering tests of the effect of shear upon subse- quent tensile stress have been made. 362 American Malleable Cast Iron In Fig. 180 certain tests of this character are tabulated. Standard A. S. T. M. tension specimens were twisted through various angles and then broken in tension. In the illustration the angle of twist under load is plotted against the tensile properties of the resulting metal. It will be noted that a rapid and continuous decrease in elongation is encountered with increasing torsional deformation. The tensile strength first rises rapidly to a maximum and then decreases still more rapidly. The location* of the maximum Load In Pounds Ai Center oggS3S .osSoooooo ' - 9 -1 -* - ' Stress Deflection Dioqram Of Malleable Iron InCro55 Bend nq With And Without Previous Cold Work ^ - -< ** ^ & r _ * ^ r* 1 * * ^5p trr\ en c rf / ^* *o -- 5p ec ^= rru p- nA^ ^ *' f* 1 -3 ^ S s s' ^ S X * ^ ^ JOl .01 .OS .04 .05 Ob 07 .00 DeflnlncbeslnlO .09 .10 .11 .12 .15 .14- 15 .6 "Span Fig. 180. Stress deflection diagram of malleable in cross bending with and without previous cold work tensile strength corresponds approximately to the torsional yield point, as may be seen from the torsional stress strain diagram. The curve suggests a hardening of the metal due to the forma- tion of amorphous metal followed at higher strains by disrup- tion at the grain boundaries. Failure in tension after great torsional strain did not result approximately normal to the axis but in a spiral surface ap- proximately normal to the helix angle into which the originally straight elements of the specimen have been strained. There is a suggestion here that distortion is not due to pure shear. We have considered the effect of a series of stresses of known intensity and direction upon a ductile material. Another Plastic Deformation 363 important condition is that in which, instead of a series of known stresses the specimen is required to undergo a series of known increments of energy. Impact testing by a series of equal or increasing blows, is the principal application of this type of plastic deformation. 90 180 70 360 450 540 680 7EO 6)0 900 990 1080 Angle Of Torsion (Degrees) In 4' Goqe Length Fig. 181. Effect of torsional deformation upon subsequent tensile strength of malleable In this case the intensity of maximum stress is a function both of the energy input of the blow and the elastic and plastic de- formation of the specimen. The latter factor depends upon the previous plastic deformation of the specimen and hence is a function of the magnitude and number of the preceding series of inputs of energy. 364 American Malleable Cast Iron Since malleable is often subjected to repeated impact in serv- ice and occasionally in testing, this condition is of special im- portance in connection with a study of that metal. If the load deformation curve of a given specimen under plastice deformation were capable of mathematical definition in terms of its dimensions and properties and the rate of applica- tion of the load, a mathematical study of this problem would be feasible although probably quite complex. However, the problem may be simplified by assuming that Pef/ecfrtn Fig. 182. Absorption of energy from successive impacts we have experimentally determined the load-deformation dia- gram of a given specimen under given conditions. The load- deformation diagram in every respect is similar to a stress- strain diagram except that the co-ordinates are actual load and actual deflection instead of unit stress and unit strain. We can conceive that for a given specimen such a graph might be autographically produced under rates of application of load as rapid as are encountered in impact testing. Referring to Fig. 182 let OLU represent the load-deforma- tion curve described above, L being the elastic limit and U , the ultimate strength and tan the modulus of elasticity. Then the energy imparted to the specimen at any given load and deforma- tion for instance is the area below the curve beginning at O Plastic Deformation 365 and ending at a n . For example, OLa a 2 etc., a n b n . If this energy input be large enough the point a n will then reach U, the energy being the represented by OLa^a 2 etc., UV and this energy will produce rupture under impact. Therefore, if impact is produced by a single blow, the en- ergy of rupture is measured by the entire area below the curve as shown above. A blow having an energy of impact of OLM or less will not produce a plastic deformation, the specimen will return to its original form after the load is removed and will have absorbed no energy. If the energy of impact be equal to OLaJ)i for example, when the load is removed the deformation will decrease along a^ (parallel to OL) and a permanent set Oci will remain. The energy OLa^c^ will have been used up in plastic deformation and the elastic limit will be raised to a 1 and the deflection at the elastic limit to c-J)^. The new load de- flection curve becomes caa etc., UV. Thereafter any impact of energy not greater than c^ajb^ will produce elastic deformation only. Suppose the second impact is equal to C 1 a 1 a 2 6 2 then by similar measuring the new load deflection curve becomes C 2 a 2 a 3 a n etc., UV the third impact moves it to c 3 a 4 a n etc., UV and so on, and after n blows it becomes c n a n UV and finally perhaps WUV in which case a blow equal to or greater than WUV will break the specimen. Suppose now that we assume an equal energy input with each blow. Then OLaJb^ C 1 a 1 a 2 & 2 = C 2 a 2 a 3 b 3 etc., cn- 1 an- 1 an bn. It is obvious by inspection that up to the point of maximum load G each succeeding one of the similar tri- angles caji, C 2 a 2 b 2 etc., is of larger area than its predecessor. These triangles represent the portion of the energy of impact expended on elastic deformation. Consequently a smaller per- centage of the constant increment of energy is available for plastic deformation with each succeeding blow up to that pro- ducing maximum deflection. Beyond this point an inqreasing- ly larger amount of each energy increment is available for plastic deformation. Finally if ^c^a 2 c 2 + C 2 a 2 a 3 c 3 c n .^an ^a n c n etc., is com- mensurate with OLGUW the specimen absorbs on the last blow energy equivalent to UVW '. 366 American Malleable Cast Iron The specimen has then absorbed plastically the energy OLGUV which it would have absorbed if broken by a single impact. Since, however, the area Cn-^n^an c n is always less than the area c,,.^,.^^ b the energy absorbed by the metal at each blow is measurably less than the total energy of im- pact, a large part of the energy of impact being returned by the specimen during its elastic recovery. Obviously since there is a definite amount of energy not ab- sorbed by the specimen at each blow a smaller percentage of the energy of impact is absorbed the lighter the blow. If the energy of rupture be measured by the aggregate of the energy of the entire number of blows to produce rupture this sum will be higher the smaller the individual blows. Consequently testing a metal by successive impacts can yield quantitatively compar- able results when all the specimens are identical in form and quality in addition to the constancy of the hammer blow. Of course this condition is impracticable of attainment, the quality being unknown before the test. In practice this means that only carefully prepared speci- mens of similar material are capable of fairly accurate com- parison by repeated impact test. One or two further conclusions may be gained from the study of the diagram. Energy equivalent to the area OLM is absorbed by the specimen elastically. The material will with- stand an indefinite number of impacts of this magnitude with- out permanent deformation. Were a similar triangle FGH drawn with its apex at G , this area will represent the maximum elastic absorption of energy the specimen can sustain when by repeated impact the elastic limit has been raised to the ultimate strength. Any increment of energy less than this will never fracture the piece but will produce a maximum deflection after a given number of blows which will not be further increased by further repetitions of the impact. The area OLM LM. LM tan 0=modulus of elasticity X square of elastic limit. The area FGH = GH. CH tan 0=modulus of elasticity Plastic Deformation 367 X square of ultimate strength. From the above we may calculate the blows required to make an impact test workable on a given specimen. The deflection at each successive blow can be determined graphically under given conditions from the diagram. An impact test in which the energy increment increases with each blow can be studied in a similar manner. In that case there is no possibility of coming to a maximum deflection without frac- Fig. 183. Load deformation diagram of. specimen subjected to alternate impact ture for the increased energy of the succeeding blow would carry the deformation beyond G. In such a test there is great danger that the last blow will be equivalent to far more than the energy WUV and the unabsorbed energy of the blow will be credited also to the specimen. We may generalize to the effect that no method of repeated impact can correctly measure the energy of rupture of a duc- tile metal. In a similar manner we may study graphically the 368 American Malleable Cast Iron effect of alternate impact in opposite directions, although we may be confronted with the difficulty of securing the necessary load deformation curves. In Fig. 183 U^L^OLU is the original curve for the specimen. An increment of energy Oa^b^ de- forms it to a and raises the elastic limit to that point. The load-deformation diagram then becomes a- l O- l L L JJ' 1 ^ and an in- crement of energy in the opposite direction to the first O l L 1 1 a 2 b. 2 produces a load of a 2 b 2 and a deformation O^b^ The new elastic limit becomes a 2 and the new diagram a 2 O 2 U 2 . The next increment of energy is diagramed as O 2 a 3 B 3 and so on. It will be seen that each impact in one direction appar- ently decreases both the ultimate load and elongation in the op- posite direction an expression of the weakening caused by a negative plastic deformation. In the absence of stress strain diagrams under dynamic loads we may turn as the best available substitute to the vari- ous stress strain diagrams given throughout these chapters and from them and the dknensions of the specimens estimate the probable load deformation curves to be used. It is obvious that those materials in which the elastic limit is quite high accompanied by a high elongation are these which will well resist repeated impact. The high elastic limit will dissi- pate a large amount of energy in elastic deformation at each blow while the high elongation provides a large amount of re- serve energy for plastic deformation before rupture takes place. The Young's modulus of all ferrous materials is practically the same, hence the deformation at the elastic limit is in direct proportion to the elastic limit. In steel high elastic ratio is ob- tained only at the expense of elongation and vice versa. The various graphs for malleable, indicating a constant and high elastic ratio and an elongation increasing with strength account for its excellent behavior under repeated impact even when of sufficient magnitude to produce plastic flow. In this connection, incidentally the yield point of metal is the governing factor in ferrous materials for the small reduc- tion in the area representing energy clue to the curvature of the stress-strain diagram between the proportional limit and yie!cl point is negligible. Plastic Deformation 369 Plastic deformation has been discussed mainly because of its great importance in the utilization of malleable. No one realizes more than the author the unsatisfactory state of knowledge and the lack of precise numerical data. If this chapter has enabled the reader to form even a qualitative image of the resistance of the metal above the elastic limit that is all that can be expected. An infinite amount of further study will be required before concrete mathematical analyses will be possible. XIX THERMAL AND ELECTRICAL PROPERTIES WHILE it is true that materials of construction in gen- eral are used to resist mechanical stress, yet there are service conditions when other properties, such as ther- mal, chemical or electrical, for instance, are of greater conse- quence. The most important condition of this kind arises in the use of malleable as a material for field frames of electrical apparatus, where the magnetic characteristics of the metal are much more important than the mechanical strength. It is a well known fact that if a coil o wire is wound around a piece of iron and a current is passed through the coil, the iron becomes magnetic. This property of iron, which it shares in a very limited degree with a few other metals, is of im- portance in electrical machinery. If the power to become magnetic is the quality desired, evidently the metal which forms the strongest magnet with the same coil and current is the most valuable. Therefore it is desirable to determine the de- gree to which a given material possesses this valuable property. Avoiding a discussion of the electrical principles and of the mathematical reasoning involved in the study of magnetism, it is sufficient to say that the intensity of magnetization, repre- sented by the symbol H f and expressed in gausses (lines per square centimeter) can be calculated from the dimensions of a magnetizing coil and measurement of the current. When an iron core is inserted in the coil it will be found that the inten- sity of magnetic field is much greater than the calculated value H. This higher value, known as magnetic induction, is sym- bolized as B and is measured in the same units as H. The ratio of B to H, that is, the number of times stronger the magnet is with the iron core than without any core, using the coil only, is called the permeability of the material and is the variable represented by the Greek letter /*. It is further found that the value of /* depends not only 372' American Malleable Cast Iron upon the material 'being used but also on the value of H at which the experiment is made. In general, the permeability of a material first increases as H increases, soon reaches a maximum and then falls off, first rapidly and then more and more slowly. The value of /* for an indefinitely strong field is prob- ably 1 for all materials. Owing to experimental difficulties determinations close to the zero value of B are not very reliable. The behavior of a magnetic material is usually repre- sented by a so-called magnetization or "B-H" curve in which the value of H, the strength of the magnetic field, is plotted horizontally and the magnetic induction in the iron, B,, which is equal to v-H, is plotted vertically. The fact that /* is vari- able, depending on B and hence on H, gives this curve a gen- eral form which rises from the origin (H = 0, B =0) first ait a rapidly increasing rate as H increases and then more slowly until it becomes horizontal when H is infinite. As a matter of fact the curve becomes nearly horizontal fairly soon, and /the "knee" in the curve, somewhat resembling the yield point in a tensile stress strain diagram, represents practi- cally the maximum flux density which can be attained in a given metal. This value varies widely in different metals and is quite definite in each metal having almost the significance of a physical constant. This characteristic for malleable iron is shown in curve A, Fig. 184. The specimen was in the form of a closed ring about 6 inches in diameter and having a rectangular section 0.33-inch thick radially and 0.9-inch wide. The per- B meability, A* = H for various values of //based on the data for the ring described above, is shown in curve B. The values of /* as related to B are plotted in curve C. When a material has been magnetized and the magnetic field H is then reduced, the magnetic induction B in the iron decreases but not at the same rate as it increased with increasing values of H. When H is reduced to there usually remains a considerable magnetic induction and it is only Thermal and Electrical Properties 373 after H has reached a definite value in the opposite direction to that first developed that B falls to 0. This lag of induction behind magnetizing force is due to hysteresis. The value of B when H is reduced from a high 500 10000 CO 5000 500 10 20 3 40 H in C.G. 5. Units Fig. 184. Magnetization and permeability curves of malleable cast iron value to is called residual magnetism, and the negative value of H required to bring B to is called the coercive force. It is quite possible to plot a curve, similar to a B-H curve beginning with a fairly high value of H, lowering H gradually to 0, then increasing it in the opposite direction until a nega- tive value is reached equal in magnitude to the original posi- tive value, then back through to the starting point. Such a 374 American Malleable Cast Iron curve forms a closed loop of distinctive form called a 'hysteresis loop. The area of this loop represents energy consumed in magnetizing and demagnetizing the specimen. Materials strong- ly retaining their magnetism, and therefore suitable for per- manent magnets have a larger hysteresis loop due to great residual magnetism and coercive force. Material for electro- magnets, especially where frequent changes in magnitude or sign are required in field strength have the opposite characteris- tics. This energy is dissipated as heat, either in raising the tem- perature of the iron or radiated to the surroundings. The loss is of industrial importance for service involving reversals of magnetism in that it involves a waste of energy and may re- sult in inadmissably high temperatures being reached in the magnetic circuit, possibly sufficient to destroy the insulation on the coils. The energy is lost once for each cycle of magnetiza- tion so that for alternating currents the loss depends on the frequency. It can be shown mathematically that the energy dissipated per cycle of magnetization per cubic centimeter of metal is the area of the hysteresis loop divided by 4 71 ", regard being had of course to the scale to which B and // are plotted. This value is necessarily dependent on the magnetic induction ob- tained. In Fig. 185 a condition is plotted in which saturation has practically been attained, hence calculations based on this graph would give the energy dissipated by a complete cycle. The area actually corresponds to a value of 11,388 ergs per cubic centimeter of metal. Cyclic magnetization of malleable to an inductance of 13,200 centimeter-gram-second units by the usual 60-cycle alternating current would raise the temperature of the iron a little over 2 degrees Fahr. per minure, assuming no radiation of heat. Steinmetz has determined empirically that the work done in a cycle of magnetization on any given material is approxi- mately proportional to the highest magnetic induction, B reached (in the cycle raised to a power between 1.66 and 1.70. This formula serves to derive the work done on the same material by cycles ending at different inductions. Therefore, the u v- Thermal and Electrical Properties . 375 / Current in Ampere^ n* Number of Turn;, 12000 20 - 20 Intensity of Magnetization, H 40 Fig. 185. Magnetic properties of malleable cast iron teresis loss on any given material, is a constant times J5 1 - 68 when B is the maximum induction reached in the cycle. This constant varies with different materials and is designated by the Greek letter ^. Calculation from the preceding data gives a value of 0.00136 for Steinmetz's constant. This very low value is logically due to the fact that the 376 American Malleable Cast Iron bulk of a malleable casting is a fairly pure ferrite contaminated mainly by silicon whose presence is an advantage and also to the fact that the anneal involves a heat treatment consistent with the very softest condition of ferrite possible. So far, the writer knows of no case where the electrical resistance of mal- leable is of commercial importance. It has been roughly deter- mined to be 0.000044 ohm per centimeter cube. More recent and accurate data indicate the specific resistance to be 0.0000295 ohms per centimeter cube. A part of this descrepancy no doubt is due to the heterogenous character of the material. The newer value however is much more reliable. Presumably the resistance decreases with the carbon content. The change in resistance with temperature is shown in Fig. 186, the resistance at room temperature being taken as unity. Where metal parts are exposed to weather or to the action of water or steam, circumstances arise in which the resistance of the material to rusting is of prime importance. This is particularly true under circumstances which preclude the use of paint, galvanizing and similar means for protecting the metal. This opens up the moot subject of corrosion of iron and the relative merits of accelerated tests in dilute acid as com- pared with service tests. All commercial iron alloys, except a few high-silicon metals, dissolve in acids more or less rapidly. While not at the same rate for all forms of iron and steel the deterioration is rapid enough to preclude the use of ordinary ferrous materials for corrosion resisting services. A great many acid corrosion tests have been conducted on malleable but the results are hardly applicable to the present discussion. It is generally admitted that since corrosion is an electrolytic phenomenon, the more nearly homogeneous a metal is the better it will resist corrosive action either of the elements or of acids, salt water, etc. Manganese sometimes is alleged to be an offender in start- ing corrosion. The surface of a malleable casting is always nearly carbon free; it contains rather small amounts of man- ganese, less than any material except wrought and ingot iron. Silicon is supposed to dissolve in ferrite, when present in mod- Thermal and Electrical Properties 377 erate amount. It would appear therefore that malleable should resist rusting moderately well. This general conclusion is borne out by the fact that malleable has been used for many years in the manufacture of pipe fittings, radiator nipples, etc., and complaints that the material has failed by rusting are very rare. Resistance at Temperature tr Resistance at Room Temp. o o b f S / / / i ^ j / 4 / / f f c y / iY Q f { / / / 6 I / tff ., x" & 1 x-* -X 1 x* 100 200 300 400 500 600 TOO Degrees Cent. Fig. 186. Variation of electrical resistance of malleable cast iron with temperature There is also of record the case of a malleable iron harness part which was found in excavating for a foundation. The circumstances were such as to make it certain that the article had been in the soil over 40 years, yet it had suffered buft little injury to the surface. The only service test with which the writer is familiar was conducted to determine the relative life of malleable and steel railway tie plates. Plates of both ma- terials were laid in the same track at the same time. When 378 American Malleable Cast Iron the steel plates had completely rusted away the malleable plates were still practically in their original condition. It seems rather doubt full whether in the present state of our knowledge any quantitative method exists of measuring resistance to corrosion other than a direct comparison under the conditions expected in practice. In a great many cases mechanism is required to function under temperature conditions either abnormally high or ab- normally low. The principles to which malleable owes its properties indicate obviously that malleable cannot be ex- posed to temperatures above Ac^ even momentarily, without being permanently destroyed. The question of its use at high temperature cannot be dismissed merely with the statement that it should never be exposed, even momentarily to temperatures higher than say 1300 degrees Fahr. lest by chance Ac be overstepped and a permanent change be produced in the metal. There are many cases where castings are to be used at temperatures considerably below the danger point and the designer -must guide himself by the effect of temperature on the properties of the material. Even so simple a property as the dimensions of a casting are affected by variations of temperature. Experiments by the author have shown that if L be the length of a malleable cast- ing at degrees Cent, when the casting is raised to a tem- perature of t degrees Cent, its length L t will be given by the equation L t = L (1-KOQ0006 H- -0000000125 t 2 ) Translating into terms of Fahrenheit temperature the re- vised formula becomes L t =L 32 [1+. 0000033 (/ 32)+. 00000000385 (t 32 ) 2 ] These figures are somewhat cumbersome. For engineering purposes it may be more convenient to take the expansion at various 'Fahrenheit temperatures in per cent of the length at 75 degrees Fahr. from the graph, Fig. 187. It is to be noted that the change in size of large castings where raised to mod- erately high temperatures is quite significant. Thus a cast- ing 3 feet long when raised to 600 degrees Fahr. expands over 0.1 inch which may be very important where clearances Thermal and Electrical Properties 379 are to be allowed. The author is not aware of any actual or experimental determinations of the specific heat of malleable cast iron. Since the material is a mechanical mixture of graphitic carbon and nearly pure iron we may use provisionally data calculated from the known constants of the two elements. The conductivity of a metal for heat represented by the 0.30 0.60 & i |0.40 a. K UJ s * Jj 0.20 C DC m ir I tted Curve Plate the V the Equation if=L ( where L - Length a Lf - pansion Measured '/ Percentage of Leng ilue 0. l+0.< to*c tc nS'Re that 7 000006 700006 corded '5'F t + O.OOL t+o.oa \ 1000012 OOOOOli st* >5r 2 ; / ( / ^ / / / / [/ V 2 / ^i / X ^o* ' X ) 200 400 600 800 1000 1200 Temperature, dcg Fanr Fig. 187. Expansion of malleable cast iron symbol k is defined as "the quantity of heat, in small calories transmitted through a plate 1 centimeter thick per square centi- meter of surface when the difference in temperature between the faces is 1 degree Cent. The heat transmitted through a plate of metal varies di- rectly as its area and as the difference in temperature between the faces and inversely as the thickness. The value of k varies slightly with the temperature, de- 380 American Malleable Cast Iron creasing for iron and increasing for carbon as the temperature rises. At room temperature (17 or 18 degrees Cent.) the values for k for iron and graphite are .161 and .037, respectively. (Smithsonian Physical Tables, 1921.) At that temperature malleable cast iron of 2 per cent to .006 005 X or 05 0' Fig. 188. Heat transfer from machined malleable to still water for various temperature differences total carbon should have a value of k between .1578 and .1585, depending on how readily heat can be transmitted from car- bon to iron and vice versa. On the same authority for the interval between 100 and 720 degrees Cent, the value of k becomes .202 for iron, .306 for graphite, and between .198 and .204 for malleable iron. The values are higher than certain approximate experimental values determined in the author's laboratory. Malleable heated above A, will have its thermal conductivity permanently de- Thermal and Electrical Properties 381 creased since this constant decreases with the combined car- bon content. The specific heat of a substance is the quantity of heat in small calories to raise the temperature of one gram 1 degree Cent. Iron at 37 degrees Cent, has a specific heat of .1092 (loc. cit.) and graphite at 11 degrees Cent, a specific heat of .160. As a mechanical mixture of 98 per cent iron and 2 per cent graphite and neglecting corrections for a change of specific heat with temperature, the specific heat of malleable at room tem- perature should be .1102. The value probably is quite accurate, since cast iron of about 3%. per cent Cent, has a specific heat of .1189, The specific heat rises with the temperature. In view of the approximate character of these deductions and of their intended application a detailed study of the rela- tion between temperature, thermal conductivity and specific heat seems "unwarranted. All ferrous metals grow softer and weaker at elevated temperatures. Accordingly it becomes important to know the quantitative effect of temperature upon strength in order that where very high temperatures are unavoidable, due allowance may be made in design for the changed physical properties at the higher tempera Lures. Since the tensile properties can be more definitely measure.'l than any other, studies on the effect of temperature on strength have usually been made on tensile specimens. The author has conducted experiments of this character by breaking very care- fully made specimens at temperatures from 80 to 1450 de- grees Fahr. The data up to 1200 degrees Fahr. the highest commer- cially safe temperature to provide against the possibility of heating up to a temperature which will permanently affect the product are shown in Fig. 189. It will be seen that malle- able cast iron has tensile properties equal to those it possesses at room temperature at all temperatures from 100 to 800 degrees Fahr. Above 900 degrees the strength decreases rapid- ly and at 1200 degrees the maximum allowable temperature, the 382 American Malleable Cast Iron metal is onfy one-fifth as strong as at room temperatures. Pre- sumably very 'similar relationships will be observed under other loads, compression cross bending, etc. Temperature affects the magnetic properties of iron. For large values of H, B decreases as the temperature increases; the reverse is true for very small values of H. The effect of the temperatures is not strongly marked at room tempera- -100 200 400 600 800 1000 1200 Temperature j deg. Fahr Fig. 189. Effect of temperature upon tensile properties of malleable tures but increases rapidly as' the temperature goes beyond 1200 degrees Fahr. Presumably the behavior of malleable is in ac- cord with these principles. Actual measurements are lacking. The specific heat of malleable, that is the number of heat units required to raise a given weight of that material 1 de- gree in temperature as compared with the heat units to raise an equal weight of water 1 degree varies from 0.11 at 75 degrees to 0.165 at 800 degrees Fahr. The intervening curve is near- ly straight, being but slightly concave upward. The values are calculated from the specific heats of iron and carbon. Malle- able, being a mechanical mixture of these two elements, can have this constant calculated in that way. As the name implies, the thermal conductivity of a metal Thermal and Electrical Properties 383 is the rate at which it will conduct heat. The constant is de- fined in terms of the quantity of heat conducted per unit of time through a cross section of unit area of a slab of unit thickness whose opposite sides differ by unity in temperature. The quantity of heat conducted varies directly as the area of the conductor and as the temperature difference between its ends and inversely as its length. However the thermal conduc- SJ& % J36 4 ./ -^La6