ELEVATED TEMPERATURE STATIC PROPERTIES OF WROUGHT CARBON STEEL Prepared for the Metal Properties Council by G V Smith ASTM SPECIAL TECHNICAL PUBLICATION 503 List price $3.00 04-503000-40 ^! AMERICAN SOCIETY FOR TESTING AND MATERIALS 1916 Race Street, Philadelphia, Pa 19103 © BY AMERICAN SOCIETY FOR TESTING AND MATERIALS 1972 Library of Congress Catalog Card Number: 74-185538 NOTE The Society is not responsible, as a body, for the statements and opinions advanced in this publication Printed in Alpha, N J IVIarcli 1972 Introduction In terms of tonnage, carbon steel is the most important material employed in boiler and pressure vessel service This material, limited in the present context to less than 0.35% carbon, may be referred to as "simple carbon steel", as it sometimes is, only in a facetious sense, and, in fact, it exhibits complexities of behavior that match those of alloy steels Carbon steel is employed in a wide variety of applications, and although some progress towards simplifying the specification picture has been made in recent years, the 1968 edition of the ASME Boiler and Pressure Vessel Code still lists a total of 32 individual specifications pertaining to carbon steel and many of the specifications provide for several grades (Many engineers believe that the multiplicity of specifications for steel is unwarranted.) Carbon steel is employed at temperatures ranging from below room temperature to as high as 1000''F in ASME Code applications, and even higher in non-code applications Consequently, in establishing allowable stresses, not only are conventional short-time tensile properties required, but also the creep and rupture properties In some applications, fracture toughness, in the sense of transition at ambient temperature, and/or fatigue strength may be of interest, but these latter properties are beyond the scope of the present review In this review, consideration will be given to the effects of small quantities of certain alloying elements, e.g molybdenum and niobium, which by narrow definition would not normally be viewed as present in carbon steel And, we shall restrict our interest almost entirely to the ferrite-pearlite microstructural condition; in principle, it is possible to develop other microstructural conditions, though in practice, the cooling rates that would be required cannot be achieved except in sections generally less than those of practical concern From investigations carried out over many years, we know that the elevated temperature properties of carbon steel may be sensitively affected by a number of variables, including chemical composition, deoxidation practice and processing Nevertheless, when the data available from tests of materials of commercial manufacture are examined, as for example, in a recently published report on carbon steelt^J prepared by The Metal Properties Council (MPC), it becomes difficult or impossible to discern these individual effects Information concerning some of the variables now thought to be important may be lacking, and always there is an uncontrolled, simultaneous variation of different factors And, the problem of isolating the effects of individual variables is compounded by the large inherent scatter in behavior even amongst nominally identical lots An important consequence of the sensitive dependence of elevated temperature strength upon manufacturing variables is that the scatter band of strength is very wide; this, in turn, is reflected in the level of working stresses permitted, which must recognize not only this scatter, but more importantly, the minimum of the scatter band If the individual variables could be identified and their importance assessed, it should be possible, in principle, to prepare tighter *Consulting Engineer, Ithaca, New York specifications leading to a reduction in the scatter and to an increase in the minimum expected level Because it has not been possible to achieve this goal by study of the available test results, MPC is giving consideration to planning a statistical-design test program for this purpose A necessary first step is to put in proper perspective what is already known; this review endeavors to perform such a service In reviewing the properties and behavior of carbon steel at elevated temperatures, it is convenient to consider independently yield and tensile strength on one hand, and creep and rupture strength on the other A third section considers ductility and toughness at temperature Yield and Tensile Strengths Establishing Allowable Stresses Before giving consideration to the factors affecting yield and tensile strength at elevated temperatures, it may be of value to note how these properties are employed in setting allowable stresses at temperatures below that level at which creep becomes important In the ASME Code, the maximum allowable stress is the least of the following properties, each multiplied by a fraction which may differ amongst the different Sections of the Code: 1) specified minimum yield strength at room temperature 2) specified minimum tensile strength at room temperature 3) yield strength at elevated temperature 4) tensile strength at elevated temperature Which of these actually limits the allowable stress in a particular instance depends upon the relative levels of yield and tensile strengths, and how they vary with temperature (and of course upon the multipliers) In contrast to the ASME Code, the draft codes of the International Standards Organization (ISO), and European Codes generally, provide that the allowable stress below the creep range shall be the lesser of the specified minimum tensile strength at room temperature, multiplied by a fraction, and the minimum yield strength at temperature multiplied by a fraction Thus, the ISO Code, unlike the ASME Code, is not concerned with the level of tensile strength at temperature, and, except as the criterion on tensile strength at room temperature may be limiting, thereby leads to higher allowable stresses at elevated temperatures, than under the ASME Code, for materials having a high ratio of yield strength to tensile strength From the foregoing, it will be evident that, in furthering the objective of higher allowable stresses, we must be interested in improving strength not only at elevated temperatures but at room temperature, as well Thus, in reviewing the subject of the elevated temperature properties of carbon steel, we shall first give consideration to the factors affecting room temperature yield and tensile strengths Yield and Tensile Strengths at Room Temperature In a very general way, strength of carbon steel at room temperature may be expected to increase with increasing carbon, increasing fraction of pearlite (which is related to carbon content), with increasing manganese and/or silicon, with increasing quantity of the "residual" elements, e.g nickel, copper, molybdenum, and with decreasing grain size A given variable may have a greater effect upon yield strength than tensile strength or vice versa; thus, as will be shown later, grain size has a much greater influence upon yield than upon tensile strength Other, more subtle or indirect effects, related to deoxidation practice and heat treatment, and which will be considered later, may be important, and affect not only strength at room temperature, but the temperature variation of strength as well.* The effects of the variables cited upon the strength of relatively low carbon pearlitic steel at room temperature have been treated extensively in many publications, e.g in reference 3, but usually in a qualitative or, at best, semiquantitative manner However, an important series of investigations,''"' conducted in England in recent years, and employing the methods of multiple regression analysis, has provided useful quantitative assessments of the effects of various compositional and microstructural variables In the first report of these investigations, Pickering and Gladman'* developed, for carbon-manganese steels containing up to 0.25% C and 1.5% Mn, the following relations that define the dependences of lower yield point (L.Y.P.) and tensile strength (T.S.) upon the microstructural features, grain size and fraction of pearlite, and upon two of the alloying elements commonly present in carbon steel, namely silicon and manganese: T.S (tons per square inch) = 19.1 + 1.78 •1/2 (% Mn) +5.35 (% Si) + 0.10 (d 0.253 (% pearlite); and L.Y.P (tons per square inch) = 6.7 + 2.11 (% Mn) + 5.44 (% Si) + 0.255 (d'-^^^), -1/2 ) is the reciprocal square root of where (d the grain diameter in inches, (% pearlite) is the volume fraction of pearlite, and (% Mn) and (% Si) are weight percentages Further studies made it possible to incorporate the effects of other variables of chemical composition, including an important effect associated with the amount of free nitrogen, % N , (i.e., nitrogen uncombined with other elements, such as aluminum, also present) and in a recent paper,^ the results of the regression analyses were reported as follows, where the true flow stress at a true strain of 0.2 (roughly 20 per cent) is assessed, rather than tensile strength: yield strength (tons per square inch) = 3.3 + 2.2 (% Mn) + 5.3 (% Si) - 14 (% S) + 24 (% P) + 8.8 (% Sn) + 180 (% NJ + 1.4 d'-1/2 flow stress at a strain of 0.2 (tons per square inch) = + (% C) + 4.4 (% Mn) + 9.6 (% Si) - 22 (% S) + 56 (% P) + 11 (% Sn) + 3.9 (% Cr) + 260 (% N^) + 1.1 d •1/2 *And at a higher level of sophistication, the interlamellar spacing of the pearlite may be a variable, although it is a less important one at the relatively low carbon levels that characterize pressure vessel steels than in higher carbon steels In these equations, the effects ascribed to the several elements are those other than the possible effects upon grain size and fraction of pearlite It is of interest to note that whereas the tensile strength (or flow strength beyond yielding) is importantly dependent upon carbon content (or pearlite fraction) within the limits studied, i.e up to 0.25% carbon, yield strength is independent of these factors The important effect of grain size upon yield strength, illustrated in Figure 1, has been a factor motivating engineers,"' particularly in countries that have adopted the allowable stress criteria of the draft ISO Codes, to place special emphasis upon development of structural steels having fine grain size It will be recalled that under the ISO Codes, higher yield strength steels tend to be rewarded with higher allowable stresses at elevated temperatures to a greater extent than might be true under the ASME Code Fine grain size is also especially beneficial to improved fracture toughness Some of the alloying elements that contribute to strength may have adverse effects upon toughness, or they may be relatively costly, and, of course, can be intentionally added only in restricted amounts if the resulting material is not to be viewed as an alloy steel rather than as a carbon steel Control of grain size can be effected by employing additives and/or by control of processing The grain refining effects of small amounts of aluminum, niobium, vanadium and titanium additions to C-Mn steels in the normalized condition were extensively studied by Irvine, Pickering and Gladman.^ They reported that marked grain refinement resulted from the relatively small additions of 0.05% Al, 0.02% Nb or 0.1% V; however, titanium was ineffective unless present in an amount greater than 0.1% Also, as a result of carbide precipitation, niobium and vanadium caused not only grain refinement but also caused strengthening over and above that associated with decreased grain size Although vanadium carbide (at the indicated concentration) is completely soluble at relatively low austenitizing temperatures, niobium carbide exhibits a solubility which increases with increasing temperature Consequently, unlike carbon steel or vanadium steel, the strength of a niobium steel for a given grain size depends upon the austenitizing temperature, as illustrated in Figure A particular virtue of grain size control by way of aluminum addition is that nitrogen may be effectively removed from solid solution, depending upon heat treatment, with a beneficial influence upon fracture toughness at ambient temperatures Studies such as that cited, and others, have resulted in the introduction of commercial fine grain steels employing one or more of the indicated additives A number of basic guiding principles have been ennunciated by Irvine, et al." Small additions of Al, Nb and V are effective in grain refinement because these elements form fine nitride and/or carbide precipitates which act to inhibit normal grain growth The formation of these precipitates is abetted by a normalizing heat treatment It is not always appreciated that a "fine grain" steel may not be fine grained in the hot rolled condition, depending upon the heating temperature for rolling, upon the concentrations of the agents responsible for grain size control, and upon the "finishing" temperature of hot rolling Thus, an aluminum deoxidized steel (fine-grain practice), finished at a relatively high temperature may be expected to exhibit coarser This has been found to be true of carbon steel, and ferrite grains than a silicon-deoxidized steel the "trend curve" showing the variation of tensile (coarse grain practice) finished at a relatively strength of carbon steel with temperature may show low temperature striking differences related to differences in composition and manufacturing procedures.(^J This possibility of grain size control by control of processing, cited earlier as the second of The most prominent feature of the temperature the two possible approaches to grain size control, dependence of yield and tensile strength of carbon may prove in some instances to be the more economisteel is the occurrence of dynamic strain aging, cal alternative in thicknesses up to to 1/2 which, in a susceptible steel, manifests itself as inches, and "controlled rolling" has come into a levelling off, or as a reversal, in the trend of widespread use Amongst the processing variables decreasing tensile strength with increasing temperathat have been found to be important, in addition ture Figure Since the stage for strain aging to finishing temperature, already cited, are the during an elevated temperature tensile test is set composition of the steel, the amount of deformation by the plastic strain occurring in the test itself and the temperature range over which it is imposed, (except when there is residual cold work from prior the rate of recrystallization and grain growth processing*), the effect upon yield strength would (which of course depends upon the composition), and be expected to be much less marked than for tensile the time available for these, and finally the rate strength, or even unobservable The peak in of cooling through the transformation range after tensile strength, which may be at a higher level completion of rolling (which depends in part on the than at room temperature, tends to occur at about thickness) The effects of these variables in both 450 to 500°F, for conventional strain rates, and carbon-manganese and grain-refined, low carbon is displaced to higher temperature with increased structural steels, and their interrelation, have strain rate A minimum in elongation or reduction been extensively examined by Irvine and associates.^ of area may be expected at the temperature at which Amongst the more important principles that were the tensile strength exhibits its peak identified are the following: Strain aging in carbon steel is associated with the interstitially dissolving elements carbon The amount of deformation must be sufand nitrogen, interacting with dislocations in the ficiently large, generally 50%, to give a ferrite phase With carbon always exceeding the fine recrystallized austenite grain size solubility limit, differences in strain-aging susceptibility of commercial structural steels are The finishing temperature for carbon steel to be associated primarily with differences in must be low, on the order of 1475°F, but "available", or free, nitrogen Nitrogen which is not so low that ferrite is present before present inadvertently (but may also be deliberately the deformation is ended; for niobiumadded) tends to react with elements such as alumitreated steels, the finishing temperature num and silicon, especially the former, that may be may be as high as 1650°? added for deoxidation [In a number of respects, There should be no holding in the later the interaction of the deoxidant with nitrogen is stages of rolling, particularly in plain of greater importance than with oxygen, e.g in carbon steels respect to grain size control.] Nitrogen also reacts, though not strongly, with niobium and Although not as yet widely employed, mention vanadium that may have been added for grain size should be made of a growing interest in the use of control The extent to which the nitrogen of a accelerated cooling to enhance the strength of given steel reacts and is thereby unavailable to carbon steel at room and elevated temperatures, cause strain aging depends upon the concentrations particularly steels containing larger amounts of of the reactants and upon temperature and time manganese and silicon Thus, ASTM AS37 provides in Thus, whereas, an aluminum deoxidized steel may be its Grade B for quenching and tempering, with expected to have minimal strain aging tendency when martensitic structures achieved in sections up to in the normalized condition, because heating at about 5/8" thick Accelerated cooling can lead to about 1600°F provides a favorable opportunity for strengthening, even if martensite is not formed, the reaction of nitrogen with the aluminum, the because of the lowering of the transformation same steel hot rolled from a temperature at which temperature, which in turn leads to a finer carbide aluminum and nitrogen are in solution may be dispersion expected to exhibit a relatively high degree of strain-aging susceptibility Figure (to be discussed later) Yield and Tensile Strength at Elevated Temperatures Reheating of a hot rolled, aluminum deoxidized In a very general way, elevated-temperature steel for tempering or stress-relief also provides yield and tensile strengths vary with the values of a favorable opportunity for immobilization of these properties at room temperature, and therefore, nitrogen, the reaction proceeding at a rate which to a first approximation, the former may be preincreases with increase of this temperature dicted from the latter Such an approximation may Silicon, commonly used as a deoxidant for "coarseprove to be fairly reliable within the relatively grained" steels, may also effect immobilization of narrow limits of a given commercial grade, and a nitrogen during tempering or stress-relief, but the data normalizing procedure that involves ratioing reaction does not proceed so rapidly nor, on the elevated temperature strength of an individual heat whole, to as complete an extent, and in fact, to the room temperature strength of that same heat Figure shows only a slight effect of tempering or has proved to be an effective tool in evaluating stress-relief upon the trend curve for steel made the temperature dependence of strength; in fact, the true dependence can be obscured if the data are *It seems probable that at least some of the scatter not first normalized.(^^ However, the exact in strength at elevated temperatures, particularly dependence of strength upon temperature may differ in yield strength, is to be attributed to strain significantly from one grade to another, and in aging associated with the presence of prior cold some instances even within the same nominal grade work, perhaps in some instances from cold straightening operations to coarse-grain practice As indicated earlier in this review, information concerning some of the variables known to affect strain-aging susceptibility are often lacking Hence, in the recent MPC evaluation f^-> of the elevated temperature yield and tensile strengths of carbon steel, it seemed possible to adopt only the relatively simple categorization of Figure 3, which shows mean trend curves (expressed in ratio form) derived by polynomial regression analysis of the data scatter bands Several aspects of the strain aging phenomenon previously indicated are evident Firstly, steel made to fine-grain practice may exhibit nearly as great a tendency to strain aging as that made to coarse-grain practice, when tested in the untempered condition.* Secondly, tempering the fine-grain material results in a considerably lessened tendency to strain aging, whereas it has only a slight effect upon material made to coarsegrain practice Another feature of Figure to which attention might be drawn is that the differences amongst the four categories are relatively small at temperatures both below and above the range (approximately 250 to 750°F) in which strain aging manifests itself This suggests that, for given strain-aging susceptibility, improvement of elevated temperature tensile strength can be effected directly by effecting improvement in room temperature tensile strength Since the ultimate objective of this review of factors affecting elevated temperature strength of carbon steel is related to the setting of allowable stresses, it may be of interest to inquire whether and to what extent differences in strain aging, such as are evident in Figure 3, should be recognized in the setting of stresses Of course, this question is of interest only in relation to codes, such as the ASME Boiler and Pressure Vessel Code, in which tensile strength at elevated temperatures is one of the criteria, along with yield strength, determining allowable stresses It is generally held that the purpose of limiting the allowable stress to a fraction of the yield strength is to insure against general plastic distortion of a structure, whereas the limitation to a fraction of the tensile strength is in the interests of protecting against bursting Since bursting would presumably only be a hazard under emergency or rapid, pressure buildup conditions, the attendant strain rate would be substantially greater than that employed in the conventional tensile test, and the time available for strain aging inadequate within the temperature range normally associated with strain aging.** For this reason and also because it would be quite impractical to employ an allowable stress greater than that which could be sustained at all lower temperatures, no recognition is given by the ASME Boiler and Pressure Vessel Committee to the rise in tensile strength at intermediate temperatures in setting allowable stresses In principle, it might be possible under certain circumstances to deliberately employ prior *Nearly all of the data for fine-grain, nottempered, material represented the as-rolled condition It is possible that normalized material, even though not tempered, would have exhibited a lesser tendency to strain age than indicated for the not-tempered category **The temperature of peak susceptibility is shifted progressively to higher temperature by increasing strain rate strain aging to increase strength at room temperature (at the expense of toughness), and the improvement in strength thus effected might be expected to persist to some 600-700°F However, one very practical limitation to such an approach is that the stress-relief treatment that many vessels receive (required under the ASME Code for welded vessels exceeding a specified minimum thickness) would obliterate strenghening developed by strain aging Mean yield strength ratio trend curves corresponding to the four categories of Figure are reproduced in Figure 4.(lJ The slight perturbations in the individual curves are of questionable significiance, in view of the extensive scatter of the individual yield strength plots from which they were derived; yet, there is some measure of order to the perturbations and some resemblance to trend curves developed by the British'^^J from more systematically generated data, to be discussed later It is also uncertain whether the differences amongst the yield strength ratio trend curves (on the order of ±10 per cent from an average) are significant, and further, more-carefully planned tests would be required to elucidate this question.* Recognizing the likelihood that tempering would be employed during or after fabrication of many vessels, and that the differences in trend curves are of questionable significance, a common trend curve for the two tempered conditions was computed "^ ' This curve, which showed an essentially linear decrease with temperature was employed in the summary comparison of the MPC evaluation.^^^ It would be reasonable to expect that the tendency of carbon steel to strain aging is altered when other elements tending to interact with carbon and/or nitrogen are present Manganese is commonly present in carbon steel and many other elements may be present inadvertently Although the scope for deliberate addition of other elements, within the context of carbon steel, is necessarily limited, it is of interest to take note of an extensive investigation by Glen ^^^' of the effects of alloying elements on the elevated temperature strength of normalized low-carbon steels Glen reported not only a modification in the strain-aging behavior in the presence of alloying elements, but that the addition of certain elements, including manganese, chromium, molybdenum, tungsten and copper, resulted in an additional strain-aging effect at temperatures higher than that in plain-carbon steel However, multiple strain-aging peaks were not observed in the MPC evaluations of 1/4 Cr-1 Mo steels.f^^^ Rather, the single temperature of peak tensile strength was observed to shift to about ysCF (from about 450 to 500 for carbon steel); also, the peak was only slightly broader than in carbon steel *Assessment of the elevated temperature dependence of yield strength is hampered by the inherent scatter related to: (1) the experimental difficulty of measuring small strains, especially at elevated temperatures; (2) possible nonaxiality of loading; (3) sensitivity to strain rate; (4) possible presence of residual stresses, e.g from cold straightening, or frran specimen preparation; and (5) possible strain-aging susceptibility When yielding occurs with an upper and lower yield point, it is necessary, of course, to use lower yield point, rather than upper, as the equivalent of yield strength Assessments of the elevated temperature strengths of various kinds of carbon steel have also been made in a number of British investigations These were undertaken in the light of the deliberations of the International Standards Organization, with the consequence that attention has focussed upon yield (proof) stress, with little attention devoted to tensile strength, which under the ASME Code criteria frequently limits the allowable stress of ferritic steels At the 1963 International Creep Conference, Glen, Lessells and Barr'- ^ reported the results of a study of one lot of each of four plate steels, all manufactured by the open-hearth process; steel A was semi-killed, steel B was deoxidized with silicon, steel C was deoxidized with silicon and aluminum, and steel D was deoxidized with silicon but treated with niobium The steels had nominally similar carbon (0.14 to 0.16%) and manganese (1.1-1.2%) contents and "similar" room-temperature tensile strengths, 59 to 68 ksi Nitrogen (total?) was 0.005-0.006% in all steels The steels were normalized frcm both 850''C (1562°F), which resulted in a relatively fine microstructure in all four steels, and from 950''C (1742°F), which resulted in a coarsened structure for steels A and B, whereas the two grain size controlled steels retained their fine structure The normalizing treatments were followed by a relatively short (3 hours) or a relatively long (36 hours) stress-relief heat treatment at eOO'C (1112''F), the former simulating normal practice Plots of yield strength vs temperature for each of the four conditions of heat treatment are shown in Figures 5-8 After normalizing followed by the three-hour stress-relief heat treatment, all except the aluminum-deoxidized steel exhibited a relatively strong tendency to strain aging at intermediate temperatures, even to the extent of exhibiting a peak in strength;* the yield strength of the aluminum deoxidized steel decreased progressively in a relatively smooth manner After prolonged stress-relief heat treatment, the behavior of the semi-killed and aluminum deoxidized steels was essentially unaltered; however, the two silicon-killed steels now exhibited a behavior similar to the aluminum deoxidized steel, an effect attributed to immobilization of "active" nitrogen by precipitation as silicon nitride The tendency for leveling of strength in the range 300 to 450°C (575 to 842°F) for the steels not deoxidized with aluminum, evident in Figure 6, was suggested to be related to the presence of "active" nitrogen, probably in association with manganese If this is correct, as seems not unreasonable, then a statement elsewhere in the paper that the effect of grain size becomes negligible above about 300°C (575''F) may deserve further examination, since the true grain size effect might be obscured by differences in active nitrogen content At the 1966 Eastbourne Conference on High Temperature Properties of Steels, Glen, et al^^^) reported on the progress of an analysis of the results of a very large body of elevated temperature tensile tests conducted by the British Steelmakers' Creep Committee Detailed statistical analyses of the results were made, including the derivation of confidence limits for the variation of proof stress with temperature For purposes of evaluation, a comprehensive scheme of classification, with both main- and sub-categories, was adopted as follows: *This behavior appears to be nontypical, and therefore might be viewed with at least a measure of skepticism In steel without prior cold work, the plastic deformation setting the stage for strain aging must be introduced during the tension test, and relatively little stage-setting would be expected by the time the proof stress is reached In carefully conducted tests of representative steels, Miller(^"J did not observe significant reversal of yield strength at intermediate temperatures in susceptible steels and viewed this as expected behavior Nor, were such effects reported in some subsequent British investigations (to be discussed) *Since there are reasons for believing that, under certain conditions, heat treatment may be an important variable, the apparent failure of Glen, et alC^^) to clearly separate data according to treatment, and to take cognizance of heat treatment in Figure 9, detracts from the confidence to be put upon their conclusions It might also be noted that whereas the evaluation methods cannot be faulted, in principle, the results obtained have, in some instances, seemed to be unreasonable, as suggested in reference 14 Main classification: A Semi-killed carbon and carbon manganese steel B Silicon-killed carbon and carbon manganese steel C (a) Aluminum treated carbon and carbon manganese steel, with generally more than a 0.02% soluble aluminum and 0.1 to 0.35% silicon (b) Aluminum-killed carbon steel, with less than 0.01% soluble aluminum and low silicon D Semi-killed, niobium-treated, carbon manganese steel E Silicon-killed, niobium-treated, carbon manganese steel F Rimming steel Subclassification; steel containing: 0.3% up to but not including 0.6% manganese 0.6% up to but not including 1.0% manganese 1.0% up to and including 1.5% manganese, L Low residuals (i.e., less than 0.02% molybdenum) H High residuals (i.e., more than 0.02% molybdenum) However, the analysis of data had by the date of the Eastbourne Conference, 1966, not yet been extended beyond the main classification Most of the steels were in the normalized condition, but the results from a number of hot-finished tubes or as-rolled plates were included The paper is somewhat unclear concerning stress relieval, indicating that "where appropriate", stress relieving was simulated by heating for hours at 600°C after normalizing.* A comparison of mean yield or proof stress vs temperature curves is shown in Figure for four of the main categories of steel for a room temperature tensile strength level of 28 tons per square inch (53 ksi) It is of interest to note that the semikilled and silicon-killed steels are essentially identical throughout the temperature range (and perhaps therefore might have been joined in a common yield strength category) It may also be of interest to note that none of the curves show the reversal in trend reported by Glen, Lessells Barr (13) for other than aluminum-treated steels Figure The authors suggested that in the normalized condition most of the nitrogen in the aluminum-treated steel was immobilized as aluminum nitride, and that this explained why the alumintmitreated steel exhibited a greater falling-off of strength with temperature than did the steels not aluminum-treated The studies of the British Steelmakers' Creep Committee resulted in 1964 in British Standard 1501, covering "Steels for Fired and Unfired Pressure Vessels", including (after ref 8) the types and grades of plates depicted in Figure 10; the grade designations define the specified minimum tensile strength in tons per square inch; grade 28 corresponds to 63 ksi, grade 32 to 72 ksi Note should be taken of several cooperative American investigations, one reported by Miller^^'^J and previously cited, one reported by Worth(-53) ^^j one by Spaeder '• ' In addition to the observations concerning the interplay between stressrelief heat treatment and deoxidation practice in determining strain^aging susceptibility, already mentioned Miller concluded that the tensile strengths of the plate and pipe steels that were studied were comparable at 800 and at lOOO'F (i.e., above the range of dynamic strain aging), whether made to fine-grain or coarse-grain practice, and whether as-rolled or stress-relieved (This has been indicated also by the ratio trend curves of the MPC evaluation Figure 3.) Miller also chose to depict yield strength of either the plate or pipe materials as decreasing uniformly with temperature; Figure 11 shows the results for plate A final observation of interest was that the tensile ductility of the plate and pipe heats made to coarse-grain practice was appreciably higher than that of the heats made to fine-grain practice, whether they were in the as-rolled or stressrelieved conditions Spaeder^^^^ reported the results of a limited cooperative comparison of two grades of carbon plate steel (specified minimum tensile strengths of 55 and 70 ksi), one heat of each made by the basicoxygen process, the other by the open hearth process The results of the investigations together with other available information led to the conclusion that the basic-oxygen steels in the hot rolled condition and in the normalized and tempered condition have about the same tensile properties in the range 80 to lOOO'F aS open hearth steels, and possess similar strain-aging characteristics Note might also be taken of recent interest in the use of accelerated cooling for enhancement of the elevated temperature strength of carbon steel As previously mentioned ASTM A537, introduced in 1965, provides that Grade B will be quenched and tempered At the 1966 Eastbourne Conference, Lessells and Barr^^^^ reported substantial increases in elevated temperature yield strength of carbon-manganese steel relative to the normalized condition, and suggested considerable potential for such steels at temperatures up to at least SSO'C (662°F) in plate thicknesses not over 1/2 inches Basic Investigations As indicated ejirlier in this review, it is difficult or impossible to draw inferences concerning the effects of individual variables of composition or structure from studies of commercial materials, such as those that have been described, owing to the likelihood of simultaneous variation of significant variables, and in many instances to a lack of information about some of the factors of recognized importance For these reasons, an investigation such as that reported upon by Irvine, Murray and Stone,1°^ in which a number of individual variables are isolated, is of particular interest The investigations of Irvine, et al, directed towards the development of an improved boiler plate steel, included study of the effects of nitrogen, molybdenum, vanadium and chromium The important effect that nitrogen exerts upon creep strength of C-Mn steels had been established earlier and will be discussed in a later section of this review Mention should also be made of an earlier study of carbon-manganese-nitrogen alloys by Baird and Jamieson(l^) which revealed that nitrogen caused important but complex strengthening effects Illustrative of their findings are the results shown in Figure 12 The results for Fe and Fe-Mn were obtained on material from which carbon and nitrogen had been removed by annealing in moist high purity hydrogen Since a prime objective of the work of Irvine, et alf^) was to develop a relative improvement in elevated temperature proof stress at a constant level of room temperature tensile strength, the authors chose to analyze their results in terms of the ratio of those two quantities (PS/TS) This basis of analysis has, of course, certain deficiencies when the tensile strength level also changes, as it might reasonably be expected to Nor is the ASME Code interest served when the elevated temperature tensile strength values are not considered The effect of nitrogen was investigated by adding increasing amounts of aluminimi to a steel containing 0.12% C,1.5% Mn and 0.018% N, the effect of the aluminum being to reduce the effective (or free) nitrogen The heat treatment was not revealed but presumably involved normalizing Figure 13 indicates a progressively more rapid fall off of PS/TS with increasing aluminum, that is, decreasing free nitrogen.* The effect of molybdenum was investigated in two different experiments In the first, 0.25% Mo was added to an aluminum-killed steel having 0.12% C, 1.3% Mn, 0.14% Si, 0.02% Al, and 0.005% N As shown in Figure 14, the addition was without substantial effect on PS/TS at about 200°C (392"'F), but an important and progressively greater effect was evident with increasing temperature Presumably, the addition also resulted in increased levels of yield and tensile strengths In a second experiment, molybdenum and also vanadium and chromiimi, as well, were added individually to a high-nitrogen, silicon-killed steel containing 0.12% C, 1.3% Mn, 0.15% Si and 0.03% N The results are given in Figure 15 In this base composition, in which the *The value of PS/TS at room temperature in Figure 13 seems high relative to general experience! proof stress seems to be nearly constant in the range from 300 to more than 450°F, increasing molybdenum is shown to have a beneficial effect; note that as little as 0.08% exerts an important influence, especially at about 300°C Chromium had less effect on PS/TS than molybdenum, especially at the higher temperatures The deleterious effect of vanadium was suggested to be an indirect effect, resulting from the immobilization of the nitrogen by vanadium Again, the lack of information concerning the effects of the several elements upon tensile strength at room temperature, and at elevated temperature, is regrettable The effect of nitrogen in the range 0.0080.033% added to a silicon-killed steel (in which nitrogen would be expected to be more effective than in aluminum-killed steel) containing 0.1% C, 1.2% Mn, 0.2% Si and 0.25% Mo is shown in Figure 16a for the normalized condition and in Figure 16b for the normalized and stress-relieved condition (3 hours at 600°C [1112''F]) For the normalized condition, as little as 0.013% nitrogen improves PS/TS, and further nitrogen addition has only a minor further effect Heating for stressrelief results in some loss of effectiveness of the nitrogen in raising PS/TS, owing presumably to partial immobilization of the nitrogen by combination with silicon Irvine, et al(^) concluded from the results of the experiments described above that an economically practicable boiler-plate steel having improved strength could be achieved by the addition of molybdenum and nitrogen to a fine-grained steel, but that to insure that the nitrogen was present in an effective form, grain refinement would be better achieved by employing niobium, which does not have the strong affinity for nitrogen that aluminum does A further advantage of employing niobium is that such a steel can be produced in the semikilled ("balanced", in British terminology) condition with consequent improved ingot yield As evident in Figure 17, such a steel does exhibit, in either the normalized or normalized and stressrelieved condition, improved PS/TS relative to semi-killed, niobium-treated steel (1501-213) The superiority of the C-Mn-Mo-N-Nb steel was confirmed by tests of an ton commercial heat Creep and Rupture Strengths The scatter of creep and rupture strengths of nominally identical carbon steel is awesome the recent MPC evaluation^^J showed the minimum (90% confidence level) stress for a secondary creep rate of 0.01% per 1000 hours to lie more than 40% below the mean strength Although problems related to test procedure contribute to scatter, it seems well established that large differences in creep and rupture strength of carbon steel are to be associated in part with differences in composition and manufacturing practice within the limits possible under current purchase specifications.l^^>26) Thus, if this degree of scatter is to be reduced, and if advantage is to be taken for heats of carbon steel having superior strength, it is necessary to isolate and understand the effects of the variables of chemical composition and processing that determine strength Efforts directed towards developing such understanding have been underway for some four decades since creep was recognized to be of engineering importance Most of the early interest centered on creep strength, as opposed to rupture strength, and it became evident in time that substantial variations in strength were to be associated with differences in type of melting furnace, and deoxidation and heat treatment practices Efforts at understanding have been hampered by the problems inherent in assessing the long term strength of interest for many applications, from tests of relatively short duration Some of the early tests were as short in duration as 24-48 hours and the results of such tests and of some longer tests, as well, must be viewed with a measure of skepticism, since it is a well-known, but seldomvoiced observation, to those who conduct creep tests, that the creep-time curve may exhibit both irregularities and irreproducibility in its early stages (See for example Freeman's discussion to reference 10, and also Spaeder.^^) A number of reviews of the early studies have been published (e.g see references 17 and 18), and no useful purpose would be served by again reviewing them here in detail Effect of Deoxidation Practice The early studies had shown that steels deoxidized with aluminum using amounts greater than about pounds per ton of steel, and hence having relatively high residual (uncombined) aluminum, might generally be expected to exhibit low creep strength However, such steels were found to exhibit creep strength as great as that observed for silicon-killed steels, provided that the aluminum-killed steels were austenitized above their coarsening temperatures, and cooled fairly rapidly.(^^) In a similar vein, Millerl^^J observed no difference in 1000°F rupture strength (10,000 hours) between steels made to coarse-grain and to fine-grain practice when the steels were tested in the as-rolled or stress-relieved condition.* Miller also reported similar creep strengths for these types of steels when assessed at creep rates faster than about 0.001% per hour, but for lower rates, the steel produced to finegrain practice crept more rapidly; however, in discussing Miller's paper Freeman, who had tested some of the steels reported upon by Miller, cast some doubts on the observed behavior at low creep rates by reporting that at low stresses, both the silicon and aluminum-killed steels had exhibited erratic behavior with alternately high and low creep rates as the tests progressed Other early studies had shown that steels produced in the electric arc furnace or by the Bessesmer process generally exhibited greater creep strength than steels produced by either the basic or acid open hearth processes However, the basic underlying fundamentals remained elusive No single deoxidation practice nor heat treatment was discovered which would always insure high (or low) creep strength Understanding remained in a confused state until the mid-fifties, when a central role for active nitrogen was first suggested, a view that in subsequent elaboration has come to be widely accepted as providing a means of rationalizing the experimental observations Thus, *From the grain size, it seemed clear that Miller's "fine-grain" steel had been finished-rolled at a relatively high temperature Not always recognized, the actual grain size of steels in the hot rolled condition depends upon the finishing temperature and may be unrelated to the deoxidation practice 500 AL killed steel (2lb/ton) •^100- •o in UJ K a 10 UJ UJ o N 920*0 (lees^F) J- 0-4 0-8 Mn,% 1-2 FIG 19 — Effect of manganese on the creep rate of aluminumkilled and silicon-killed carbon steel under a stress,of.8 tons per square inch (18 ksi) at 450°C (842°F) Glen^ ^ 31 100 Si killed steel ( - - % Si) 50 30 l-l7i>Mn in V) Ui N900-920OC Q: 10 -- es o E •o Q «, "- * ^ID O - ^ o o G G G > • -u •H Q: G ^ G o o CO o cr o ora — G(% G G O o o o o o o 00 ^ oIS> o ^ oCVJ o o o o" o Q- a, S o o o o" o ° - = o O D < n o m Q) B •iH J-i ^ 4J •H o o Q o © o a T3 hUJ o o 4J •H iH •H •U O Q & LU a: Q s (- CU > ^ 4-1 VJ 4-1 i H O^-' o g 4-1 cfl •r-l PM iJ o CO O > u-l O H 1 O B o El Q El rj o 1— Q ZD QC O in • a: X B El to Or ^ (U n s-i d • H P=M < C3 Q o O O El o o 4-1 o M o3 S J o 00 L o o 'J- iN33 a3d 45 o CM ^