I

s?m9%z32zc

Austeriite Formation Kinetics DuringRapid Heating in a Microalloyed Steel

J. D. P~skar+,R. C. Dykhuizen+, C. V. Robino+,M, E. Burnett$, J. B. Kelleyt

‘Sandia National LaboratoriesAlbuquerque, NM 87185

‘The Timken CompanyCanton, OH 44706

Key Words: Austenite, Microalloy, On HeatingKinetics, Model, Gleeble, Dilatometry

INTRODUCTION

The formation of austenite at rapid heatingrates is an important aspect of many metallurgicalprocessing and fabricating, schemes for steels. Forexample, hot working, heat treating, and welding allrequire or result in heating into the austenite oraustenite plus ferrite phase fields. At the presenttime, there is widespread interest in modeling theseprocesses as an aid in optimization and control ofpost-process microstructure, properties anddistortion, Additionally, these models will enablesteel producers to identifi optimized compositionsand microstructure for various classes ofapplications. For these models to be applicable,they must describe the phase transformation kineticsassociated with both the on-heating and on-coolingtransformations, and these descriptions must beexperimentally validated, In general, the formationof austenite in steels has received less attention thanthe decomposition of austenite, although there havebeen a number of experimental[l-G] and numerical

‘studies[7-’2]of the process. These studies haveyielded significant insight into the transfoqnationfrom both mechanistic and computational

Ree..pgnperspectives, but there are@$n “mita - and

iU&@difficulties in applying this si t6

ge scaleprocess modeling. *“

s Tj

As discussed by Gavard et al.[’~, and Akbayet al.‘9]the formation of austenite differs from itsdecomposition in two principal ways. First, in thecase of diffision-limited on-cooling transform-ations, the driving force for the reaction increaseswith increasing undercooking below the equilibriumtransformation temperature, while difision ratesdecrease with increasing undercooking. The balancebetween driving force and diffusion rates results inthe classical C-curve kinetic behavior, in which theoverall transformation rate experiences a maximumat intermediate undercoolings. In contrast, for theon-heating transformation, both the driving forceand diffision rates increase with temperature abovethe equilibrium transformation temperature, so thatthe rate of transformation continuously increaseswith temperature; Also, for the on-coolingreactions from homogeneous austenite, the kineticscan be fully described in terms of the compositionand austenite grain size. However, such asimplification is not possible for the formation ofaustenite since a wide variety of starting micro-structure are possible.

Recently, a numerical model was developedto describe the formation of austenite in pearlitichypoeutectoid plain carbon steels.[*4] The modelextracts the austenite formation kinetics directlyfrom dilatation data for continuous heatingexperiments. A low carbon steel, 1026, was used todevelop and validate the model which is based on atwo step description of the process. The first steputilizes a classical description of the decompositionof the pearlite constituent, while the second stepuses a one-dimensionaI diffbsion model to describethe growth of the austenite into the proeutectoidferrite.

Microalloyed steels are an important class ofsteels that are similar to plain carbon steels, but theytypically ~ontain small ~o~ts of strong c~bideforming elements such as niobium, vanadium and

DISCLAIMER

This report was prepared as an account of work sponsoredby an agency of the United States Government. Neither theUnited States Government nor any agency thereof, nor anyof their employees, make any warranty, express or implied,or assumes any legal liability or responsibility for theaccuracy, completeness, or usefulness of any information,apparatus, product, or process disclosed, or represents thatits use would not infringe privately owned rights. Referenceherein to any specific commercial product, process, orservice by trade name, trademark, manufacturer, orotherwise does not necessarily constitute or imply itsendorsement, recommendation, or favoring by the UnitedStates Government or any agency thereof. The views andopinions of authors expressed herein do not necessarilystate or reflect those of the United States Government orany agency thereof.

- ,.:,m -— .,-,. .- ..

DISCLAIMER

Portions of this document may be illegible

ih electronic image products- images areproduced from the best availabie original

document.

.,

- ,.. ., ,..-.., , .-/>, .~,..... .. , ,-,,....-,... ., ... ....t..—- ,. .,, >-. —

f

titanium. The carbide formers serve to addprecipitation strengthening, grain refinement andcontrol over the transformation temperatures, whileadding minimal additional raw material cost.Depending on the processing route, these steelshave microstructure that are similar to the plaincarbon steels (e.g. ferrite-pearlite). Given thesimilarity, the purpose of the present work was toevaluate the extent to which austenitization modelsdeveloped for plain carbon steeIs are applicable tothe microalloyed steels. Applying the low carbonsteel model to the transformation kinetics ofTimken 1054V1 seamless steel tubing will be thefocus of this paper. Timken 1054V1 alloy is anexperimental composition being developed forvarious applications requiring a surface hardness of60 HRC. Of particular interest is examining bothas-pierced and normalized tubing. The twoconditions were made from the same heat andtherefore have the same composition, but themicrostructure vary significantly in terms ofpearlite volume fractions, prior austenite grain sizes,and potentially, microalloy carbide distributions.

EXPERIMENTAL

Materials and Processing

The heat of microalloyed 1054V1 used inthis study was melted from scrap iron in a 150 tonelectric fi.umace,ladle refined and strand cast into 28

cm by 37.5 cm blooms. The blooms were thenreheated and rolled to M 20.3 cm diameter roundbar, reheated to around 1230°C and pierced into a17.8 cm OD by 1.9 cm wall tube, followed by aircooling on a hot bed. Part of the tubing wassubsequently production normalized by reheating to900°C followed by air cooling. The chemistry ofthe final product can be fowd in Table I. Samplesfrom both the hot rolled (as-pierced) and thenormalized tubing were then sectioned to remove6.35 mm diameter by 127 mm long dilatometryspecimens. The mechanical properties of the tubingin both conditions can be found in Table II.

The rnicrost&cture of the hot rolled tubingis composed of coarse grained pearlite withproeutectoid ferrite at the prior austenite grainboundaries, Figure l(a). The normalized tubing iscomposed of fine grained pearlite with ferrite at theprior austenite grain boundaries, Figure l(b). Therelevant grain size and constituent volume fractionmeasurements can be found in Table III, and it isimportant to note the significant differences in therelative fractions of the constituents.

Dilatometry

To” characterize the phase transformations,dilatometry experiments were petiormed using aDSI Gleeble 1500 therrnomechanical simulator. Ahigh resolution dilatometer was used to measure thediametral dilatation of solid 6.35 mm diameter

Table I Steel chemistry of 1054V1 material used in this study.Element c Mn P s Si Cr Ni Mo Cu Al v NPercent 0.55 0.78 0.01 0.032 0.24 0.14 0.09 0.02 0.17 0.02 0.10 0.012

Table II Mechanical properties of 1054V1 material.

Condition HardnessTensile Yield Percent Percent CVNStrength Strength Elong. RA Impact

BHN HRc MPa MPa JouIes

Hot Rolled 275 25 917 555 14.5 35 11.5

Normalized 240 19 820 510 20 47 20

Table III Microstructural properties of 1054V1 material.Condition Prior Austenite Grain Size Ferrite Pearlite Ferrite Grain Size

ASTM (pm) (percent) (Percent) ASTM (w)Hot Rolled 3.25 (116) 9.5 90.5 11 (8)

Normalized 7.5 (25) 28.4 71.6 11 (8)

-—... ..——.— -—. ... . .

#

I-’g -1

Figure 1. Photomicrographs of Timken alloy 1054V1initial microstructure. The hot rolled (as-pierced)microstructure (a) contains a coarse grain ferrite withproeutectoid ferrite at the prior austenite grain boundarieswhile the production normalized condition (b) contains amuch finer grained pearlite with significantly moreproeutectoid ferrite present.

specimens using the low stress modified dilatometrytechnique, ’15] Specimens in the hot rolled andnormalized conditions were heated with linearlyprogrammed temperature ramp rates fiorn 50°C/s to500°C/s to simulate the heating rates that occurduring rapid thermal cycle processes. The rapidlyquenched specimens were tubular with a 6.35 mmouter diameter and a 4.55 mm inner diameter thatwas internally quenched from the peak temperaturewith helium gas. This produced a cooling rate ofapproximately 250°C/s between 800”C and 500”C.All tests were conducted in an argon atmosphere.

Model

A model was previously developed thatrelates the on-heating phase transformations to

Figure 2. Idealized initial microstructure consisting of aspherical pearlite nodule of radius RI surrounded by a shellof ferrite of radius R=

volume changes that are measured by dilatometry.The main featies of the model are presented belowand additional detail can be found in the work byDykhuizen et al.[14]

The model begins by assuming a singlerepresentative grain that is spherical in geometryand contains a pearlite colony within a shell offerrite, see Figure 2. The grain size is representedby the prior austenite grain size, Rg which is one ofthe req~ed inputs to the model. Another requiredinput for the model is the volume fraction ofpearlite, FP, which is experimentally determinedfrom micrographs. The average size of the pearliteregion within a grain, RI is calculated by:

R, = Rg(Fp); (1)

As the steel is heated above the AC1temperature, the pearlite is assumed to transform toaustenite based on Avrarni kinetics. [”] Thegoverning equations for the reactions are as follow:

“ += (~= (T) – A)zdqzy(””l) (2)

,

where

I,=-’”(Az?iA)’K(T)L J

and

(3)

()

PK(T) = exp CZ,-T (4)

where A is the fraction of pearlite transformed andAmm as the maximum amount of pearlite that cantransform to austenite as a fimction of temperature.h practice, Ammis a step function that equals 100%above AC1and 0°/0below. The reactitm order, n, isassumed to be three based on previous workperformed by Speich and Szirrnae. ‘3] The 0 term, afictitious time, is set equal to the time that isrequired to obtain the current extent of reaction atthe current temperature. The K(T) term is atemperature-dependent rate parameter that in theorycould be determined from a time temperaturetransformation (TTT) curve for eutectoid pearlite.To eliminate the need for a detailed TTT curve foreach alloy, a functional form that is consistent withSpeich and Szirmae[3] is assumed and fit withexperimental data which provides two of the fittingparameters that are used in this model, u and ~.

A separate diffusion-li@ted model is usedfor the surrounding ferrite shell, a modelingapproach that is similar to that used by otherresearchers.[17’

The radial location of the boundary between theuntransformed ferrite and the austenite is denotedby r. The term r can range from RI to R-. Theterm R,nm is calculated to assure that the carbon

content of the resulting austenite is not below theAC3line. The left hand terms represent the carbonconcentrations with cfo being the initial carbonconcentration of the ferrite, ci(T) is the austenitecarbon concentration at the interface, and Ac/ AXisthe carbon concentration gradient at theferrite/austenite interface. The diffusion coefficientof carbon in austenite is D(T)[]8]. A fitting term, y isthe final parameter in the above expression.Essentially y is used to compensate for thedifferences in the actual prior austenite grain shapeand the assumed spherical shape, as well as theassumed simplifications of carbon diflhsion inaustenite.

While the fhnd~ental principals of theDykhuizen et al.[14]model, summarized above, wereused in this study, several changes to the modelwere necessary to simulate the microalloyed steels.The amount of pearlite that forms in a steel issensitive to the cooling rate of the austenite. Inparticular, rapid cooling of austenite can result in alarger than predicted equilibrium volume fraction ofpearlite. The non-equilibrium value of pearlite hasa carbon content that is lower than the eutectoidcomposition. In this study, the effect of cooling ratewas readily seen as the hot rolled samples hadgreater than 90 VOI’%0of pearlite compared with thenormalized that had 71 VOlO/Oof pearlite. Therefore,to properly simulate non-equilibrium pearlitevolume fraction steels, it was necessary to modi~the model.

The calculation of the carbon content of theaustenite that was created fi-omthe pearlite colonieswas modified to account for the non-equilibriumvolume fractions. When the temperature risesabove the & line, austenite is created at theeutectoid carbon content. “ To accomplish this,cementite and ferrite are transformed to auslenite inproportions that leave the untransformed pearlitewith excess ferrite. As the temperature rises, thecarbon content of the austenite reduces as describedby the A= line. The austenite carbon contentcontinues reducing until it reaches the averagecarbon content of the pearlite colonies. Then, thecarbon content of the austenite created from thepearlite remains constant at this value.

-. .:.,,,,, -,.”.1 , f ,.. . ,: .- ;.*.

In the new model, the state variable A nowrepresents the fraction of cementite within thepearlite colonies that has transformed to austenite,not the fraction of pearlite (although in practicethese two quantities are nearly equal). The volumefraction of ferrite in the pearlite colonies that hastransformed to austenite is then determined from theaustenite carbon content as described above.Finally, the proeutectoid ferrite is not allowed totransform until the austenite carbon content hasreduced to the pearlite carbon content eliminatingthe possibility of creating austenite at carboncontents lower than specified by the&3 line.

RESULTS AND DISCUSSION

Samples were tested using highGleeble dilatometry as described above.

resolutionA typical

dilatation curve for the formation of austenite in ahypoeutectoid steel is shown in Figure 3. As thespecimen is heated at a constant rate from roomtemperature, the thermal expansion of the materialis the dominant feature causing the diameter toincrease with a nearly constant positive slope.However, at approximately 735°C the pearlitebegins to form austenite as the Acl temperature isreached. The sharp contraction (vertical drop) inthe data is associated with the formation of austenitefrom the pearlite. The formation of austenite thencontinues as the proeuctectiod ferrite reacts to formaustenite, albeit at a much slower rate, with thereaction continuing to approximately 860”C atwhich point the AC3temperature is reached. Thecompletion of the austenite formation is observed inthe data, as the expansion of the material becomeslinear above approximately 860”C with a slope thatis steeper than that seen in the original ferrite-pearlite specimen. The above interpretation of thedilatation curve is somewhat simplified as there isactually a point where the pearlite is formingaustenite at the same time as the proeutectoidferrite. Detailed analysis of the curves wasperformed using the model presented previously, asit is necessary t~,. use the model to accuratelyanalyze the curve.L’4J

The solid specimensGleeble dilatometry and the

were fust testedresults examined

usingusing

‘:pz~Proeutec\oid’”Ferrite’+wstenite

0.022 t,,l,.,.;.,.,1..,l.,l .,650 700 750 800 8S0 900

Temperature (°C)

Figure 3 Typical dilatometry curve obtained from ahypoeutectoid steel heated at a constant rate. The pearlitein the specimen begins to form austenite once the AC1temperature is reached. The pearlite completes reactingquickly, and the proeut~ctoid ferrite then begins to formaustenite at a much slower rate. The formation of austeniteis complete once the Ad temperature is reached.

the austenitization model. The various samples allhad the same chemistry, but two different prioraustenite grain sizes and volume fractions ofpearlite. Analysis of a series of three dilatometrycurves (at 50, 100, and 200°C/see) for thenormalized material yielded fitting parameters of a= 232, ~ = 236000, and y = 1.93. The experimentalcurves and model fits are shown in Figure 4(a) forAL/L versus time and for AL/L versus temperature.The model fits are very good for the normalizeddat~ implying that the model assumptions arereasonable for the 1054V1 normalized material.

Figure 5 shows the fraction austenite as afunction of temperature for the normalized materialheated at 10O°C/see, and shows the relativecontributions from the pearlite and ferriteconstituents. “ Fi@re 5 illustrates the rapiddecomposition of the pearlite that is associated withthe short diffusion distance (approximately half thepearlite interlamella spacing). The longer tailassociated with the completion of thetransformation of the ferritic regions is moregradual since the temperature has to txaverse thetwo phase region between the AC1and aAC3lines.

Because three fitting parameters are used inthe model, a good fit to one set of dilatation curves

-.. —... %-,. , . . . ,. -.. -. . .-. >...L .,, ,

0.014

0.013

50.012

#:0.011

0.01

0.009

5m=:

100OC/sec 50OC/sec -

.

. (a) ~I I I I

o 1 2 3 4 5 6Time (see)

700 750 800 850 900 950Temperature (“C)

Figure 4. Comparison of normalized experimentaldilatation data (solid) to model predictions (dashed) forthree different heating rates: (a) time dependence and (b)temperature dependence. For clarity, the temperaturedependent curves are displaced in ALIL.

1

0.8a ~------------------ ----------,.=c Pearlite+ 0.6

,’,,

2 !

.~ 0.4

f!L

‘U I

--------------0.2 -----------------

P FerriteI054v1 Steel, 100OCLs

o ~~ 1 I I ! I

740 760 780 800 820 840 860Temperature (“C)

Figure 5. Volume fraction of austenite formed as afunction of temperature for the normalized material heatedat the 10O°C/s heating rate. The relative contributionsfrom the initial ferrite and pearlite phases are also shownas predicted by the model.

is not necessarily indicative of its generalapplicability. Additional experimental trials weretherefore conducted at heating rates both within andoutside of the range used to develop the parameterset. The data from these trials was then comparedwith calculated results obtained by using the’parameters developed from the initial experiments.The results of these comparisons are shown inFigure 6, and the calculated curves are in goodagreement with the experimental data. Thus, thefitting parameters obtained with one set of dataworked quite well with other sets of data for heatingrates within the initial range, and extrapolatedreasonably to a heating rate of 500°C/s. Using theparameters for one set of data to accurately predictanother set of data confirms the experimentalrepeatabiliv and suitabili~ of the parameter set for

normd;zed material. - The prediction for the

0.014

0.013

0.012

5

0.011

0.01

0.009

5g

8

1 1 , , , f

Ff;Oc’secL7!

tb!l!!!!t!o! ! !.!l!!!!l!!.l(a),.0 1 2 3 4 5 6

Time (see)

I I I 1

I I I I

700 750 800 850 900 950

Temperature (“C)

Figure 6. Comparison of experimental dilatation data(solid) to model predictions (dashed) for a different seriesof three normalized tests.: (a) time dependence and (b)temperature dependence. The plots show very good modelprediction for the set of new specimens. For clarity, thetemperature dependent curves are displaced in AL/L.

500°C/s transient, though quite reasonableconsidering the rapid cycle, is not as well predictedas the other transients. The reasons for this reducedaccuracy are not filly clear, but may be related tothe mobility of the alloy elements in themicroalloyed steel. As discussed below, the effectof these elements is not explicitly described in thecurrent model, but may affect the kinetics thoughtheir slower (than carbon) difision rates andpartitioning to the various phases.

The fit of the normalized 1054V1 datayielded parameters that are similar to those obtainedin the original study of 1026 steel[i4] (u = 201, ~ =200000,and y = 2.23). This implies that thetransformation mechanisms are similar for the twosteels (and their respective starting microstructure).It is interesting to compare predictions for the1054V1 steel based on the parameter set developedfor the 1026 steel. Figure 7 shows these predictionsand indicates that though the predictions arereasonable, they are less accurate than the fitsobtained using the 1054V 1 parameters as shown inFigure 3. Stated differently, the curves of Figure 7illustrate the expected transformation behavior for1026 steel given the same thermal cycle as the1054V1 steel. For the three heating rates, the 1026parameters generally predicted a faster reaction,Figure 7(a), at lower start temperatures, Figure 7(b)for the formation of austenite than what isexperimentally observed in the 1054V1. Althoughthese figures are complex and difticult to interpretdirectly because of the thermal arrest during thetransformation, [’4>]9]it is reasonable to conclude thatthe austenite formation kinetics of the 1054V1 areretarded relative to the 1026 steel. There are severalpossible sources for the differences in the kinetics.First, the current model does not explicitly includepearlite spacing. Differences in the pearlite spacingbetween the original 1026 and the normalized1054V1 could affect the dissolution rate of thepearlite, In addition, it is also likely that differencesin the kinetics can be attributed to the rnicroalloyadditions. With respect to the pearlite decom-position, partitioning of the carbide formingelements to the carbide within the pearlite mi~t beexpected to kinetically retard decomposition of thepearlitic carbide. Moreover, the alloying elementswould be expected to influence the diffusion rate of

0.014

0.013

0.012

5g 0.011

80.01

0.009

0.008

5g

:

L“’’’’’’’’’’’’’’’’’’’’’’” ““-lL J

50 OC/sec :

“. .. ... . ... (a) j

I I 1 I Io 1 2 3 4 5 @ 6

Time (see)

I I 1 1

I 0.001

..

,.

.................

I 1 I(b) :

700 750 600 850 “ 900 950

Temperature (“C)

Figure 7. Comparison of normalized experimentaldilatation data (solid) to model predictions (dashed), wherethe parameters ftom the original study of 1026 data wereused (a) time dependence and (b) temperaturedependence. The 1026 parameters generally show a morerapid rate and lower transformation start temperatures forthe 1026 parameter model predictions. For clarity, thetemperature dependent curves are displaced in ALL.

carbon,pearlite

thereby tiecting the dissolution of theas well as the dissolution of the

proeutectoid ferrite.

Model fits for microalloyed 1054V1 hotrolled steel are shown in Figure 8. The model givesreasonable representation of the data, and the fitparameters for this condition were found to be u =378, ~ = 392000, and y = 0.59. Due to the lowfraction of ferrite in the hot rolled condition, TableIII, the y parameter has little impact on the model

.—— ... ~---- .- ~rm >,=, ,~> -,P=-?, ..,,:,, W,L,*.,,. .+. OK=--=--- ->. 7.- :. ..?... . .. ,, *. -:. .s ,. .=. .- —.. . . .. . . --=---- ‘- ‘— -- .-

0,014

0.013

:0.012

=80.011

0.01

0,0090 1 2 3 4 5

Time (see)

I I I I

: ~o.ool

rlt I I I I 1700 750 800 850 900 950

Temperature (“C)

Figure 8. Comparison of hot roiled experimental dilatationdata (solid) to model predictions (dashed) for threedifferent heating rates: (a) time dependence and (b)temperature dependence. For clarity, the temperaturedependent curves are displaced in AL/L.

and can vary appreciably from 0.59 withoutsignificantly altering the results. However, evenneglecting the third parameter, the first twoparameters are also significantly different than thoseobtained from the normalized material. This can beillustrated by using the normalized parameter set topredict the hot rolled response. The comparisonshown in Figure 9 was accomplished by inputtingthe correct hot rolled grain size and pearlite fractionand using the normalized kinetic parameters of a =232, ~ = 236000, and y’= 1.93. The fit in Figure 9is quite inaccurate, with the actual austeniteformation occurring much slower in the hot rolledmaterial than is predicted by the model for the

0.014

0.013

0.012

5go.oll

:0.01

0.009

0.008.,

her’’’’’’’’’ ”’’’’’’’’””i300OC/secJ

(a) !I I I I

o 1 2 3 4 5Time (see)

I I I I I I

,.-”

.3:Isec/w ‘wsec J

::------------

,, I I I

-ST-l

700 750 800 850 900 950

Temperature CC)

Figure 9. Comparison of normalized parameters used topredict (dashed) the experimental hot rolled dilatation data(solid): (a) time dependence and (b) temperaturedependence. The, fit is very inaccurate with the actualaustenite formation occurring much slower in the hot rolledmaterial than is predicted with the normalized parameters.For clarity, the temperature dependent curves are displacedin ALL.

normalized material. For the two microstructuralstarting conditions, the major differences are in thefraction pearlite and grain size (although there mayalso be fine scale differences in terms of microalloycarbides). For the hot rolled material, the fractionof pearlite is significantly different than would beexpected at equilibrium. However, the currentmodel accounts for this discrepancy as described inthe modeling section. The formation of non-eqtilibrium pearlite fractions occurs when thematerial is transformed to pearlite and ferrite attemperatures significantly below the equilibrium

transformation temperature (as in the hot rolledmaterial), This has the effect of altering both thepearlite spacing and the relative thickness of thecarbide and ferrite larnella within the pearlite. Thecurrent model does not capture these differencesexplicitly, but they are captured indirectly in theparameters associated with the pearlitedecomposition, Therefore, it is perhaps notsurprising that different fit parameters wereobtained for the two starting microstructure.

The above comparisons reinforce theintroductory comments regarding the relativecomplexity of austenite formation in steels, anddemonstrates the sensitivity of the process to thestarting microstructure. They also illustrate thepotential difilculties associated with development ofa general model for the on-heating transformations,since such a model must account either explicitly orimplicitly (as in the current model) for the detailsand range of starting microstructure. Moreover, itwas generally observed that the austenite formationwas experimentally more repeatable in thenormalized material than in the hot rolled. Thisdifference is consistent with typical shop floorexperience, that more repeatable results areobtained from normalized material duning rapidthermal cycle processes.

Direct verification of an on-heating modelduring a rapid thermal cycle is extremely difficult.Techniques such as hot stage microscopy areneeded to directly view the formation of austenitesince austenite transforms to other phases whencooled. Unfortunately, heating rates in this studyare orders of magnitude faster than that availablewith hot stage microscopy. Another complicatingfactor is the very short time required for the reactionto complete; at most two seconds in the currentrange of heating rates. Thus, indirect methods tovalidate the model are required. In this study,experiments were performed on tubular specimensthat were partially austenitized. A 1054V1normalized tubular specimen was heated at alinearly programmed rate of 50°C/s to a fixedtemperature of 785”C. The specimen wasimmediately quenched internally with helium gas toroom temperature in an effort to transform as muchof the austenite to martensite as possible. The

quench rate was approximately 250°C/s between800°C and 500°C. Figure 10 shows the thermalcycle and measured dilatation for the specimen.The actual temperature cycle displays the typicaldeviation from a linear ramp around 720 ‘C. Thedeviation is due to the endothermic reaction ofpearlite transforming to austenite. The Gleeble1500 control algorithm attempts to, but does nottotally, eliminate the arrest.[lg]

After quenching, the tubular specimen wassectioned and etched in a 2°/0 nit.al solution formicrostructural examination. Optical microscopywas performed to examine the amount of remainingferrite. Figure 11 is an oil immersion image that

800

750

c-l0g 7(J133~alQ 650E

#’

600

I t I

,t

\.,

\ 2

0.025

0.024

0.023

0.022 ~m

0.021 5c

0.02 r

0.019

0.016\

550 t I I I ., + 0.0170 0.5 1 1.5 2

Time (see)

Figure 10. Measured thermal cycle (solid) and dilatationdata (dashed) for a hollow normalized specimen heated to785°C and quenched internally with helium gas.

Figure 11. Optical micrograph after quenching of thenormalized tubular specimen.

. ,

1

0.8

0.2

0 J 0.018-1 -0.5 0 0.5 1 1.5

Time (see)

Figure 12 Predicted fraction transformed and measureddilatation as a function of time for the normalized tubularquenched specimen.

shows a typical cross section of the partiallytransformed specimen. Blocky white areas ofundissolved ferrite can be seen in the structure. Inaddition, the structure appears to be a mixture ofmartensite, gray regions, and upper transformationproducts (bainite or very fme new pearlite), darkregions, Quantitative image analysis of severalmicrographs was attempted, but it was difficult toisolate the ferrite and produced somewhatunpredictable results. However, point count resultsof 20 fields, 50 counts per field, resulted in anaverage of 6.05°/0ferrite.

The actual temperature transient of thetubular specimen was used as input for the model.Predictions of the fractions of pearlite andproeutectoid ferrite transformed to austenite werethen calculated. The predictions as a fhnction oftime are plotted in Figure 12, along with the actualdilatometer data from the experiment. Figure 12shows that all of the pearlite is predicted totransform to austenite. However, the amount offerrite predicted to transform is only 78.3%.Therefore, the final volume of ferrite predicted is78,3% of the amount of ferrite initially present,which was 28.4°/0 of the total voh.une. Thus thetotal remaining volume fraction of ferrite ispredicted to be 6.2%, a value very close to themeasured value of 6.05°/0. .

..

CONCLUSIONS

The model parameters for the normalized1054V1 material were compared to parameterspreviously generated for 1026 steel, and thetransformation behavior was relatively consistent.Validation of the model predictions by heating intothe austenite plus undissolved ferrite phase field andrapidly quenching resulted in reasonable predictionswhen compared to the measured volume fractionsfrom optical metallography. The hot rolled 1054V1material, which had a much coarser grain size and anon-equilibrium volume fraction of pearlite, hadsignificantly different model parameters and the onheating transformation behavior of this material wasless predictable with the established model. Thedifferences in behavior is consistent withconventional wisdom that normalized micro-structure produce a more consistent response toprocessing, and it retiorces the need for additionalwork in this area.

ACKNOWLEDGEMENTS:

Sandia is a multiprograrn laboratory operated bySandia Corporation, a Lockheed Martin Company,for the United States Department of Energy underContract DE-AC04-94AL85000

1.

2.

3.

4.

REFERENCES

G. A. Roberts and R. F. Mehl, The Mechanismand the Rate of Formation of Austenite fromFerrite-Cementite Aggregates, Trans. ASM,VO1.31, pp. 613-650, (1943).

E. S. Davenport and E. C. Bain, Trans AIME,vol. 90, pp. 117-131, (1930).

G. R. Speich and A. Szirrnae, Formation ofAustenite from Ferrite and Ferrite-CarbideAggregates, Trans. AIME, Vol. 245, pp. 1063-1069>(1969).

G. Molinder, A Quantitative Study of theFormation of Austenite and the Solution ofCementite at Different AustenitizingTemperatures for a 1.27% Carbon Steel, ActsMetall., Vol. 4, pp. 565-571, (1956).

.

5.

6.

7.

8.

9.

10.

11.

12.

13.

. t

R. R, Judd and H. W. Paxton, Kinetics ofAustenite Formation from a SpheroidizedFerrite-Carbide Aggregate, Trans. A.IMIi, Vol.242, pp. 206-215, (1968).

C. I. Garcia and A. J. DeArdo, Formation ofAustenite in 1.5 Pet Mn Steels, Met. Trans.A,Vol. 12A, pp. 521-530, (1981).

P. A. Wycliffe, G. R. Purdy, and J. D. Embury,Growth of Austenite in the IntercriticalAnnealing of Fe-C-Mn Dual Phase Steels,Canadian Metallurgical Quarterly, Vol. 20, pp.339-350, (1981).

D, F, Watt, L. Coon, M. Bibby, J. Goldak, andC. Henwood, An Algorithm for ModelingMicrostructural Development in Weld Heat-Affected Zones (Part A) Reaction Kinetics,Acts Metall., Vol. 36, No. 11, pp. 3029-3035,(1988).

T. Akbay, R. C. Reed, and C. Atkinson,Modeling Reaustenitisation fromFerrite/Cementite Mixtures in Fe-C Steels,Acts Metall. Mater., Vol. 47, pp. 1469-1480,(1994).

C. Atkinson, T. Akbay, and R. C. Reed, Theoryfor Reaustenitisation from Ferrite/CementiteMixtures in Fe-C-X Steels, Acts Metall.Mater., Vol. 43, pp. 2013-2031, (1995).

C. Atkinson and T. Akbay, The Effect of theConcentration-Dependent Diflbsivity ofCarbon in Austenite on a Model ofReaustenitisation from Ferrite/CementiteMixtures in Fe-C Steels, Acts Mater., Vol. 44,pp. 2861-2868, (1996).

T. Akbay and C. Atkinson, The Infl~ence ofDiffhsion of Carbon in Ferrite as well asAustenite on a Model of Reaustenitization fromFerrite/Cementite Mixtures in Fe-C Steels, J.Materials Science, Vol. 31, pp. 2221-2226;(1996).

L. Gavard, H. K. D. H. Bhadeshia, D. J. C.MacKay, and S. Suzuki, Bayesian Neural

14.

15.

16.

17.

18.

19.

Network Model for Austenite Formation inSteels, Materials Science and Technology, Vol.12, pp. 453-463, (1996).

R. C. Dykhuizen, C. V. Robino, and G. A.Knorovs@, A Method for Extracting PhaseChange Kinetics from Dilatation for MultistepTransformations: Austenitization of a LowCarbon Steel, Metallurgical and MaterialsTransactions B, Vol. 30B, pp. 107-117, (1999)

J. D. Puskar, manuscript in preparation.

J. W. Christian: Tke Theory of PhaseTransformations in Metals and Alloys, Part I,Tergamon Press, New Your, NY, 1981 pp.525-48

A. S. Oddy, J. M. J. McDill, and L. Karlsson,Microstructural Predictions Including ArbitraryThermal Histories, Reaustenization and CarbonSegregation Effects, Can. Met. Quarter., Vol.35, pp. 275-283, (1996).

E. A. Brandes, Smithells Metals ReferenceBook Sixth Ed. , Butterworths, London, p. 13-58, (1983).

C. V. Robino, G. A. Knorovsky, R. C.Dykhuizen, D. O. MacCallum, and B. K.Damkroger, Transformation Kinetics inControlled-Power and Controlled-TergperatureCycle Testing, Accepted for Publication inProceedings of the 5thInternational Con.t?erenceon Trends-in Welding Research,cd., Pine Mountain, GA, (1998).

S. A. David,

For further information on this paper, please contactJoseph D. Puskar at Sandia National Laboratories,PO Box 5800, MS 0367, Albuquerque, NM 87123-0367 or jdpuska@sandia.gov