The Iron Age 1922-09-28: Vol 110 Iss 13

THE IRON ACE New York, September 28, 1922 ESTABLISHED 1855 ¥: Be VOL. 110, No. 13 Recent Developments in Fatigue of Metals Effects of Heat Treatment—Testing Machine Now Used— Endurance Limit Defined—Machine Design and Localized Stress BY H. F. MOORE* AND T. M. JASPER+ HE failure of metal parts of machines after thousands or millions of repetitions of a load which, for a time, apparently does_no damage is a phenomenon which has become of prime importance to metallurgists, machine designers, and materials en- gineers. The term “fatigue” has been applied to such a failure, and the testing of metals for resis- tance to “fatigue” has come to be an in the testing la- boratéry. The historical develo pment of testing for “fatigue” of met- als has been so often summarized that no such sum- mary will be at- tempted in this ar- cle, exe pt to call attention to the lemolishing of the theory that, under Fi Tepeated st ress, meta “erystal- zed,” and the demonstration, by means of the metallo- graphic microscope, that fatigue failure consists in the progressive spread of minute cracks across a member sudjected to repeated stress. During the war the subject of resistance to fatigue very great importance and, as a result of ntion called to the problem, there are in prog- imber of important investigations of various this problem. In this country important ns are in progress at the following labora- » and doubtless at other laboratories as well. became of of Illinois, investigation under the of National Research Council, Engi- & Foundation, General Electric Co., Engi- ng Experiment Station of the University of il Engineering Experiment Station, An- Ma of Ting rofessor of engineering materials, University irge of investigation of the fatigue of metals. male “sistant professor of engineering materials, fatigue «3 ‘‘inols, engineer of tests, investigation of the ig. 1.—Section of the Latest Type of Testing Machine for Making Reversed Bending Tests as Used at the University of Illinois 779 Cornell University Ithaca, N Y.. investigation under the auspices of the U. S. Bureau of Mines. Air Service Laboratories, McCook Field, Dayton, Ohio. Aluminum Co. of America. Although the above laboratories are all getting data of great value the opportunities for profitable research in fatigue of metals are very far from being exhausted. This particular field is one of prime im- portance and it would seem worthy the attention of any parties inter- ested in engineer- ing research. Within the lim- its of this article it will not be at- tempted to give de- scriptions of spe- cial testing ma- chines for making fatigue tests of metals*. It may be noted that most test results have been obtained by the use of reversed-bending tests of specimens. Probably there are many more machine members sub- jected to cycles of flexural stress than there are ma- chine members subjected to cycles of direct axial stress. Latest Type of Testing Machine Used. Fig. 1 shows a diagram of the latest form of testing machine for making reversed bending tests, as used in the laboratories of the University of Illinois. In this machine the specimen § is held by one end in the vise V and the other end is rotated in a small circle being held in the bearing B. Sidewise pressure is brought on the bearing B by a calibrated indicator spring I and this pressure can be adjusted by means of a screw. The compression of the spring, and hence the load on the specimen is measured by means of a strain gage span- *For descriptions of such machines see Unwin “The Test- ing of Materials of Construction,” Chap. XVI.; B. Parker Haigh, “Alternating Stress Endurance of Steel,” Tue Iron AGE, May 11, 1916: Moore, Kommers and Jasper, “Fatigue or Progressive Failure of Metals under Repeated Stress.” 1922 Proceedings of the American Society for Testing Materials. Wy oe ee ms » fies iat ae oe a: 498 nr ee reels See: seer Te ad ore ee ht gets oe te = = 5 ey oad wie ie iG ‘* : } hs in: am ho wbabetry Ns “itnp ose - m rb ee ee ee area tear us S. Lo oe ee io! ness pale. NE i roe + ié io a iy A Thy ae 1 aaa a sh ied | * Se ita oi a Tue Seno hee ote Sel eee wa eae OEE , +93 ~ en er x S pie oe ao obud ree Ct i o- “6 =f Seer emanee * “aes ae a re¢ ' as 3 ey ti a» ; yo ’ 7 ’ : . BP ab ats. ah : i . ry | oh . + ar 5 be tis ie. x : yee tr« “ eee Bis Me eit . f 5 : iis mt rf, i. ™ 4, ee P ils, ey}: af ~ "' ‘ 7 . 780 THE IRON AGE Fig. 2.—Typical Diagrams of Results September 28, 1999 t-4I- r + nt | paces | | | tat | o> Spec/men aid not Rr] \R- Endurance Linn# re ae —p> 7Y 4 cL../ > Carbon Stee/ be» | Dv | & | | 3] | S4—— rf —-? = | OT¥ & N x S| er R 8 spo Ss f 6 cyT Te WO | t » i iV iA TAO ] V0 V 5 Read ngs far Ric o of Termeraty re under Reversed Bend Fonlure 6 of Repeated Stress Tests. Diagram (A) drawn to ordinary co-ordin Diagram (B) gives same test results as diagram (A), but is drawn to logarithmic co-ordinates ning the gage holes GG shown near the ends of the spring. The rotating spring is carried in the cross- head C and sidewise motion of the bearing B is pre- vented by placing the bearing in a slot. Excessive dis- placement of the bearing when the specimen breaks is prevented by the rod R. The crosshead is driven by a shaft H, a pulley P and a motor not shown in the figure. The number of revolutions of the cross- head is measured by the revolution counter K which is driven by a worm on the drive shaft H. When a test to destruction is carried out the frac- ture of the specimen occurs at the necked-down part N, and the broken end of the specimen hits a screw and kicks out the lever L. This releases the spring W, which then opens the motor switch D, thus stopping the machines. This machine can be used for tests in which the rise of temperature of the specimen is measured. For this purpose fiber strips are placed between the head- stock T and the base A, between the vise V and the base A, and a fiber bushing is placed between the end of the specimen and its bearing. <A thermocouple (copper-constantan) is held against the specimen at the necked-down portion, and an opposed thermocouple is placed in an insulating bushing against the vise end of the speci- C men. From these couples wires lead to a delicate galvanometer. The fiber bushings serve to mini- mize the transfer of heat from the rotating parts of the machine. The machine produces cycles of completely reversed bending stress, and can be run at a speed of 1800 r.p.m. 5; In studying the results of re- : © peated-stress tests it is usually . ; convenient to usea graphin which £ f& § values of fiber stress computed by a the ordinary formulas of me- > 200 £40000 chanics of materials are plotted as ordinates and the numbers of cycles of stress necessary to cause failure ( are plotted as abscissae. Such a diagram has been given the name S—WN diagram, Fig. 2 shows S—-N Fig. 3.—Effect of ness, an diagrams for several typical series of tests of wrought ferrous metals in reversed bending. In Fig. 2 (a) the diagrams are plotted to ordinary coérdinates; in Fig. 2 (b) the diagrams are plotted to semi-logarithmi coordinates. In Fig. 2 it will be noted that nearly all specimens which failed during the test failed be- low 10,000,000 cycles of stress. There seems to be a fairly well defined stress below which no failuves oc- cur, and at this stress the S—N diagram has a fairl) well defined “break” and becomes horizontal, or ver) nearly so. That is, at this stress there seems to b a rapid change in the relation of fiber stress to numbe: of cycles for rupture, and below this stress no failur occurred under 100,000,000 cycles of stress (one spe men run at a stress corresponding to the horizonta line has withstood 850,000,000 cycles of stress withou rupture and another to 650,000,000 cycles.) It can hardly be said dogmatically that this limit ing stress is one which can be withstood for an infiniti length of time because of the impossibility of test ing to that point. Items of independent evidence however point to this endurance limit as the limiting value below which a ferrous metal will not fail, an¢ these may be stated as follows: 4 0 200 40 600 800 1000 200 Drawing Temperature, deg. fahr Varying Temperature of Draw on Tensile Stre! ad Endurance Limit of 1.20 Per Cent Carbon Stee! Sept mber 28, 1922 THE IRON AGE st point of evidence is contained in the £110,000 he stress-cycle curve at about 1,000,000 3 cas kk 8 * ' nd is the horizontal nature of the curve 2 ih f sreak has been passed which, as far as P ; tested, goes out beyond 850,000,000 5 80,00 7 pecimens which had run 100,000,000 E 70,000}-> = 3 = b .e endurance limit without failure were Be ee . . e into the machine at progressively o . aay .° ° | - ‘ sses for a further 50,000,000 cycles per 5 s0000be e . 3% ventually a stress of 20 per cent above 5 i ee 11] endurance limit in some instances = 40.000 e Di | without failure. This constitutes a E i. ° | yf evidence for the existence of a lim- ‘y 59,000r¢@ - s for wrought ferrous metals below = = ire will not occur under repeated stress. Sees T . ae = ens run at stress just below this limit- 2 .....| | sia cei eet Nee Be ee ee ictually seem to be strengthened by the 0 tress. Possibly this strengthening is o : 5 20 25 30 35 40 45 EO 55 60 5 work during the progress of the test. Charpy Single -blow Impact-Bending Test, ft-Ib. jurance limit therefore has been defined ht ferrous metals as the stress below tal will withstand an indefinitely large ycles without failure and is found by locat- rizontal part of the stress-cycle curve. lemperature Changes Under Repeated Stress nection with tests to failure under repeated udy of temperature changes under repeated > oO —) oO > Oo 2 Oo oS oO oO oO 2 o oO oO oO > . oOo Oo oS oS oO oO - vv oO @ oS ~ <t Proportional Limit, Tension Test, Ib persq. in relation of Endurance Limit with Proportional Limit lata of interest for wrought ferrous met- specimen is subjected to, say, a thousand stress below the endurance limit there ght rise of temperature measurable with ermocouple and galvanometer,—only a few fa degree. If, however, the specimen is a thousand cycles of stress above the mit it becomes “feverish” and shows a of temperature. Fig. 2 (c) shows a dia- results plotted with fiber stresses as ordi- se of temperature after 1000 cycles of nding stress as abscissas. The specimens the same bars of material as those from ut the specimens whose test results are g. 2 (a) and Fig. 2 (b) and the distinct the temperature curve at the endurance nly shown. perature test was first used by Stromeyer ter, England. Recently Gough of the nal Physical Laboratory has made some seem to give some promise that the de- a specimen under repeated cycles of re- ; a vending stress may also be used as a quick- ‘or of the endurance limit. This rise of rs Fig. 4 Correlation of Endurance Limit with Results of Charpy Single-Blow Impact-Bending Tests temperature is a fourth item of evidence of some dis- tinct development in internal progressive structural breakdown at the endurance limit. All the foregoing discussion is applicable to wrought ferrous metals. Data are very few for non-ferrous metals, and almost entirely lacking for cast iron and steel castings. It is by no means certain that for non-ferrous metals there is a well-defined endurance limit, or that if such a limit exist, there is a well- marked development of heat under cycles of stress at that limit. The two fields of non-ferrous metals and of steel castings offer a wide opportunity for investi- gation. Reversal of Axial Stress As noted above most of the test results of repeated- stress tests have been obtained from tests in reversed bending. Some tests have been made on the resistance of steel to reversals of axial stress—a cycle of stress from tension to compression. sritish results seem to show that for such tests the endurance limit is slight- ly lower than for reversed-bending tests, and results at Illinois seem to show that the endurance limit is mark- edly lower than for reversed-bending tests. The Brit- ish tests were mostly on normalized steel, while the results at Illinois were on sorbitic steel and on troos- titic steel as well, and for the harder steels the dis- crepancy between tension-compression tests and re- versed-bending tests was most marked. Three explanations for the discrepancy are sug- gested :-— 1.—Initial stresses, especially in heat treated steel: oS So So oS aS oO eo S S o o ° oO S oOo © oOo So Oo Oo oOo oS oO Co Oo aQ “~ +z o co OQ ~ - lension Test, Ib per $a in 0 oO ~+ eld ; Fig. 6.—Correlation of Endurance Limit with Yield Point = p re * =~ i Su teapee ee tt 7 ~ oth, ae cer nied hn os. Sormedyy aa cement wubb ata +5, ee om vie % tak a ee a oe Pi aioe ty os ~ noe B ¥ Pet ri 7. Les te Moe 782 as the outside surface of a speci- 100 000 men cools first the inside is in ten- sion after final cooling. Hence in a bending test the outside is in 580 090 initial compression and this initial compression must be overcome be- fore the crystalline grains tear apart; in a_ tension-compression test the inside is in initial tension, and the tension applied during the test would be added to this initial tension. persq.i 60 000 40 000 20 000;— THE IRON AGE 2.—Tension-compression speci- mens (or parts) are never given an exactly axial load, there is al- ways some eccentricity, and this means a higher maximum fiber stress than the computed tensile compressive stress. 3.—In a_ tension-compression. specimen (or machine part) the whole area of cross-section is subjected to a maximum stress rather than the outside fibers only, as is the case with a flexural specimen; hence there would be more opportunity for the development of a crack into a fracture when all fibers were subjected to approximately the same stress. Tests are in progress bearing on these three points, and while results so far obtained are tentative, the general indications seem to show that initial stress is an appreciable factor, especially for heat-treated steel; but that it can not explain all the observed discrep- ancies, that, since two machines of quite different de- sign and action but each applying axial load to the specimen give fairly concordant results, the effect of eccentric loading can hardly be large, and that it seems reasonable to expect that, for specimens or ma- chine parts in which a whole cross-section is sub- jected to maximum stress, the endurance limit will be somewhat lower than for specimens or machine parts subjected to repeated flexure. For rotating-beam tests the endurance limit.is found at about one-half the static ultimate; for tension-compression tests of nor- malized or annealed steels the endurance limit is found at about one-third the static ultimate (this value should be regarded as tentative); and for heat-treated steels the results of alternate tension-compression tests give still lower values for endurance limit, for very hard steels, much lower. Endurance Limit, Rotating Beam Machine, Ib Other Factors of Importance In addition to the type of loading which has an effect on the strength of a metal in fatigue, the shape of a member under load and the surface finish are factors of importance. Poor surface finish may di- minish the endurance limit by about 18 per cent. A sharp re-entrant angle may increase the liability of failure of a member 100 per cent, depending on the © oO ~ Brinell Hardness Number Fig. 8.—Correlation of Endurance Limit with Brinell Hardness 160 000-- sw Tensile Strength, Ib. per sq. in. Fig. 7.—Correlation of Endurance Limit with Tensile Strene location of the sharp corner with reference distribution. Experiments recently made to determin: fect of an occasional heavy overload bring out the following facts: The endurance limit for complete r versal of stress is consistently below the static point for nearly all the materials tested. If the mem- ber is overloaded, say, 20 times to just below the yield point, the endurance limit is decreased by about 6 pe cent only. If it is overloaded above the yield point to about 70 per cent above the original endurance limit the resulting endurance limit is lowered about 23 pe cent. The test method of overloading consists in placing the specimen in a universal testing machine and ap- plying an axial tensile load 20 times and then imme- diately placing the specimen in a rotating-beam fa- tigue testing machine. The endurance limit for speci- mens which are introduced into a bath of boiling water immediately after overloading and allowed to boil for about 1 hr. and then cooled before introducing the specimen into the fatigue machine does not seem much higher than the endurance limit for specimens placed directly in a fatigue testing machine after overloading A complete series designed to give the specimen a pe- riod of three months’ rest have been laid down so as to study the effect of rest after overstrain. the ef- yield Effect of Drawing After Quenching Results of tests of specimens of carbon steel, give! different “draws” after quenching, indicate that, in general, a “draw” up to about 600 deg. Fahr. increases the ductility of steel without appreciably lowering 1s endurance limit. The result of tests on heat-treated steel is most readily shown by Fig. 3. A suggested explanation of this phenomenon is that drawing steel after It quenched produces two opposing effects: the hardness and strengtn of individual grains of steel ar somewhat decreased by drawing, but at the same time the interna: stresses set up in the process of quenching are somewhat relieved. These two contradictory effects seem to balance each other quite closely up to a drawing tempe ture of about 600 deg. Fahr. the carbon steels tested; for higher temperatures the weakening effect overbalances the stress-relieving effect, and the steel is weakened. These tests suggest the importance of a careful heat-treatment ot vey of any given steel as 4 gu) for its effective use. ra- for ner 28, 1922 1 to 9 inclusive show the relations found be- e endurance limits for various steels and other he properties. It will be noted that the results et] harpy impact test do not seem to be correlated th the endurance limit; that the static ulti- the Brinell hardness number seems to be a dex of strength under repeated stress than astic limit, and that the rise-of-temperature es s to indicate the endurance limit for wrought ferr netals with a good degree of accuracy. T foregoing discussion deals with reversed- adie nd an important problem is the relative de- struct effect of cycles of completely reversed stress j ycles of stress not completely reversed. In re} way test data indicate that a steel will with- tand maximum stress 50 per cent above the en- imit for completely reversed stress if, in the ’ stress, the range of stress is from zero to a maximum. For intermediate cases the limiting stress es between the values given above. In no case can depended upon to withstand repeated stresses yield point. Effect of Speed of Reversal of Stress The question of the effect of speed of reversal of tress on the result of repeated stress tests is of im- Tests have been made with the Farmer ro- oO oa Oo oOo 20 000 30 000 40 000 50 000 60 000 70 000 Fndurance Limit, Rise of Temperature Machine, |b. per sq. in. Correlation of Endurance Limit by Bending Test to Failure with Endurance Limit by Rise of Temperature tating beam machine at speeds of 200 r.p.m., 1500 r.p.m. and 5000 r.p.m.; the results are given in Table 1. These results show that at a speed of 200 r.p.m. the endur- ance limit averages about 6 per cent lower when using a specimen of 0.25 in. in diameter than at the speed of 1500 rp.m. and that for a speed of 5000 r.p.m. the endur- ance limit averages no higher than at 1500 r.p.m. These variations are not large and, while for any given omparison of materials it would seem wise to make tests at the same speed throughout, the effect of speed { reversal of stress on small specimens and machine parts does not seem to be very great. For larger machine parts where the path traversed by the stress vould be longer, especially for larger shafts, there may be a question of the interference of waves of ee ss as they traverse the member. In connection ‘i large machine parts subjected to repeated changes oad (a steam turbine rotor disk, for example), it may be necessary to make a study of the transmission ind int rference of waves of stress. ‘t seems worth while to attempt to picture the “ogress of a repeated-stress failure in a ferrous metal * I spreads throughout the crystalline structure. ‘der any conceivable condition of application of ae a machine part it seems probable that some S5lais s“ ubjected to heavy localized stress at bear- infavorably located for resisting stress, minute cracks. The failure of these few THE IRON AGE 783 crystals may not be the starting of a general failure of the member, but may constitute an adjustment of the internal structure of the member as cycles of stress are applied. If, however, the stress is above sOLane TRE apene Pe Table I—Effect of Speed on Endurance Limit of Steel Under Reversed Bending Stress Endurance Limit, Lb. Per Sq. In. a a 200 Cycles 1500 Cycles 5000 Cycles of Stress of Stress of Stress Steel Per Min. Per Min. Per Min. 1.20 per cent carbon, nor- rr a 48,500 50,000 50,000 1.20 per cent carbon, quenched and drawn... 100,000 SGaee) 0 ewes 0.49 per cent carbon, quenched and drawn... 45,000 ei cae eR ere ee 24,200 26,000 26,300 Chrome-nickel, quenched and drawn ...... 64,500 65,000 3.5 per cent nickel, water quench at 1490 deg Fahr., draw at 1100 ee s 65,000 64,000 Same steel, oil quench at 1450 deg. Fahr., draw at 1000 dee. Fahr 78.000 79.000 Same steel, annealed 2.000 54,000 “eeeee the endurance limit, these minute cracks spread. The roots of the cracks constitute points of localized stress and the damage spreads much as if minute hacksaws were cutting into the material. When the cracks have penetrated to some critical depth, failure suddenly occurs as in the case of a sharp-notched test speci- men under static or impact load. This progressive failure under repeated stress seems to be a very different phenomenon from the duc- tile flowing of metal stressed beyond the yield point. Localized .stresses and localized areas of weakness are far more dangerous under repeated stress than they are under stress applied only a few times. Machine Design and Localized Stress The machine designer of parts subjected to repeated stress must first of all recognize that the ordi- nary concepts and formulas of the mechanics of materials must be sup- plemented by a consideration of localized stress. It can scarcely be over-emphasized that the ordinary formulas of the mechanics of materials neglect many localized stresses; the formulas are not wrong, but for parts subjected to repeated stress they do not give a complete concept of the dangerous stress acting. In a few cases the intensity of localized stress may be estimated quantitatively. For example the localized stress at a sharp shoulder is about double the stress computed by ordinary formulas; the localized stress at a shoulder with an ordinary fillet or at the root of the standard screw thread may be estimated at about 50 per cent above the stress computed by the ordinary formulas. In many cases, however, the amount of localized stress cannot be estimated even approximately, and in such cases the designer must avoid sudden changes of form, sharp corners, sharp bearing edges, and any other causes of localized stress. Localized stress must be considered for all members subjected to thousands of repetitions of stress, whether the stress involves a complete reversal or not. If, however, the stress (including the estimate of localized stress) is applied in cycles in which the stress does not pass through zero (the stress at the chord mem- bers of a long-span bridge, for example), then the allowable stress can usually be determined by the 80 000 90000 100 000 ‘“« . .* ‘. i » ; ¢ +) ie iow sh) fa fi. fam i> te r, 1 ee Pe Rony ‘i; ‘ si 1 ii ” ae S # ’ fe “| il t a ad » Fol ae ° br be : ri ., |e ‘7k : > * : ‘ ‘ ~ an ae aaa dat § nls aie ii : ib ’ Bia MESES Rae FS. ey = ae ae jmoore < ae cue does Sa ee tet ee Sgroatee fetal 1B. Seinen’ 255 . ¥ — *, wens Oe rN a. ae ROTI Te +> - 2 - © ~ One iighakessinn LT - one a »* tae = era Deterbgurerana In Oo? aA . + “ ~ $a at _ ahem > “ ees : cnet Mc eens Sus « & Pitas So. oe - io tg bs 2 ag Can A a ~ Dm nee oe ede ig ad wo >t af eatesiues See ~ Secs 1 Sy 784 THE IRON AGE static yield point of a material rather than by its en- durance limit under repeated stress. It would be very unfortunate if the designer enter- tains the idea that machine parts, subjected to repeated stress, cannot be safely and economically designed. September 2%, 1999 They can be so designed, but it is, however. to consider both the static stress (including a of occasional overload) and also the strength ler re. peated stress, including both the strength of «ho ma- terial and the localized stress set up in the ma part Alloy Castings from Electric Furnaces Nickel, Chromium and Chrome-Nickel Steel Made on Pacific Coast—Their Production, Properties and Heat Treatment BY LARRY J. BARTON* NE of the newer lines of metallurgical endeavor () is the manufacture of alloy steel castings to be heat-treated and used as a substitute for forg- ings, and for purposes where the ordinary grades of steel castings will not stand up under the duty im- posed. The first company on the Pacific coast to enter into this class of alloys is the Los Angeles Foundry Co., which is marketing its product under the name of ., 1000 + 3 D> s ‘1500 y 4 2 sae ai t <4POM/¢ : eons a |4 f & o } o 9 se can be made much harder than manganese when scrapped can be used for other purposes After a great deal of experimentation and research covering a great many combinations of alloys it has been found that steels containing chromium, nicke! and vanadium offered the most feasible and versatile prod ucts for the different classes of work covered. Som: castings contain only nickel, some only chromium, som varying combinations of nickel and chromium, and som Af — Approximate Changes in Critical Temperatures as Caused by Alloys Electro-Cast Products. There are many cases, in the oil field and mining work, where a superior product is desired, and quite often the shape is such as to make forging very difficult, if not impossible. For this work it has been found that a properly made and properly heat-treated alloy steel casting furnishes a material which will fit into the work desired with the greatest ease. Not only can certain conditions be met, but by changing the treatment the same metal can be used for several different purposes. This results in being able to send out to a mine several castings, all of the same chemical composition which, when treated accord- ing to the directions sent with them, will cover widely differing specifications, There are many classes of work, formerly made of manganese steel, which are now made from other al- Joys. This is due to the fact that manganese steel cannot be machined, and if there are any small varia- tions in the casting, it cannot be used unless put through the necessary grinding to alter it. Then, too, when the piece is scrapped, due to wear, it cannot be used for any other purpose. Alloy castings are made to overcome these points. They can be easily machined, *Consulting metallurgical engineer, Los Angeles, Cal, consist of different amounts of all three elements Carbon contents are made suitable to the exact use the casting. The Melting Practice All are cast from basic electric furnace, as are the company’s other products, gray iron and “Duroloid, an alloy white iron for abrasive resisting work. Thi furnace is a standard 2-ton Greene type, lined to the slag line with magnesite, and similar in detail! t any basic furnace. Steel scrap of varying composition © used for the charge which requires quite a large amoun of chemical control work in order that speciicatl’’™ may be closely adhered to. Before the scrap 1s Chats” a light coating of lime is thrown on the hearth ®© the charge is melted down under a reducing atm Jitions as far as possible in order that oxidizing con’™™ - may be eliminated and alloys in the scra <oe soon as melted an analysis is run for the elemen desired, usually carbon and manganese, certain heats nickel and chromium are While this sample is being analyzed the slé up to a perfect carbide condition and the arsgee the the heat is checked by fracture tests. As 500" * ications osphere p saved. although ‘ also desired .o js brought ess of september 28, 1922 nary analysis is obtained the alloys are added heat is ready to tap. A skim gate is used and vot ladle for pouring. Molding and Heat Treating at care must be used in the core making and » for this class of work. Due to its heavy duty ‘ust be no interior flaws such as small shrink - or blow holes, and as a result larger heads and gating must be employed than would be used inary steel casting work. This results in a yield of cleaned castings from metal poured, , true quality product. heat-treating equipment is all standard, the eing both gas and oil connected. For quench- indard oil and water baths are used. Some mental work is being done with electrical heat- at the present time we have no results which able or which have been tested over any length All temperature checking is done by means i meters and each class of steel has been accur- ecked for critical points so it bécomes an easy to keep the castings at the temperature desired. obtained by means of a central temperature to which electrical leads come from each pyrom- ich furnace has several different colored lights ruidance of the operator. When the furnace the temperature desired, a light shows green; rect, white: and when too high, red. This, with recording meters, gives an excellent checking and supervising temperatures. Nickel Steel Alloy Castings N | has the same effect upon steel as carbon, th h to a greater degree, that of lowering the rmation points. For the ordinary grades of eel used in the above casting work, those of 0 per cent nickel up to 4.00 per cent nickel, it 0.50 per cent nickel will lower the Ac, range 0 deg. Fahr. and the Ar, on cooling about 20 from those of similar contents in plain This is shown in the chart. The approximate Reduc- Tensile Elastic Elon tion Strength, Limit gation, of Area, Lb. pet Lb. per Per Per Sq. In. Sq. In. Cent Cent 77.000 13.000 24.0 32.6 ( 1500 ul 79,400 4,200 26.2 41.9 ising stro? ! L500 gz back t S2,.000 57,000 27.3 $4.6 roy 1500 112,000 98.000 19.0 7.0 temperatures of these steels run from about 1310 deg. Fahr. (Ac,), 1250 to 1230 deg. Fahr. r the 1.00 per cent nickel, to about 1250 to ahr. (Ac:), 1100 to 1120 deg. Fehr. (Ar:) 1.00 per cent nickel, depending on the differ- niscellaneows composition, carbon, manganese, el steels offer a distinct difference from plain ‘ls when heated above the critical points for of time, carbon steel tending to grow in while nickel steels tend to become finer and grain. This is an advantage in handling a are ce of work because the time of soaking may be | and at the same time the grain refining superior. This, of course, raises the cost of ut in a strict quality product this is of importance, Treatment of Small Castings all size castings the general practice is to ‘ the upper critical range, allow to soak until is even throughout the piece of work being THE IRON AGE _ 785 treated, withdraw from the furnace and either oil or water quench. These pieces are then drawn back to the temperature necessary to give the toughness re- quired. On larger sized castings it is often necessary to give a double quench in order that the transforma- tion may be complete, nickel having the tendency to re- tard this action. A bearing casting was cast from metal analyzing carbon, 0.23 per cent; manganese, 0.75 per cent; silicon, 0.21 per cent; nickel, 1.53 per cent; phosphorus, 0.035 per cent, and sulphur 0.02 per cent. ‘The first table gives a comparison of the different results which may be obtained by varying the treatments. Some tests upon nickel steel averaging carbon, 0.25 to 0.30 per cent; manganese, 0.60 to 0.70 per cent; silicon, 0.20 to 0.25 per cent; nickel, 3.25 to 3.50 per cent; sulphur, 0.01 to 0.04 per cent, and phosphorus, 0.01 to 0.04 per cent follow: Reduc- Tensil Elastic Elon tion Strength Limit, gation, of Area, Lb. pet Lb. per Per Per Sq. In Sq. In Cent Cent As cast S5.000 $6,000 20.0 4s.0 Annealed from 1450 deg 83,500 43,600 24.4 53.2 Quenched in water from 1400 deg., drawn back to 900 deg 93.000 74.600 20.7 66.4 102,640 $1,100 19.8 71.7 Quenched in water from 1400 deg., drawn back to 550 dee : 137,000 113,800 13.4 SS 143.000 106.400 12.9 »d.7 Brinell hardness numbers will range from about 175 on the annealed specimen to 400 on some of the harsher quenches. Chromium Steel Alloy Castings Chromium acts in a similar manner to nickel in changing the critical points of steels, but in the op- posite direction, causing these temperatures to become slightly higher than those of plain steels of the same carbon contents. Chromium is exactly the opposite of nickel in regard to the effects of overheating for pro longed periods, rapidly deteriorating in strength and resistance to shock. For this reason a double quench gives superior qualities to these classes of steels. Chromium steels are capable of taking great hardness without a corresponding decrease of toughness, which makes them desirable for any such purpose as crusher jaws,, wearing plates and similar uses. As this hard- ness comes directly from the double carbide of iron and chromium the steel must be softened according to the amount of machining to be done. The general treat- ment on this class of steels is similar to that on nickel steels—quench from above the critical point, either draw back in air, or re-heat and re-quench at the lower temperature. Their properties are about as follows: Reduc- Tensile Elastic Elon- tion Strength, Limit, gation, of Area, Lb. per Lb. per Per Per Sq. In Sq. In. Cent Cent Analysis: Carbon, 6.60 to 0.70 per cent; chro- mium, 0.50 to 0.75 per cent Quenched at 1500 deg., drawn back to 1300 GO scstieves hanes 79,000 64,500 34.4 66.6 Drawn back to 1050 deg. 100,600 94,300 20.1 60.0 Drawn back to 800 deg 114,000 103,400 14.3 54.4 Nickel-Chromium Steel Alloy Castings This class of steel probably represents the best general type of steel for heat-treated castings. The steel, when treated, shows the beneficial result of the toughness and ductility of the nickel together with the increased hardness and cohesion of the chromium. The best results are obtained by having a certain ratio of the two metals, usually about 2.50 per cent nickel to 1.00 per cent chromium. The general method of ocean ap niendinrmiereterer Sey SA eg pehe ge heks ie eee ae Taegan 786 THE IRON AGE treating is to quench in oil from a temperature higher than the critical point and draw back to the temper desired. For very high specifications a double quench GOUELSLUAERLLCCELLUDANERADADEOAESROGUEROEURELAOORGANIAALDUELEDOECLEGOONEDOGEEDEOONGGTONNEOGURELOULEUUAONNLOUENDEOLNGLLALORELOOGGSLLOEELI AECL DOCERELAENCLRENELOECEEL YOON OCOSEDOREEOOOERELIE Reduc- Tensile Elastic Elon- tion Strength, Limit, gation, of Area, Lb. per Lb. per Per Per Sq. In. Sq. In. Cent Cent Analysis: Carbon, 0.30 to 0.40 per cent; chro- mium, 0.50 to 0.70 per cent; nickel, 1.30 to 1.70 per cent. Annealed from 1500 deg. 66,700 37,000 f 25.8 Annealed from 1350 deg. 65,000 32,600 b 54.3 Quenched from 1500 deg., drawn to 1000 deg.... 90,000 65,000 : 61.6 Quenched from 1550 deg., drawn to 900 deg 84,400 50,050 ‘ 57.7 eneLNNNOOUNDDOONEDOONNTONNRLY cupenernanueeiuceuusasonecreutT sentannne HHUUUUEEaAOEODN CRO NONNEOOONtOLONEDOOONAONONONOD is used. Quench from about 200 deg. Fahr. above the critical point, re-quench from just above the critical September 28, i999 point, and then draw back. Their properties ar, as tabulated above. The above results go to show in a general may what can be accomplished along these lines, but by no means put forward as a text or as the a figures which can be obtained. There are many :; where castings poured from the same heat and in the same manner will show widely differing , ults, We attempt to explain this by a difference’in structyre caused by such points as a different temperature y sine poured, different texture and temperature of the mold, ete., but, as we have no microscopical test laboratory. we are undecided at this time. These points wil] be proven later when we have absolutely determined. through wider use, the true value of the product as shown by its ability, not to fulfill tests or laboratory specifications, but in the test under actual working conditions. Pearlitic and Sorbitic Manganese Steels A Plea for Castings Containing About One Per Cent Manganese— Some of the Literature on the Subject—Their Heat Treatment and Properties BY JOHN HOWE HALL* HE adoption this year of tentative specifications for steel castings for railroads by the American Society for Testing Materials, in which the up- per limit for manganese is set at 0.85 per cent, calls attention again to a most curious and persistent fal- lacy,—that steel is rendered brittle by a manganese content of 1.00 per cent or a little higher. This idea appears to have originated in the early days of the industry when chemistry was fighting to obtain a foothold in the steel works. Failure of va- rious heats of the Bessemer rail steels of ‘those days was traced to an unusually high manganese content, and a dead line was set at 1 per cent. “Men are sheep, bloomin’ sheep,” as Ortheris remarked, and since those old days but a very few steelmakers have ven- tured beyond that ancient dead line at 1.00 per cent manganese, and for some curious reason the experience of those who have penetrated into the forbidden terri- tory has not attracted the attention it deserves. To some extent, the same thing appears to hold true of phosphorus and sulphur; the upper limits for these elements were set many years ago, when the scope of metallurgical knowledge was limited indeed, and until very recent years no one has thought it worth while to re-examine carefully the premises upon which the allowable content of these impurities is based. The joint committee on phosphorus and sulphur in steel, now investigating this subject carefully and _ thor- oughly, may in time be followed by a similar commit- tee to determine the desirable content of manganese and silicon. As a young man fresh from technical school, the writer became familiar with a steel, made on a regu- lar production basis, analyzing from 0.30 to 0.50 per cent carbon, and from 1 to 1.50 per cent manganese or a little over, used quite largely for purposes where a combination of strength and toughness was essential. From his notes, dated 1902, are taken the following analysis and tests of some of this steel, made in the *Metallurgical engineer, Taylor-Wharton Iron & Steel Co., High Bridge, N. J. Bessemer converter, which show its freedom brittleness, to say the least: Carbon, Manganese, Phosphorus, Sulphur, Per Cent Per Cent Per Cent Per Cent 0.24 1.62 0.06 0.019 Tensile Elastic Elon Strength, Limit, gation, of Lb. per Lb. per Per Sq. In. Sq. In. Cent Annealed bars......... 97,780 67,230 27 Annealed bars......... 99,420 68,990 7 Oil quenched and an- nealed 118,000 98,000 24.0 It may readily be imagined that the idea that ste¢ containing over 1 per cent manganese is brittle was early dislodged from the writer’s mind. The late Maunsel White had much to do with the development of this steel as a regular commercial product. In 1908 and 1909, the late Dr. Henry M. Howe and the writer were called upon to produce a stee! for castings, made in a 3-ton bottom blown Bessemer converter, that could be used satisfactorily for gold dredge buckets. These castings were subjected to ex- tremely severe stresses in service, and to heavy wear, —so much so that to-day they are regularly made 0! 12 per cent (Hadfield) manganese steel. Experiment had demonstrated that plain cast. steel of 0.20 to 0.30 per cent carbon was too soft and weak for the service demanded, and that harder steels almost invariably failed by breaking under the tremendous stresses 0 casionally set up in the line of buckets. Experiments with nickel and chromium did not promise a stee! with sufficient strength and toughness and remembering the manganese steel mentioned above, we tried it for these castings, which in addition were to be water quenched and annealed, to give the highest possible strength and elastic limit for an extension of 25 per cent and a con- traction of area of 40 per cent or better. Buckets of this steel, protected by plates of 12 per cent mangé nese steel at the points of greatest wear, stood er service successfully, and held their own until displaced by the one-piece 12 per cent manganese stee! bucket. Sentember 28, 1922 m a book “Etude Industrielle des Alliages Metal- by F. Guillet, published in 1906, the following ex ts are taken: Of the steels containing about 0.20 per cent rbon, those of the first class (from 1 to 5 per nt manganese) possess a tensile strength and elastic limit which rises slowly with the inganese content, while the extension and con- .ction of area diminish slightly; these steels ow @ very great resistance to shock.* This int is extremely important, and to show it arly, we have broken more than 100 Fremont st pieces. The results obtained contradict ssolutely the very widespread opinion among etallurgists that manganese makes steel brittle. Ve believe we have demonstrated clearly, on the ntrary, that manganese makes steel brittle only en it is present in sufficient amount, or better, hen the sum of the carbon and manganese is gh enough, to lead to the production of mar- ensite. As long as the structure is pearlitic, steel, far from being brittle, is extremely mogeneous and possesses a resistance to shock hich one finds only exceptionally in carbon ‘ 9 * They (manganese steels) can, up to a cer- point, be substituted for nickel steels, and the more readily since for equal physical perties they are much less costly. Much less manganese than nickel, in fact, is iired to produce a given effect, about 2.2 less. . . . The pearlitic (manganese) els can be used in the place of similar nickel is; that they are not more used, is due, we to the false idea of the brittleness pro- d by manganese. There is, however, int that should be made, which is that ganese steels become brittle upon annealing readily than nickel or carbon steels. For reason, one should not think of using them case-hardening. . . To sum up, we can say the manganese steels have not yet been ntly studied from the industrial point of ics are the author’s, | steel was described by the writer in a paper, Heat Treatment of Hypo-eutectoid Carbon Steel Cast- Proceedings A. S. T. M. 1913, Vol. XIII page 514, in a paper, “Shock Tests of Cast Steels,” [THE Iron Ace, Oct. 28, 1913], Proceedings American Inst. Met. Engrs., Vol. XLVII (1913), page 482; book, “The Steel Foundry,” (1914), in a paper n, Taylor and Hall, “The Heat Treatment of eel,” Proceedings American Inst. Min. and Met. g Vol. LXII (1920), page 353 and elsewhere. A earch of technical literature would no doubt ult in scores of other references,—among others an by F. C. Langenberg, “Impact Tests on Cast Steel,” Proceedings A. S. T. M., Vol. XXI (1921), in which the great toughness of these steels le ck is especially noted. In a paper by George Lang, in Stahl und Eisen, 1911, “The Influence of Manganese on the Properties of Soft Steels,” results are given vy that manganese up to even 1.77 per cent or CVU ULELUAOU PACED ONENEDUOGNOOEDLEDDOMORONEED ONT eEtEN Ua oenneeTHUNtEnaNRCraNe THE IRON AGE 787 higher confers great toughness on steel of 0.1 per cent carbon. The following results are of interest: saute /NEDEOLLANONLLEDOROLERELH OTE ODORRDDNEES | (0) (440460106) )eumeNRERANEER EDN D TT TEY ny EUpUSMEBRREERNN sexy Her /oUNe Physical Tests of Rolled Steels with Varying Manganese Content Reduc- Manga- Elastic Tensile Elon- tion nese, Limit, Strength, gation, of Area, Per Lb. = Lb. a Per Per Hard- No. Cent Sq. In Sq. In. Cent Cent Shock ness : ef 46,000 62,600 27.85 53.5 23.2 119 2 0.44 47,600 65,000 29.2 62.5 29.4 115 3 0.68 49,700 66,200 29.4 63.9 28.1 115 4 0.79 49,200 64,500 27.85 65.9 33.4 118 5 1.02 51,600 68,000 28.65 64.2 31.8 123 6 1.27 54,300 71,300 27.9 68.1 26.4 132 7 1.32 54,000 71,200 28.0 67.7 34.0 133 8 1.77 57,900 82,500 25.9 54.7 25.1 156 9 1.84 58,400 $5,000 26.0 46.75 23.6 162 10 2.23 61,000 92,600 21.2 43.4 9.5 198 1l 2.47 63,400 103,000 18.8 39.4 2.7 211 Physical Tests of Steel Heated to 900 Deg. C. and Quenched, Having Varied Manganese Content 0.29 42,900 78,700 19.6 61.0 23.6 146 2 0.44 47,900 82,500 19.18 62.0 27.0 165 3 0.68 48,600 84,100 16.0 62.2 29.4 154 4 0.79 49,600 85,000 14.1 56.1 30.0 157 5 1.02 54,200 96,800 15.52 53.0 27.2 179 6 1.27 61,550 107,900 15.24 48.2 18.4 230 7 1.32 66,000 111,200 16.3 47.8 18.1 237 8 1.77 86,000 155,000 9.0 44.6 14.8 273 9 1.84 89,000 157,300 6.7 41.7 14.5 288 10 2.23 91,000 164,500 4.91 26.6 12.3 326 11 2.47 92,200 166,600 3.1 18.9 11.7 144 Another reference worth quoting is an article “An Alloy Steel of Special Composition,” in the Journal of the American Society of Naval Engineers, abstracted in THE IRON AGE, Dec. 30, 1915, page 1553. Im this article is described a steel of the following composition: Carbon, 0.30 to 0.35 per cent. Silicon, 0.25 to 0.35 per cent. Phosphorus and sulphur, below 0.05 per cent Manganese, 1.00 to 1.20 per cent Nickel, 1.50 to 1.80 per cent Copper, 0.50 to 0.80 per cent which gave physical properties comparing favorably with those of 3 per cent nickel steel. Some of the test results appear in the table below. From a letter of the late Dr. Henry M. Howe, writ- ten in 1909, the following is quoted: The line N.P. (plotting elastic limit against extension) represents the upper limit of steel forgings of unusually high quality which I gathered many years ago, . . . you will see from this that our coupons (of 0.25 per cent C., 1.25 per cent Mn. cast steel, quenched and an- nealed) are better than the best published forg- ings at the time I made the compilation. I have in my notes 255 sets of published results of steel castings which I have taken the trouble to compare with these results. Out of these 255, only three cases come up to the line N.P. (rep- resenting high quality forgings) and these three are not so good as the average of our coupons by a good deal It thus seems that we are making the best castings of which we have any knowledge From a paper by the late Dr. Henry M. Howe, “Manganese in Steel Making,” Proceedings A. S. T. M., 110000 AAURDUERONERA SERED SDERED IER: HH ROS cerOHLNE ce8mmE Physical Tests from Coupons Cast on Body of Castings Tensile Yield Reduc- ; Man- Phos- Strength, Point, Elon- tion Sulphur, ganese, Silicon, phorus, Nickel, Copper, Lb. per Lb. per gation, of Area, Bend, PerCent PerCent PerCent PerCent PerCent PerCent Sq. In. Sq.In. Per Cent Per Cent Degrees 0.036 0.81 0.31 0.034 1.60 0.76 92,283 55,767 22 37 120 0.037 0.93 0.36 0.04 1.44 0.85 92,334 55,003 21 31 120 0.030 1.02 0.38 0.035 1.15 0.49 88,616 54,494 25 40 120 0.034 1.04 0.36 0.033 1.14 0.52 92,691 56,277 22 34 120 0.035 1.11 0.33 0.042 1.34 0.60 95,543 59,078 22 37 120 0.032 0.93 0.35 0.043 1.12 0.77 90,042 57,550 23 37 120 0.04 1.18 0.35 0.046 1.60 0.61 92,691 55,003 22 36 120 0.042 1.15 0.35 0.035 1.58 0.69 94,830 54,748 22 36 120 0.035 1.02 0.30 0.047 1.47 0.42 $5,510 56,022 23 34 120 0.029 1.02 0.35 0.049 1.10 0.80 88,718 55,512 23 38 120 0.031 1.15 0.38 0.046 1.60 0.56 96,510 59,587 23 37 120 ¢ 0.036 1.18 0.31 0.049 1.37 0.80 94,117 50,929 23 36 120 0.036 1.03 0.35 0.039 1.47 0.63 90,144 55,225 23 34 120 0.035 1.03 0.40 0.04 1.73 0.21 96,052 59,087 25 41 120 : 0.036 1.21 0.36 0.049 1.24 0.65 91,366 55,003 25 44 120 94 i 0.032 1.15 0.34 0.043 1.54 0.70 81,343 60,096 25 41 120 945 * 0.033 0.96 0.39 0.042 1.27 9.64 87,343 53,457 23 36 120 948 + 0.032 1.15 0.37 0.087 1.80 0.62 92,181 55,288 23 36 120 ° 0.032 0.79 0.42 0.048 1.69 0.65 87,191 . 54,239 23 38 120 *F : ‘i m4 s * rf ’ Re . ner we, r Oe I aera a emp mg canes eae ~ + ON ae ae a et a te : 4} a po ER os ee Ame eae SO SEI Ret ee A 788 THE IRON AGE September 28, 1999 é vol. XVII (1917), page 5, abstracted in THE IRON As we have already stated, these are but a few o; : Ace, for Aug. 2, 1917, page 239, the following is taken: many references which might be quoted on ban : | sae . teresting subject, but all tell the same sto) t f Turning now to carbon steel, the retarding sl ‘ s “i ‘ im effect of manganese on the structural changes msageae in excess of 1.00 per cent is a by ' : i . shows itself by leading in general to ee struc- a detriment, at least to low carbon steels. me. ture, to finer ferrite masses, finer network struc- . a3 ; : ture and finer pearlite, indeed probably often The writer has referred before to the suc iif : to the replacement of lamellar pearlite with plication of a similar steel in car coupler knuc k - sorbite This greater fineness leads to better 7 ‘he es : Fa : quality in general, and to a higher elastic limit of the large manufacturers, = aware of the u ae in particular, though of course with a corre- steels for motor shafting subject to severe ser\ Beasts sponding sacrifice of ductility The great value sf . ‘ ‘ ; ss ay of manganese for this purpose has not begun to with the employment of a steel of about 0.50 ’ + v4. receive the attention it deserves. It is probable carbon, with 1.00 per cent manganese i? apt! f° fe that a manganese content of, say, 1.25 per cent, . a i A gh: ; with a correspondingly lessened carbon content, for oxygen, hydrogen, and other Gases, an ap} ii ‘ae may be use