3982 Bollard Avenue Cincinnati, Ohio 45209
Although much has been written about engines, comparatively little had been reported about the iron and steel used in their construction. This article seeks to remedy that situation.
Today's chemistry and physics surpass what was known in the late 1800s and the earliest years of the twentieth century. All the same, the designers and builders of boilers and engines had amassed an impressive array of scientific facts and formulas. This essay presents the story of steel and iron as it would have been told around 1900.(In other words, take heed! A few of the 'facts' may not be facts anymore.)
In the last decade of the nineteenth century, publishers brought out several textbooks for mechanical engineers to follow in taking advanced college courses and in working for companies which built engines. These weighty tomes provide a historical record of what was deemed of utmost importance for students and practitioners of engineering to know at the height of the industrial era. Such books devote numerous chapters to descriptions of iron and steel. Apparently, the term 'iron' served as (a) a generic classification for a wide range of metallic alloys including steel, (b) a more specific designator of certain metals which differed from steel in molecular structure and (often) in manner of production, and (c) the very specific signifier of the silvery-white, malleable, ductile element symbolized by Fe. With meanings from general to specific, 'iron' in old textbooks can be, and occasionally is, a highly ambiguous word.
Despite such confusing terminology, the engineering books from around the turn of the century provided much useful instruction to readers born not long after the Civil War. Most begin their treatment of the subject of iron by examining the nature of stresses, both longitudinal and transverse. Under the former heading belong tensile stress, which resists a pulling force, and compression, which resists a crushing force. Beneath transverse stresses may be classified (a) shearing stress, which resists cutting across, (b) bending stress, which resists breaking across, and (c) torsional stress, which resists twisting (Thurston 46). Knowledge of such forces enabled engineers to design stronger boilers and engines. With stress as a force acting upon a structure, strain, according to Robert H. Thurston, was the resultant change of structural form (47). Unfortunately, general agreement on such a seemingly simple matter as defining 'stress' and 'strain' could not be reached:
An external force applied to a body, so as to pull it apart, is resisted by an internal force, or resistance, and the action of these forces causes a displacement of the molecules, or deformation. By some writers the external force is called a stress, and the internal force is a strain; others call the external force a strain, and the internal force a stress: this confusion of terms is not of importance, as the words stress and strain are quite commonly used synonymously, but the use of the word strain to mean molecular displacement, deformation, or distortion, as is the custom of some, is a corruption of the language. (Kent 236)
Then as now, science progressed despite disagreement. On the subject of strain, Thurston and William Kent parted company.
Large numbers of engineers necessary to keep pace with industrial growth were learning from textbooks, not from apprenticeships, as had been customary prior to the late 1800s. The thick volumes they studied helped them to comprehend categories of load: (a) working load, which is applied to a piece proportioned to bear the force securely, (b) dead load, which is applied without shock and endures without change (example the weight of a bridge), and (c) live load, which is applied abruptly with inconsistent stresses (example a train crossing a bridge) (Thurston 47). In turn, students discovered the intricacies of 'elastic limit.' Kent explained:
The elastic limit is defined as that point at which the deformations [in a bar of metal being tested for tensile strength] cease to be proportional to the stresses, or, the point at which the rate of stretch (or other deformation) begins to increase. It is also defined as the point at which the first permanent set [morphologically-fixed elongation] becomes visible. The last definition is not considered as good as the first, as it is found that with some materials a set occurs with any load, no matter how small, and that with others a set which might be called permanent vanishes with lapse of time, and as it is impossible to get the point of the first set without removing the whole load after each increase of load, which is frequently inconvenient. The elastic limit, defined, however, as the point at which the extensions begin to increase at a higher ratio than the applied stresses, usually corresponds very nearly with the point of first measurable set. (236-37)
Thurston agreed with the latter definition that the elastic limit is the point at which the rate of distortion more or less suddenly changes, with the piece approximating permanent change of form; however, Thurston quickly added that the limit is not so observable as a point but more as a 'curve' of change (51). Then, continued Thurston, if the stress is relaxed, the piece recoils. Once all load is removed, recoil stops, and the piece, having been stretched beyond its elastic limit, assumes a permanent set without recoiling all the way to its original position. Elongations under stress and sets after stress are proportional to loads. Also, Thurston advised, metals respond to live loads differently than to dead loads or to loads applied slowly (53). Kent stated, 'Under a load applied suddenly the momentary elastic distortion is equal to twice that caused by the same load applied gradually' (238). Matthias N. Forney concluded that a bar of iron or steel can withstand up to about one-fifth the amount of force which would break the bar and still return to its original length after stretching. He pointed out that repeated applications of excessive force can further elongate the bar, with each new set approaching the breaking point (also known as 'ultimate resistance' or 'ultimate strength'). Along the way, the metal becomes more brittle, less able to withstand a sudden strain (188).
Working independently, Thurston and Commander L. A. Beardslee of the United States Navy, discovered that, if wrought iron were subjected to a stress in excess of its elastic limit but not beyond its ultimate resistance and then were allowed to rest by releasing it from all stress or by holding the test piece at a given intensity of force, the wrought iron's elastic limit and ultimate resistance would increase considerably (Kent 238). In 1873, such a discovery was news. Engineers benefited from knowing that stress followed by rest could increase the resistance of wrought iron. Designers and builders of engines studied stresses, loads, elastic limits, and resistance quite carefully. They learned that a compression lowers metal's ability to withstand a future tension (Kent 239). They remained well aware that one big stress would break a bar of metal but so would several smaller repeated stresses; they recognized that iron shafts had snapped from a succession of little shocks which otherwise would seem harmless (Kent 240). They concentrated on such data so as to know how best to plot factors of safety.
Kent recommended that a factor of safety of no less than six be applied in situations when stresses would alternate between tension and compression. When stresses would be unpredictable and complex, as in the forces acting on a crankshaft of a reversing traction engine, the factor of safety would need to be much higher perhaps as high as forty! Kent also suggested that builders increase any sections of metal liable to deteriorate through corrosion (315-16). Clearly, engineers around 1900 hoped to design machines which would last.
Thurston argued that the strongest metals for constructing steam boilers exhibited (a) tenacity (obviously), (b) ductility, (c) elasticity, and (d) resilience. Boiler iron or steel, he contended, 'must have ductility, even if tenacity is sacrificed to some extent. . . .'(56-57). Pulling together such broad subjects as stress, load, limits of elasticity, and breaking points, Thurston summarized:
The problem of proportioning parts to resist shock is seen to involve a determination of the energy, or 'living force,' of the load at impact, and an adjustment of proportion of section and shape of piece attacked such that its work of elastic or of ultimate resilience, whichever is taken as the limit, shall exceed that energy in a proportion measured by the factor of safety adopted. (57)
In the U.S., normal factors of safety used in designing steam boilers varied from 3 up to 5. A so-called factor of safety of 6 (often a 'talking point' in catalogues advertising engines) might or might not apply to the boiler's real integrity; usually, the 6 referred to the full tensile strength of the iron plate, not to that plate bent and riveted. In boilers, the stresses on roundabout seams and the more severe stresses on longitudinal seams caused the actual factor of safety to be lower than that of a pristine, unriveted plate (Power 38-39). The calculation of efficient but appropriate factors of safety challenged the designers and builders of engines:
When heavy shock is anticipated, calculations of energy and resilience are necessary, and these demand a complete knowledge of the character, chemical, physical, and structural, of every piece involvedin other words, a complete knowledge of the material used, of the members constructed of it, and of the circumstances likely to bring about its failure. (Thurston 57)
Small wonder, then, that the most successful engine manufacturers employed teams of chemists and hired mechanical engineers having enviable backgrounds in physics.
In the 1870s, such manufacturers wanted to decrease their reliance on wrought iron as the metal of preference for boiler construction and to switch to steel; however, steel presented two formidable obstacles. At 'black heat,' a term defining a stage in the process of gradually increasing the temperature and thereby softening metal, mild steels became brittle and could not be worked without difficulty (Holmes 400). Secondly, while the elastic limit of steel exceeds that of iron, steel remains less ductile than ironand ductility is essential in boiler plate (Thurston 59).Once technology overcame the problem occurring at black heat, factories began to experiment with mild, reasonably ductile steels for boilers.
Builders of engines complained how the quality of both iron and steel varied considerably from lot to lot and even within the same sample; the physical character and structure often differed so markedly from batch to batch that the naked eye could detect the dangerous variations (Thurston 59). By about 1900, the processes for producing industrial metals became more refined.
Cast iron was obtained from a fluid mass, wrought iron from a pasty mass. Practically obsolete by the turn of the century, wrought steels represented experimentation to derive a useful metal from puddled steels. True steels were produced by the expensive crucible process, by the Bessemer method, or by the Siemens-Martin open-hearth technology; such steels were cast, malleable irons, in the generic sense of the term (Kent 364). Fusing low-carbon steel with charcoal or cast iron in a graphite crucible yielded high-grade 'crucible steel.' The process named after Sir Henry Bessemer (1813-1898), the British inventor and metallurgist with over a hundred patents in his name, made steel by blasting compressed air through molten iron to burn out excess carbon and impurities; factories employing the Bessemer method featured the renowned 'converters'large, pear-shaped containers wherein the molten iron became transformed into steel. In the open-hearth process, flames which were deflected downward from the roof superheated molten iron in a bowl-shaped depression within a hearth. On striking the liquified metal, the intense gases eliminated unwanted carbon while burning out impurities. The Siemens-Martin method of open-hearth steel-making bore the names of its inventors: Karl Wilhelm Siemens (1823-1883), German-born engineer who moved to Great Britain and became Sir Charles William Siemens, and Pierre E. Martin, French engineer who lived until 1915 and made numerous improvements in the process of adding wrought iron and steel scrap to molten iron. Incidentally, Siemen's brother, Ernst Werner von Siemens, advanced the science of telegraphy and associated electrical apparati, while Sir Charles invented a regenerative steam engine and designed a steamship for laying long-distance cables. In any event, the Siemens-Martin process created a reverberatory furnace in which deflected flames heated molten iron and in which the steel could be tested, modified, and improved while still molten.
No matter what the process of obtaining high-quality iron or steel, the amount of carbon in the final product offered a means of defining each metal; that is, cast iron contained over three percent carbon, cast steel anywhere from .06 to 1.5 percent, and wrought iron from .02 to .10 percent (Kent 364). Iron and steel companies commonly graded pig iron (which is nothing more than crude iron cast in blocks) by fracture. Producers in eastern Pennsylvania identified five basic grades, shading from gray to white. Southern firms, in general, classified pig iron in ten grades, although certain plants listed even more categories. For example, iron with greater silicon content might earn special designations. Pigs varied tremendously; anthracite and coke pig iron differed markedly from charcoal foundry pig (Kent 365). By the late 1800s, scientists knew that, in cast iron, carbon exists in two forms: (a) in chemical union as 'combined' carbon (not discernible except as increasing the whiteness of the fracture in 'white iron') and (b) mixed mechanically with iron as graphite (visible, varying from gray to black and causing the fracture to be light gray to dark gray). Metallurgists noted that, when melted iron contained all the carbon it could hold, the presence of silicon would expel the excess carbon. By itself, silicon would lend strength to a casting, but, with great amounts of carbon, the silicon would serve ultimately as a weakening agent. After enough silicon would be added to produce solid castings, any extra amount (up to a certain level in ordinary practice) would cause additional graphite to form, thereby producing brittleness. Decrease of combined carbon and the consequent increase of graphite would lessen the strength. Given additional graphite, bending would increase. When no more graphite could form but silicon would continue to be added, a brittle stiffness and lighter grain would result. Again, by itself, silicon would shrink castings, but the increased graphite caused by the silicon would decrease shrinkage. The less the shrinkage, the softer the casting; the more the shrinkage, the more brittle the casting (Kent 365-69).
Designing engineers knew that, for cast parts, the chemical composition should be checked carefully; the data obtained then would be compared with the specific gravity of the metal so as to judge its character. A good foundry iron for casting should be gray, free from phosphorus, and relatively low in silicon; its density should be between 7.25 and 7.28 (Thurston 59). Metallurgists realized that large quantities of phosphorus would weaken cast iron but that, in amounts ranging from a half to one percent, phosphorus would be beneficial in preventing shrinkage. Sulfur, however, would increase shrinkagebut only when carbon and silicon levels were low. Sulfur would tend to drive out silicon and carbon and to increase shrinkage and chill while decreasing strength. Increasing the amount of silicon, however, would tend to neutralize sulfur's effects. 'Chill' meant the sudden cooling at the casting's surface producing a whitish color thereas in the case of contact between the metal and the cold surface of the mold. Such chilling would harden the surface. Only in certain instances (such as in the making of wheels) was chill deemed desirable. Silicon would defeat the tendency to chill by forming more gray graphite. Manganese beyond fifty percent would increase brittleness and cause the metallic mass to crack while cooling; beyond ninety-eight percent, the mass would crumble. Manganese also would augment shrinkage. An increase of only one percent would raise the shrinkage twenty-five percent and increase the hardness forty percent (Kent 365-69).
Thus generally, elements which would increase tenacity would decrease ductility and resilience. Up to a point, phosphorus would add strength, as would manganese; beyond a certain level, they would weaken or make brittle the casting. Silicon and carbon also would tend to harden ironbut only in moderation. Occasionally, ductility would decrease more quickly than corresponding increases in tenacity!
Builders of boilers and engines found that the tenacity of cast iron varied from ten thousand pounds per square inch (psi) to more than fifty thousand psi (Thurston 60). Additionally, they learned that silicon levels differed considerably from lot to lot and within even a single bed of a cast of pig iron. Fracture was seen to deviate in size of grain from bed to bed, even when chemical analyses showed the beds to be the same (Kent 369-70). Much had to be done to strive for uniformity and to determine acceptable standards.
The quality of wrought iron resisted exact determination even more than did that of cast iron. So-called 'malleable cast iron' was a crude wrought iron which had been subjected to a process of decarbonization; its tensile strength was not great, and it was used to make only such items as handles, wheels, and pinions (Kent 375). Milling pressures, temperature variations, and mechanical handling so greatly modified other forms of wrought iron that designing engineers had to go to much trouble to put it through tests before judging it safe to use. Some wrought irons boasted a tenacity of forty thousand psi, others eighty thousand (these latter more accurately termed 'puddled steels'), and certain wrought irons resembled cast iron in being hard and brittle (Thurston 59-60, 62). In short, while steel could be analyzed chemically and much be known with certainty, wrought iron could be analyzed with considerable care and yet nothing known with any real assurance (Kent 377). Still, wrought iron was preferred over cast iron for most uses (Thurston 127). Although cast iron would remain durable even in the presence of corroding elements (acids do not decay it rapidly), benefited from being compact, was immune to dangerous lamination, conducted equally well throughout, and could be produced at low cost, it would crack from shrinkage strains and local variations of temperature and was 'always treacherous and unreliable' (Thurston 62).
Designing engineers, then, had to test metals physically and chemically, observe the openness or closeness of the grain, evaluate the shade of color (particularly of the fracture), try to determine the depth of chill, and become so skilled as to be nearly intuitive while checking and judging properties (Thurston 61). By 1900, mechanical engineers were welcoming the use of more and more 'steels' in boiler making. Such steels were called 'low,' 'soft,' or 'mild' and consisted of 'ingot iron,' in fact. Their characteristics included homogeneity, tenacity, and ductility. Their resilience set them apart from 'true steels,' which were much harder. The mild steels enjoyed tenacity greater than that of common iron and could be rolled into thin or thick plates, thereby making them desirable for strong, durable boilers (Thurston 63). Builders continued to scrutinize the quality of such steels:
In all rolled steel [as in Bessemer plates] the quality depends on the size of the bloom or ingot from which it is rolled, the work put on it, and the temperature at which it is finished, as well as the chemical composition. (Kent 390)
Soft steel ingots for boiler plates had only ten thousand up to twenty thousand psi tensile strength until they were rolled, when they then gave 55,000 to 65,000 psi (Kent 392-93). Carbon in good steel boiler plate should be less than one-fourth of one percent; in fact, it often was less than one-tenth of one percent. Designing engineers would reject boiler plate which, after being heated red-hot and suddenly cooled, was found to have hardened perceptibly; despite variations of temperature, good steel should not take a temper (Thurston 64). One steel-manufacturing firm in Pittsburgh provided six grades of steel: Extra Fire-box (0.02 phosphorus), Fire-box (0.03 phosphorus), Extra Flange (0.04 phosphorus), Flange (0.60 phosphorus), Shell (0.80 phosphorus), and Tank (1.00 phosphorus). For firebox plates, which needed the best steel, builders specified the following ranges of tolerance: carbon (0.15- 0.25), phosphorus (not to exceed 0.03), manganese (0.40 preferred and not to exceed 0.55), silicon (0.02 preferred and not to exceed 0.04), and sulfur (not to exceed 0.05) (Kent 399).
In the early 1800s, good boiler plate was made of 'charcoal iron,' produced from pig iron in a charcoal blast-furnace with no fuel other than pure wood charcoal being used in the presence of the metal. The growing scarcity of such charcoal caused these irons to become expensive. Costly or not, firebox iron had to be made of the best platestrong enough to wear well and ductile enough to be flanged. Under no circumstances could firebox iron be brittle. By the mid-1800s, much iron boiler plate came from puddled or scrap iron. On the one hand, wrought iron was formed into piles, which were heated to a full welding temperature, then were passed through a heavy roll-train. With the burning coal not in the presence of the iron during the heating phase, sulfur and phosphorus would not be absorbed into the metal. Wrought-iron plate retained some slag which came from the puddle-ball; therefore, surfaces coated with oxide became flattened inside the plate and led to dangerous lamination. On the other hand, steel plate often derived from the Siemens-Martin process, which stirred the molten metal to prevent cinders and which (like the Bessemer process) introduced ferro-manganese to help eliminate unwanted oxides and to carbonize the metal correctly. When somewhat cooled, the ingot then was taken to be rolled. It first was reheated to a uniform temperature then foiled. Steel boiler plates frequently were made twenty feet long or longer and ten feet wide (Thurston 95-97). The engineers who designed boilers and engines had to know all iron-and steel-making processes and know them well!
Such engineers also knew that, of all stressful phenomena, temperature would have the greatest effect on iron and steel under load, and their designs treated temperature variations with great deference (Thurston 83). Even steam pressures, as powerful as they could become, posed less of a threat to the metal of a boiler than the rapid elongations and contractions brought about by temperature (Forney 188). A rise in temperature of only one degree would cause a bar of ordinary boiler iron to elongate 'by the same amount as would a tensile stress of the intensity of about 190 lbs. per square inch'; thus, if a bar were held rigidly at each end and not allowed to move while heated a. mere ten degrees, it 'would be subjected to a force of compression equal to 1900 lbs. per square inch' (Holmes 399). Such wildly fluctuating effects explain why the long bar-stays running the length of a locomotive-style boiler in a space hotter than the outside shell had a tendency to droop and, according to one authority, were 'perfectly useless' (Holmes 397). Manufacturers and those who ran engines respected the menace of rapid changes in temperature. Bill Lamb of Lexington, Kentucky, remembers that railroad engineers fired boilers as slowly as possible taking many hours to do so. Once heated, the firebox was kept at as constant a temperature as could be managed through banking the fire when the locomotive was idle.
In terms of allowable pressures, designing engineers expected the tensile strength of wrought-iron plates to be approximately fifty thousand psi and of mild-steel plates to be around sixty thousand psi. Steel plates should not be less than fifty thousand and not more than seventy thousand psi; if more than seventy thousand, they would be brittle, according to tum-of-the-century texts on mechanical engineering (Forney 188-89, for example).
A review of some of the catalogues which engine manufacturers printed to advertise equipment for agricultural use reveals the evolving story of iron and steel. In 1891, the Harrison Machine Works of Belleville, Illinois, claimed, 'The very best Tennessee iron is in the boilers. The castings are made from selected pig iron.' The Harrison Jumbo boiler's outside shell was 'the best' Charcoal Hammered Number 1 iron of fifty thousand psi tensile strength, while the firebox and tubes heets were 'the best' flange iron of 65,000 psi. Clearly, Harrison used wrought-iron plates for the boiler proper. Incidentally, Charcoal Hammered iron was not always hammered as advertised. Also, it made acceptable shells but was not good for flanging. In 1893, C. Aultman & Company of Canton, Ohio, were building their own boilers of mild steel with a tensile strength of sixty thousand psi. By 1895, the Buffalo-Pitts Company of Buffalo, New York, used shell steel of sixty thousand psi for the boiler cylinder (9/32 of an inch thick) and for the furnace casing (5/16 of an inch thick). They put flanged firebox steel into their firebox (9/32 of an inch thick) and into their tubesheets (3/8 of an inch thick). The Buffalo-Pitts catalogue of 1910 called attention to the fact that their boilers were 'the best' open-hearth firebox and flange steel and could safely carry 150 psi working pressure.
In 1903, the Northwest Thresher Company of Stillwater, Minnesota, were using steel of sixty thousand psi tensile strength. In their return-flue engines of eighteen horsepower or smaller, the main flue was 5/16 of an inch thick, the shell of an inch, and the tubesheet 3/8. In the 20 horsepower and larger engines, the main flue was 3/8 of an inch thick, the shell 5/16 of an inch, and the tubesheet 7/16. Interestingly, in their twenty-five horsepower simple and thirty horsepower compound engines, the main flue had three sections, the central portion 13/32 of an inch thick, the outer ends 3/8 of an inch.
Robinson & Company of Richmond, Indiana, were still building their own hand-riveted boilers of extra-heavy marine steel in 1906:
It is the only hand riveted boiler we know of today, and it grades with other hand made articles, being considered 20 per cent more durable, and costing, as we well know, 20 per cent more to construct than machine riveted. And the workmanship on it throughout is of the same careful, thorough-going sort which places this boiler in a class by itself. . . .(4)
Likewise, the J. I. Case Company of Racine, Wisconsin, were building their own boilers in 1907; theirs were of open-hearth flange steel of sixty thousand psi tensile strength. The steel used in Case boilers had to pass rigid chemical and physical tests in Case's own laboratories. In 1909, their catalogue noted that the seams of Case boilers were hydraulic riveted.
Of all the catalogues of companies manufacturing agricultural steam engines, the Port Huron Company's 1908 advertising brochure lavished the most pages on describing the construction of their boilers, which could carry a working pressure up to 175 psi. The shell, wagon top, and dome were of flange steel, while the firebox casing, the crownsheet, the throatsheet, and the tubesheets were of more costly firebox steel. The Port Huron, Michigan, company conducted their own inspection at the steel mill because steel manufacturers might stamp metal with a grade which the chemical and/or physical properties of that metal would not deserve. Port Huron established a narrow leeway for tensile-strength specifications: 55,000 up to 63,000 psi. In its tests, the firm rejected plates the elastic limit of which was reached before half of the ultimate strength had been applied. Port Huron's catalogue mentioned that 'red short' steel, which might crack while being worked at red heat, often could be blamed on excessive sulfur; therefore, Port Huron allowed only 4/100 of one percent of sulfur in flange steel and 3/100 of one percent in firebox steel. Port Huron stated:
That our sulphur (sic) specifications are very rigid is evidenced by the fact that one of the largest plate mills in the country could not furnish our 'Fire Box' grade, and the mill from which we regularly obtain our plates had at three different times asked us to modify our specifications as to sulphur, which we have steadfastly declined to do.(39)
Port Huron's fourteen horsepower single simple engine and sixteen horsepower single compound engine had boilers with shells 5/16 of an inch thick and with tubesheets of an inch thick; the company's larger engines had boilers with a shell 3/8 of an inch thick. In catalogues of later years, Port Huron called attention to the fact that, while the company's larger engines had boiler dome-shells 5/16 of an inch thick, the tubesheets and throatsheet measured half an inch in thickness. Given the safety factor which Port Huron employed in designing boilers, actual strength of the boiler was 'five times as great' as the allowable pressure indicated.
In 1910, the Geiser Manufacturing Company of Waynesboro, Pennsylvania, specified a tensile strength for their boiler plate between 55,000 and 65,000 psi. Their tubesheets were half an inch thick, their shells between and 3/8 of an inch, depending on the size of the engine. At the same time, the Reeves Company of Columbus, Indiana, offered boilers made of flange-steel plate tested to sixty thousand psi tensile strength. The Reeves catalogue announced, 'Not a pound of shell-steel plate is used in these boilers' (4).
In 1912, the Frick Company of Waynesboro, Pennsylvania, used open-hearth homogeneous flange steel of between 55,000 and sixty thousand psi tensile strength for their boilers. In 1913 and 1914, the Nichols & Shepard Company of Battle Creek, Michigan, featured boilers having good steel plate 7/16 of an inch thick in the wagon top of their small engines and half an inch thick in their twenty-five and thirty horsepower engines; such thickness in the wagon top ensured strength where all of the principal brackets attached. The tube sheets on Nichols & Shepard engines measured half an inch thick.
Also in 1913, the Huber Manufacturing Company's return-flue boilers had an outer shell between and 3/8 of an inch thick, according to the size of the engine. By 1914, the Marion, Ohio, company's marine-steel shells measured 5/16 of an inch thick in their smallest engine (sixteen horsepower) and \ in their largest engine (thirty horsepower). The main fire flue thickness, meanwhile, ranged from 3/8 of an inch to 7/16, and the heads ranged from 3/8 to a full half inch.
At about this time (1914), most engine-manufacturing companies were using open-hearth steel of sixty thousand psi tensile strength and hydraulic double-riveted butt straps. The Waterloo Manufacturing Company of Waterloo, Ontario, however, used a triple-riveted seam on the boilers of their sixteen and eighteen horsepower engines. The 1916 Waterloo catalogue pointed out that their boilers were constructed in their own factory. Adhering conscientiously to the principle of truth in advertising, the A. D. Baker Company of Swanton, Ohio, stated that their boiler plate had been tested to 55,000 psi tensile strength (not the customary sixty thousand).
The history of boiler manufacturing from the late 1800s to the early 1900s shows a trend away from wrought iron and toward better and better steel. No matter how excellent the boiler plate, however, the boiler was only as strong as its weakest rivet! Designing engineers knew that iron or steel rivets had to be especially good quality: strong, ductile, and able to withstand severe deformation 'at all temperatures.' Rivets under a diameter of 3/8 inch commonly were 'worked cold, and this is the most trying test of quality possible' (Thurston 114). The tenacity of good rivets necessarily equaled that of good steel plate, sixty thousand psi. The head of a good rivet would not harden or become brittle beneath the blows of a hammer; a good rivet of 5/8 inch diameter could be 'doubled up and hammered together, cold, without exhibiting a trace of fracture.' Split in two, a good iron rivet would show close, smoothly-curved grain with no slag, a good steel rivet absolute uniformity with no trace of grain. Steel rivets were so pure as to take a mirror surface when polished (Thurston 114-15).
Rivets usually had a straight or slightly-tapered body with a head 1.5 or 1.6 times the diameter of the shank. The shank was two to four percent smaller than the hole it would fill. The thickness of the head measured between seventy and seventy-five percent of the diameter of the body (or shank) of the rivet. The length of the shank was two and a quarter or two and a half times the diameter of the hole (Kent says 1.8 times for double-riveted butt joints), and the hole's diameter often equaled the double thickness of plates held together by the rivet. Hand-hammers and riveting machines drove the small end down to form a cone-shaped or hemispherical head (Thurston 115).The Mechanical Engineers' Pocket-Book gave designing engineers detailed descriptions of a variety of rivets useful in all manner of situations.
Thurston told mechanical engineering students that good hand-riveting surpassed bad steam-riveting. He also warned against careless handling of rivets which could create a fin on the shank in between the sheets being-riveted. Such a fin would hold sheets apart and cause leaks. To prevent the formation of fins, builders should carefully ascertain that plates were well adjusted and in strong contact before heading up the rivets, according to Thurston, who preferred that, in hand-hammering, a cup-shaped die be used. He claimed that the hemispherical head produced by the cupped die presented a form stronger than that created by a cone-shaped die. He recommended that workers learn how to hit hand-hammered rivets with the fewest number of blows possible. Given a choice between steam-riveting and hydraulic-riveting, Thurston would select the latter because its slow application of pressure would do less damage to the plate than steam-riveting would do. He feared that, unless carefully monitored, machine-riveting of any kind was more likely to cause defective rivets than hand-riveting would cause (125-26). Kent cautioned mechanical engineers that hand-riveted joints were liable to present the serious defect of 'visible slip,' which could begin at a comparatively small load and render the joint apt to fail under sudden strain (355). Regardless of the method of riveting, rivets generally confronted shearing stresses (Thurston 115). In iron, the metal resisted shearing differently according to the plane in which the stress was acting. Shearing strength was 4/5 of the tenacity, if the shear were in a plane perpendicular to the direction in which the iron plate was rolled and if the tension were applied in a plane parallel to the direction of rolling. Kent added:
The shearing resistance in a plane parallel to the direction of rolling is different from that in a plane perpendicular to that direction, and again differs according as the plane of shear is perpendicular or parallel to the breadth of the bar. In the former case the resistance is 18 to 20% greater than in a plane perpendicular to the fibres, or is equal to the tenacity. In the latter case it is only half as great as in a plane perpendicular to the fibres.(363)
The shearing resistance of iron rivets likewise varied from full tenacity, through eighty percent tenacity, to merely fifty percent tensile strength!
In the early years of boiler manufacturing, only a single row of rivets joined a lap seam. Eventually, builders saw the wisdom in using a double row of rivets, and, before long, triple-riveted lap seams made their appearance. Last came the butt-strap joints which, in their most perfected form, consisted of several rows of rivets joining straps inside and outside the boiler where the two edges of the boiler plate met but did not overlap. Forney believed that well-constructed lap joints were as secure as any butt-strap joints (208), and Kent reported results of tests wherein the efficiency of lap joints was, for all intents and purposes, the same as that of butt-strap joints (359-60). The 1913 Case catalogue featured a double-riveted lap seam on sixty horsepower (and smaller) engines, a triple-riveted lap on seventy-five and eighty horsepower engines, and a double butt-strap on the 110 horsepower engine.
In writing about single-riveted joints, Thurston concluded that they have, in general, about sixty percent the strength of a solid sheet, whereas double-riveted lap joints enjoy about seventy percent the original tenacity of the plate before riveting. If properly distributed, additional rows of rivets would bring the joint even closer to the tensile strength of the boiler plate itself, Thurston stated (117). The Power Catechism instructed engineers that a triple-riveted lap seam contains the maximum number of rows possible because adding another row of rivets would bring about further widening of the space between rivets (the 'pitch'); consequently, the joint would be liable to leak. By increasing the pitch, a designer would be lessening the number of rivets, thereby increasing the stress on each; rivets could begin to shear. If a mechanical engineer were to stick to three rows of rivets but then try to add strength by increasing the number of rivets per row, that engineer would reduce the section of plate between the rivets and would increase the chances that the plate would pull apart (41-42).
According to Thurston, the placement of rivet-rows in the sheet should balance the potential shearing of rivets against the potential breaking away of the edge of the sheet or a tearing along the lap (117). In such lap joints, the plate could crack between rivet holes, could split from the rivet toward the edge of the sheet, could crack on a line parallel to the rivet row on the plate just inside either overlapping edge (especially where the plate would be scored from caulking), or could become crushed from the rivet toward the sheet's edge because the pressure inside would be attempting to 'unroll' the cylindrical shape of the boiler (Forney 193; Port Huron catalogue 7). Barring weakness anywhere in the plate itself, a designing engineer could assemble a host of careful calculations so as to build a boiler the rivets of which, under excessive pressure, would pop out before the boiler would explode, thereby creating miniature 'safety valves' to release steam. Among the 'rules of thumb' which mechanical engineers followed were the facts that a lap usually extended beyond the rivet holes 1.5 times the rivets' diameter (Thurston 117) and that the pitch in the butt-joint ordinarily amounted to 4.1 times the diameter of the rivet hole (Kent 358).
Punching out the holes for rivets increased the strength of a sheet made of extremely soft and ductile metal because the 'flow' of the metal outward from the hole stiffened the surrounding area; on the other hand, punching injured hard iron or steel. Thurston recommended drilling holes in steelwhich indeed was the practice of most builders. A few, however, punched a small hole in steel then reamed or counter bored it to the correct diameter. Thurston added that the section of solid material left between rivet holes in a single row of rivets in either hard iron or steel plates (having excessive levels of manganese or phosphorus) lost tensile strength most in lap joints with punched holes, as well as in sheets riveted by steam rather than by hydraulic pressure. The strength of rivets in such plates, however, was greater against shearing. A zig-zag pattern in a double row of rivets was best for improving the tensile resistance of the material between all of the rivets in a lap joint, but a double-strapped butt joint prevented the bending of the main plate and nearly doubled the shearing strength of each rivet (123-24). Thurston suggested that steel be annealed after each operation to restore lost ductility (64). Forney, though, believed that annealing steel plates after punching could be mishandled and might lead to more harm than good (192).
Considering the effects of rivets and riveting, designing engineers came to recognize that the real factor of safety of well-made boilers should not be less than six and ought to be eight or ten!
In the most general sense, a 'false' factor of safety of six is reduced to 3.36 (0.56 x 6 = 3.36) for single-riveted lap seams or to 4.2 (0.70 x 6 = 4.2) for double-riveted lap seams (Thurston 122-23). Mechanical engineers also had to consider the molecular and chemical changes occurring in metal over long periods of time:
The first change to be referred to is that gradual and imperceptible one which, occupying months and years, and under the ordinary influence of the weather going on slowly but surely, results finally in important modifications of the proportions of the chemical elements present, and in a consequent equally considerable change of the mechanical properties of the metal.(Thurston 81).
Thurston explained how especially vulnerable iron is to oxidation:
A loss of strength and of ductility is very often observed in the iron of which boilers are composed, as they advance in age, due to the progress of oxidation. . . .within and between the laminae of which the sheets may be composed. The plate may be thus very nearly destroyed, at times, before this action may be detected. In some cases the iron may be nearly all destroyed, and only a sheet of oxide may remain; while the boiler, if not working under high pressure, may still appear sound. Such deterioration is often a source of great danger.(149)
Good cast iron could last for at least a century submerged in salt water but would become soft enough to cut with a knife. Exposed to the air, iron was found to rust and to deteriorate more quickly than steel. Ironically, the smoke from the 'iron horse,' or railroad locomotive, rapidly ruined the iron in railway bridges because the combustion of sulfur in coal acted as a corrosive agent and became even more deleterious when bridge iron became coated with soot (Kent 385-86).
Other changes occurring over time in iron and steel resulted from stress. Textbooks for engineering students defined 'fatigue' as the weakening of metal by repeated stresses. Giving the metal an opportunity to rest would relieve the effects of certain internal stresses, but rest (or, for that matter, annealing) only occasionally relieved actual fatigue. If not excessively weakened, steel razors were known to recover keenness after a time of rest. Thurston stated: 'Boiler-makers frequently search old boilers carefully, when reopened for repairs after a long period of service, to find any tools that have been lost and so improved' (81). Likewise, springs were found to regain tension after resting, and scythes exposed to the weather for a long season apparently were restored. Iron and steel could best withstand stationary loading; to a lesser degree, they could endure tension alternating with relief; but least of all could they recover from alternate and equal tensions and compressions (Thurston 80). The fluctuating temperatures of boilers and the pounding rhythms of steam engines gave iron and steel their ultimate tests.
Gone are the days when designing engineers, armed with pencils and slide rules, brought complex bodies of knowledge to bear on how to improve boilers and steam-powered mechanisms. Vanished are most of the factories which, occupying entire city blocks, focused a growing country's attention on the progress of agricultural industry from iron ore to engines. Today, one strolls around the remaining brick edifices of an old-time manufacturing giant and wonders what became of its might and grandeur. If one stands still long enough, however, one's imaginative mind can begin to hear the hammering of rivets, the pressing of iron sheets between dies to compel them to take the shapes of dometops, handhole plates, and tubesheets, and the thud of the 'tup' or hammer dropping six feet to knock out small parts. By closing one's eyes, one can picture the newly finished traction engines under their own steam chuffing forth beneath an azure sky.
Engine-company catalogues (reprints) too numerous to list
separately Forney, Matthias N. Catechism of the Locomotive. New
York: Railroad Gazette, 1897.
Holmes, George C. V. The Steam Engine. New York: Longmans, Green, 1900.
Kent, William. The Mechanical Engineers' Pocket-Book. New York: Wiley, 1895.
Lamb, William Morris. Telephone conversation Nov. 20, 1994.
The Power Catechism. New York: McGraw, 1897.
Thurston, Robert H. A Manual of Steam-Boilers: Their Design, Construction, and Operation. New York: Wiley, 1904.