Sie sind auf Seite 1von 9

Available online at www.sciencedirect.

com

Acta Materialia 56 (2008) 1482–1490


www.elsevier.com/locate/actamat

Finite element simulation of quench distortion in a low-alloy


steel incorporating transformation kinetics
Seok-Jae Lee a, Young-Kook Lee b,*
a
Research Institute of Iron and Steel Technology, Yonsei University, Seoul 120-749, Republic of Korea
b
Department of Metallurgical Engineering, Yonsei University, Seoul 120-749, Republic of Korea

Received 17 October 2007; received in revised form 22 November 2007; accepted 27 November 2007
Available online 22 January 2008

Abstract

The uncontrolled distortion of steel parts has been a long-standing and serious problem for heat treatment processes, especially
quenching. To get a better understanding of distortion, the relationship between transformation kinetics and associated distortion
has been investigated using a low-alloy chromium steel. Because martensite is a major phase transformed during the quenching of steel
parts and is influential in the distortion, a new martensite start (Ms) temperature and a martensite kinetics equation are proposed. Oil
quenching experiments with an asymmetrically cut cylinder were conducted to confirm the effect of phase transformations on distortion.
ABAQUS and its user-defined subroutines UMAT and UMATHT were used for finite element method (FEM) analysis. The predictions
of the FEM simulation compare well with the measured data. The simulation results allow for a clear understanding of the relationship
between the transformation kinetics and distortion.
Ó 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Finite element method; Martensitic transformation; Transformation kinetics; Distortion; Low-alloy steel

1. Introduction influence of transformation plasticity on the distortion of


a carburized ring specimen [4]. Arimoto et al. [5] have
Heat-treating processes have traditionally been used to explained the origin of distortion and the stress distribution
greatly enhance the mechanical properties of steel parts in quenched cylinders by accounting for the phase transfor-
such as bearings, gears, shafts, etc. Unfortunately, heat mation. Ju et al. [6] have studied the martensitic transfor-
treatments such as carburizing, quenching and tempering mation plastic behavior during quenching.
often cause excessive and uncontrolled distortion. This Because martensite is the major phase produced during
type of distortion is still a major issue in the production the quenching of the steel parts, a reliable prediction of the
of quality parts. Many research groups have examined martensitic transformation kinetics is indispensable for the
the causes of distortion and found that the phase transfor- computational simulators of the distortion such as
mations as well as thermal stresses that occur during the HEARTS [7], SYSWELD [8], DEFORM-HT [9], DANTE
heat treatment play an important role. [10] and COSMAP [11].
Denis et al. [1,2] have investigated the effects of stress on Koistinen and Marburger’s equation [12], dating from
the phase transformation kinetics and transformation plas- 1959, is still widely used for the prediction of martensite
ticity. Inoue et al. have studied the relation between phase kinetics. Their equation was obtained by fitting the mar-
transformations and residual stresses [3], as well as the tensite volume fraction, measured by X-ray diffraction, as
a function of temperature below the martensite start tem-
perature (Ms) in various iron–carbon steels. Although the
*
Corresponding author. Tel.: +82 2 2123 2831; fax: +82 2 312 5375. equation was originally developed using iron–carbon steels,
E-mail address: yklee@yonsei.ac.kr (Y.-K. Lee). many researchers have cited it without any modification

1359-6454/$34.00 Ó 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2007.11.039
S.-J. Lee, Y.-K. Lee / Acta Materialia 56 (2008) 1482–1490 1483

for low-alloy steels containing alloying elements such as [13] first introduced this equation. It is based on the work
chromium, nickel and molybdenum. In addition, the of Zener [14] and Hillert [15].
equation generates a C-curve shape for the martensite vol-
dV
ume fraction plotted against the cooling temperature below ¼ f ðchem; N ; Q; DT Þ  gðV Þ ð1Þ
Ms. In contrast, for most low-alloys steels the martensitic dt
transformation kinetic curve exhibits a sigmoid shape. where V is the volume fraction of the product phase at a
Although many researchers [1–6] have attempted to process time t, chem means the effects of alloying elements
clarify the exact relationship between phase transforma- on the diffusive mobility, N is the ASTM grain size num-
tions and internal stress, few studies that clearly explain ber, Q is the activation energy for the transformation and
the interaction between transformation kinetics and distor- DT is the undercooling below the equilibrium transforma-
tion have been conducted. Therefore, the purpose of the tion temperature. g(V) is a function of V relating to the
present study was to investigate the relationship between overall kinetics rate. The differential form of Eq. (1) is con-
transformation kinetics, focussing on martensitic transfor- venient, since it allows the combination of the kinetics
mation and distortion using an AISI 5120 steel, which is model for phase transformations with a constitutive mate-
widely used for diverse automobile parts. rials model so that the stress–strain matrix can be calcu-
In the present work the Ms point and a martensite kinet- lated within the finite element analysis.
ics equation for steel are proposed. The equation considers The dimensional change of the dilatometric specimen
austenite grain size (AGS), chemistry and the shape of the due to the transformation during heat-up results from the
kinetic curve. The Ms point and the equation were validated crystal structural change from body-centered cubic (bcc)
with experimental data from the literature. A finite element ferrite and pearlite to face-centered cubic (fcc) austenite.
method (FEM) analysis was performed, using thermal This dimensional change, which is a transformation strain,
and mechanical properties obtained from thermodynamic can induce stresses within the specimen that could affect
calculations and the literature. Quenching experiments subsequent transformations during the heating and cooling
using cut cylinders were conducted. The experimentally mea- processes. Thus, the kinetic model of the transformation
sured temperature and distortion data were used to explain during heating has to be considered in the heat treatment
the relationship between the transformation kinetics and simulation for the accurate prediction of the final distor-
distortion within the FEM simulations. tion. The austenite volume fraction is obtained by applying
a lever rule to the change in dilatometric curves assuming
2. Transformation kinetics model an isotropic transformation and insignificant cementite
effect. According to the lever rule, the volume fraction of
2.1. Diffusive transformation transformed phase at a given temperature is calculated by
the ratio of the measured transformation strain to the total
Dilatometric specimens of AISI 5120 steel were transformation strain, which is the gap between the extrap-
machined into small plates of 10  3  1 mm3 from a olated linear thermal expansion lines of parent phase and
hot-rolled bar. Table 1 lists the chemical composition of fully transformed phase at that temperature. Based on
AISI 5120 steel. The initial microstructure of the dilatomet- Eq. (1), the optimized kinetic equation for the diffusive
ric specimen was a mixture of ferrite and pearlite produced transformation on heating of AISI 5120 steel is given by
by furnace cooling. The specimens were austenitized at  
dV A 242742
900 °C with heating rates ranging from 1 to 50 °C s1, 4:45
¼ 8932  ðT  Ae1 Þ  exp  0:14
VA
and held for 10 min in a vacuum. The specimens were then dt RT
cooled to room temperature at cooling rates from 1 to  ð1  V A Þ3:07 ð2Þ
50 °C s1 by blowing nitrogen gas. A dilatometer was used
to measure contractions and expansions during the heating where Ae1 is an equilibrium eutectoid temperature, R is the
and cooling. The sensor force needed to hold a dilatometric gas constant (8.314 J mol1 K1) and VA is the volume
specimen (7.9 kPa) was too small to produce plastic trans- fraction of austenite. The values for the parameters in
formation phenomena. The cooled specimens were Eq. (2) were based on the austenite volume fraction
mechanically polished and etched using 2% Nital. obtained from an optimization program.
A common differential formula to characterize the diffu- The phases formed by diffusive transformation during
sive transformation was used in this study. Kirkaldy et al. cooling are classified as ferrite, pearlite and bainite, while
martensite forms via a diffusionless transformation. Thus,
it is impossible to apply a lever rule to obtain the product
phase fractions in a cooling process. The volume fractions
of the product phases were obtained using a routine [16]
Table 1
Chemical compositions of AISI 5120 steel (wt.%) that converts the transformation strain measured from a
dilatational curve to the volume fraction of each phase.
C Mn Si Cr P, S Fe
This conversion routine calculates more reasonable volume
0.21 0.89 0.24 1.25 <0.01 Bal.
fractions of product phases compared to the lever law, and
1484 S.-J. Lee, Y.-K. Lee / Acta Materialia 56 (2008) 1482–1490

is used in developing the kinetic models of both the diffu- Ms temperature of Fe–1 at.% C steel by Izumiyama
sive and diffusionless transformations during cooling. The et al. [22] is around 470 °C, while that of the same steel
kinetic equations of diffusive transformations were obtained by Ackert and Parr [23] is about 150 °C. This
obtained by optimization as follows: difference in the Ms temperature of the same steel possi-
Ferrite transformation bly comes from a difference in austenite grain size
  (AGS), which strongly affects the nucleation and growth
dV F 3:48 59093
¼ 91073  ðAe3  T Þ  exp   V 0:10
F of martensite.
dt RT
Some experimental results regarding the relationship
2:97
 ð1  V F Þ : ð3Þ between the AGS and the martensitic transformation have
Pearlite transformation been reported in Fe–Ni and Fe–Ni–C alloys [24,25]. The
  results indicate that the AGS has a significant effect on
dV P 2:12 40384 martensite formation. The Ms temperature rose with
¼ 24647  ðAe1  T Þ  exp   V 0:42
P
dt RT increasing austenite grain size especially in Fe–Ni–C alloys.
 ð1  V P Þ
1:46
: ð4Þ The relationship between the Ms temperature determined
from dilatational curves and the ASTM grain size numbers
Bainite transformation of low-alloy steels is investigated in this study, where it is
 
dV B 3:10 39538 found that the Ms temperature increases with decreasing
¼ 91111  ðBs  T Þ  exp   V 0:53
B ASTM grain size number.
dt RT
3:68 In order to obtain the experimental data regarding Ms
 ð1  V B Þ ; ð5Þ
temperature and martensitic kinetics, dilatometric tests of
where Ae3 , Ae1 and Bs are the transformation start temper- 29 low-alloy steels were conducted. The specimens were
atures of ferrite, pearlite and bainite, respectively. Vi is the heated to austenitizing temperatures ranging between 850
volume fraction of product phase i. and 1050 °C and held for a maximum of 90 min. In order
Fig. 1 shows a dilatometric curve measured at the cool- to obtain only the martensite phase from austenite, the
ing rate of 50 °C s1 and the volume fractions of product specimens were quenched to room temperature by blowing
phases calculated by the conversion routine. The micro- helium gas into the dilatometer chamber. The average cool-
structure of the sample was confirmed by optical ing rate between the austenitizing temperature and the Ms
microscopy. temperature was greater than 170 °C s1. The cooling rate
was slowed below the Ms temperature due to the latent
2.2. Martensitic transformation heat generated during the martensitic transformation.
For the measurement of the AGS, the quenched dilatomet-
A number of empirical formulae have been proposed ric specimens were etched in a saturated picric acid solution
to predict the Ms temperature as a function of the chem- after mechanical polishing with a 1 lm diamond suspen-
ical composition of steels [17–20]. The effect of the alloy- sion. Based on these data, the authors propose a new pre-
ing element on the Ms temperature of iron-based binary dictive equation of Ms temperature as functions of both
alloys has been investigated in many studies, and Liu chemical composition and the AGS of low-alloy steels as
et al. [21] have summarized the results. The measured follows:

Fig. 1. Dilatometric curve of AISI 5120 steel measured at a cooling rate of 50 °C s1 and its predicted volume fractions of product phases by the
conversion routine.
S.-J. Lee, Y.-K. Lee / Acta Materialia 56 (2008) 1482–1490 1485

Fig. 2. Comparison between the Koistinen–Marburger equation and Eq. (7) from this study with the measured (a) M50 and (b) M90 temperatures where
the martensite fractions are 50 and 90 vol.%, respectively.


M s ð CÞ ¼ 402  797C þ 14:4Mn þ 15:3Si  31:1Ni comparison between the K–M equation and Eq. (7).
The chemical composition and the AGS of the selected
þ 345:6Cr þ 434:6Mo þ ð59:6C þ 3:8Ni
steels from the published isothermal transformation
 41Cr  53:8MoÞ  G ð6Þ diagrams are quite different from the experimental con-
where each element is in weight per cent and G is the ditions used to formulate Eq. (7). The comparison
ASTM grain size number. between two kinetic equations with the measured M50
The K–M equation [12] and some similar equations and M90 temperatures is shown in Fig. 2. For the M50
[26,27] have been previously proposed to predict martensite temperature, the two equations reveal insignificant differ-
kinetics in steels. The Ms temperature is directly affected by ences. However, for the M90 temperature, Eq. (7) shows
the AGS, indicating that the kinetics of martensite trans- a very good agreement with the measured M90 tempera-
formation is also influenced by the AGS. However, these tures, while the values predicted by the K–M equation
previous kinetics equations, including the K–M equation, differ significantly.
do not contain an AGS term or factor. The new kinetics
equation for the martensitic transformation of low-alloy 3. Material properties
steels, which includes the effect of AGS, undercooling
below Ms temperature and chemical composition, was The thermal conductivity calculated by Miettinem’s
made based on the converted martensite fractions from formulae [29] is used in this study. He proposed equa-
the dilatational curves. The new kinetics equation is given tions to predict the thermal conductivity of alloyed steels
as at the liquidus temperature, at the austenite decomposi-
tion temperature, and at 400, 200 and 25 °C. He
dV M remarked that the thermal conductivity is usually not
¼ K  V aM  ð1  V M Þb
dT known for each individual solid phase but rather for
:191
G:240  ðM s  T Þ the solid as a whole.
K¼ The values (J mol1 K1) for heat capacity (CP) were
9:017 þ 62:88  C þ 9:27  Ni  1:08  Cr þ :76  Mo
a ¼ :420  :246  C þ :359  C 2 calculated using Thermo-Calc [30], assuming the steel to
be in equilibrium. The heat capacities of austenite, ferrite
b ¼ :320 þ :576  C þ :933  C2 and ferrite+cementite as a function of temperature for
ð7Þ AISI 5120 steel are
where VM is the volume fraction of martensite, C is carbon austenite
content in weight per cent and T is the temperature below C P ¼ 93:82 þ 25:162T 0:5  0:378T þ 0:0000717T 2 ; ð8Þ
the Ms temperature in degrees Celsius.
The M50 and M90 temperatures, where the martensite ferrite
fractions are 50 and 90 vol.%, respectively, were C P ¼ 8938:11 þ 444417:483=T þ 786:886T 0:5
obtained from the published isothermal transformation
diagrams [28] of 37 low-alloy steels for more reliable  20:662T þ 0:00529T 2 ; ð9Þ
1486 S.-J. Lee, Y.-K. Lee / Acta Materialia 56 (2008) 1482–1490

ferrite + cementite 4. Experiments and FEM simulation of an asymmetrically


0:5 cut cylinder
C P ¼ 1091:734 þ 3768:92=T þ 175:576T
 5:742T þ 0:00227T 2 ð10Þ Fig. 3 shows the shape and dimensions of an asym-
where T is temperature in Kelvin. A simple rule of mixtures metrically cut cylinder of AISI 5120 steel. The asymmet-
is applied to obtain the heat capacity for multiphase ric design is helpful for the investigation of the
conditions. relationship between transformation kinetics and distor-
The enthalpy change due to a phase transformation, i.e. tion during quenching. The asymmetrically cut cylinder
latent heat, causes heat absorption or heat generation of was austenitized at 860 °C for 10 min and quenched in
the system. In this study, the latent heat of the diffusive oil at 17 °C. K-type thermocouples and a multichannel
transformations is calculated based on the thermodynamics recorder were used to measure surface temperatures dur-
of the transformation. The latent heat for ferrite formation ing the heat treatment. To obtain a heat transfer (convec-
is calculated at the temperature at which the austenite tion) coefficient during oil quenching, a cylinder of AISI
decomposition is thermodynamically complete, while the 304 stainless steel (10 mm diameter  100 mm long) was
latent heat of the pearlite formation is calculated by the austenitized at 860 °C for 10 min and quenched in the
rule of mixtures between the latent heats of the cementite same oil. 304 stainless steel was selected because no
and ferrite formation. Although bainite is composed of latent heat is generated by phase transformation during
ferrite and cementite-like pearlite, the latent heat of the the heat treatment. The published thermal properties of
bainite formation contains an additional shear energy AISI 304 stainless steel [38] and the measured surface
value of 600 J mol1, which was reported by Nanba et al. temperatures were used to determine the convection coef-
[31]. The calculated latent heats of ferrite, pearlite and bai- ficient. The convection coefficient was calculated as a
nite are: DHF = 5.95  108, DHP = 5.26  108 and DHB = function of temperature by the inverse algorithm shown
5.12  108 (J m3), respectively. in Fig. 4.
Only a few studies have reported the latent heat of mar- The FEM simulation was performed using ABAQUS
tensite formation. Recently, Cho et al. [32] suggested the [39] and its user-defined subroutines UMAT and
following equation to calculate the Gibbs free energy UMATHT. The hexahedral element (C3D8T) was used
change as the latent heat of martensite transformation: and the total numbers of nodes and elements were 2205
  and 1632, respectively. Two-step conditions were specified
C T  Ms
DG ¼ DG 1  ð11Þ for the simulation: the heating process of the asymmetri-
T 0  Ms cally cut cylinder from room temperature to 860 °C was
where DGC is the Gibbs free energy change between austen- simulated by convectional heat transfer (300 W m2 K1)
ite and martensite at the Ms temperature. According to for 15 min followed by quenching simulation for 5 min
Kunze and Beyer [33], DGC is 2100 J mol1 for the forma- using the convection coefficient obtained from the AISI
tion of plate martensite and (1200 + 3128yCr + 29260yMn + 304 stainless cylinder.
6470yNi + 21000yC) J mol1 for lath martensite, where yi is
the site fraction of element i. T0 is a thermodynamic equi-
librium temperature at which the chemical free energies of
austenite and martensite are equal and is usually expressed
as T0 = 1/2(Ms + As) [34]. As is the austenite start temper-
ature from martensite and Andrews’ formula [19] is used to
calculate the As temperature of AISI 5120 steel. The latent
heat of the martensite formation of AISI 5120 steel is
calculated using Eqs. (7) and (11). The value for the latent
heat for martensite formation is DHM = 3.14  108 (J m3).
The published stress–strain curves of AISI 5120 steel
with different microstructures are used for the stress anal-
ysis [35]. The stress–strain curves were generated as a
function of temperature using Instron and Gleeble
machines. In addition, the hardness is also an important
means to assess the mechanical properties after heat
treatment. In this study, the empirical formula proposed
by Maynier et al. [36] was used to predict the hardness
after cooling. The values of the phase transformation
plasticity of AISI 5120 steel are taken from recently
measured data [37]. The thermal and transformation
expansions are calculated by the equations used in the Fig. 3. Shape and dimension of the cut cylinder specimen of AISI 5120
previous work [16]. steel.
S.-J. Lee, Y.-K. Lee / Acta Materialia 56 (2008) 1482–1490 1487

Fig. 6. Phase fractions predicted at two different positions of the sample


and the predicted and measured hardness values at the central cross-
section of the quenched cut cylinder.

Fig. 4. Calculated convection coefficient of oil quenching using AISI 304 cooling, the bainitic transformation occurs and is followed
stainless steel. by the martensite transformation. The predicted relative
amount of martensite is 77% on the edge and 71% in the
center. This difference is due to the different cooling rates
5. Results and discussion
throughout the thickness of the sample producing different
bainite fractions prior to the start of martensite formation.
Fig. 5 shows the comparison between the surface tem-
The measured average hardness of the central cross-section
peratures obtained by the simulation and the measured
of the quenched asymmetrically cut cylinder is about 42
temperatures at different positions of the asymmetrically
HRC. The predicted hardness based on the Maynier’s for-
cut cylinder during oil quenching. During this quenching,
mula at the same position is within ±2.2% of the measured
the measured temperature changes at four different points
hardness.
are similar. The predicted cooling profiles show good
Distortion of the asymmetrically cut cylinder before and
agreement with the measured ones. Unfortunately, how-
after oil quenching was quantitatively measured at nine dif-
ever, the latent heat, which occurs during oil quenching,
ferent points along the longitudinal direction at the center
is too small to cause a significant temperature change
of the outside surface of the cylinder using a coordinate
because of both the low carbon content of the steel and
measuring machine with a minimum resolution of
the relatively small volume of the cylindrical specimen.
100 nm. Fig. 7 shows the predicted distortion to have very
Fig. 6 shows the predicted microstructural changes on
good agreement with the measured distortion. The maxi-
the edge and in the center of the asymmetrically cut cylin-
der during oil quenching. Within 2 or 3 s of the start of

Fig. 7. Predicted and measured distortions of the asymmetrically cut


Fig. 5. Comparison between the predicted and measured surface temper- cylinder, which was bent in the opposite direction of the cutting plane (axis
atures at each different position of the cut cylinder during oil quenching. 1 direction) after quenching.
1488 S.-J. Lee, Y.-K. Lee / Acta Materialia 56 (2008) 1482–1490

mum distortion is approximately 500 lm, which could pose Fig. 8 shows the relationship between distortion and
a problem in terms of dimensional stability during a com- microstructural change in the vertical section during oil
mercial heat treatment process. quenching. The specimen was bent in the normal direction

Fig. 8. Relationship between the distortion and microstructure changes in the asymmetrically cut cylinder specimen during oil quenching: (a) bainite and
(b) martensite.

Fig. 9. Effect of phase transformations on the distortion of the cut cylinder during oil quenching: (a) distortion with transformations and (b) distortion
without transformations.
S.-J. Lee, Y.-K. Lee / Acta Materialia 56 (2008) 1482–1490 1489

Fig. 10. Variations in the axial stress (rz), radial stress (rr) and hoop stress (rh) during oil quenching of the cut cylinder: (a) without phase transformations
and (b) with phase transformations. The subscripts (C, S and cut S) indicate the node positions of the cut cylinder at which the stresses were calculated.

of the cutting plane (axis 1 direction) at the beginning of stresses become greater as shown in Fig. 10b. The increased
the oil quench and shortened in the longitudinal direction stress variation is related to the bainite and martensite
due to thermal contraction. When the bainite and martens- transformations combined with thermal contraction of
ite transformations started, the additional transforma- the asymmetrically shaped cylinder (Fig. 8).
tional strain and the strain due to thermal expansion Fig. 11 compares the effect of the martensite kinetics on
affected the distortion of the asymmetrically cut cylinder. quenching distortion using two different kinetic equations:
Additionally, the position-dependent transformations have the K–M equation and Eq. (7). The same material proper-
an influence on the distortion direction. Finally, the distor- ties and initial and boundary conditions were used for the
tion changed to the direction opposite to the cutting plane distortion simulation. Even if the predicted distortion was
(opposite to axis 1 direction) due to the transformation. in the same direction (opposite to axis 1) after quenching,
The original causes of the distortion are not only the asym- the relative amount of distortion of the quenched cut cylin-
metric shape of the cut cylinder but also the additional der would be quite different. The predicted distortion calcu-
transformation strains. lated using the K–M equation does not reach 200 lm,
Fig. 9 shows the effect of the transformation strain on while the distortion predicted using Eq. (7) is greater than
the distortion, which was investigated by computer simula- 500 lm. The accuracy of the martensite kinetic equation,
tions. The distortion simulations were performed with the
same initial and boundary conditions but different trans-
formation strains. Fig. 9b shows the quenching distortion
without transformation strains, indicating the distortion
due to the continuous contraction of austenite in the direc-
tion normal to the cutting plane (axis 1 direction).
Fig. 10 provides a comparison of the variations in the
axial stress component (rz), radial stress component (rr)
and hoop stress component (rh) during oil quenching of
the cut cylinder with the phase transformation effect being
considered. Without consideration of the transformation
strains, as shown in Fig. 10a, the tensile stress at surface
and the compressive stress at the center of an austenitic
specimen are generated at the beginning of oil quenching
because the surface temperature drops faster than the inner
temperature. With continued cooling, the cooling rate at
the surface is decreased while that at the center is increased,
and the temperature difference between these two tempera-
tures is reduced by a few degrees. The compressed stress at
the surface and the tensile stress at center are generated
when the cooling rate at the center becomes greater than
Fig. 11. Effect of martensite kinetics on the final distortion after oil
that at the surface. However, when considering the trans- quenching. Two different kinetic equations were compared: the Koistinen–
formation strains, the stress variation is more complicated Marburger equation and Eq. (7) proposed in this study. The measured
and the amounts of the maximum compressive and tensile values at P1–P9 were referred to previously in Fig. 7.
1490 S.-J. Lee, Y.-K. Lee / Acta Materialia 56 (2008) 1482–1490

Eq. (7), proposed in this study, demonstrates the need to [3] Inoue T, Wang Z. Mater Sci Tech 1985;1:845.
have a reliable kinetics model for phase transformations [4] Yamanaka S, Sakanoue T, Yoshii T, Kozuka T, Inoue T. In:
Proceedings of ASM heat treating conference & exposition. ASM
if accurate distortion is to be predicted. International; 1998. p. 657.
[5] Arimoto K, Horino T, Ikuta F, Jin C, Tamura S. In: Proceedings of
6. Conclusion ASM heat treating conference & exposition. ASM International;
1998. p. 253.
The relationship between transformation kinetics and [6] Ju DY, Zhang WM, Zhang Y. Mater Sci Eng A 2006;438–440:246.
[7] Inoue T, Ju DY, Arimoto K. In: Totten GE, editor. Proceedings of
distortion during oil quenching of AISI 5120 steel has been the 1st international conference on quenching and control of
investigated. Experimental results were compared with distortion. Materials Park, OH: ASM International; 1992. p. 205.
computational simulations using ABAQUS with its user- [8] SYSWELD. A predictive model for heat treat distortion. Southwest
defined subroutines UMAT and UMATHT. To predict Research Institute; 1992.
accurate martensite volume fraction during quenching, a [9] Arimoto K, Lambert D, Li G, Arvind A, Wu WT. In: Wallis RA,
Walton H, editors. Proceedings of the 18th conference on heat
new Ms temperature and kinetics equations of diffusive treating. Materials Park, OH: ASM International; 1998. p. 639–54.
and diffusionless transformations are suggested. These [10] Ferguson BL, Petrus GJ, Pattok T. In: Proceedings of the 3rd
equations consider the influences of austenite grain size, international conference on quenching and control of distor-
alloy elements and the shape of the kinetics curve. The tem- tion. Materials Park, OH: ASM International; 1999. p. 188.
perature change during oil quenching and distortion of an [11] Ju DY, Ito Y, Inoue T. In: Proceedings of the 4th international
conference on quenching and control of distortion. Materials Park,
asymmetrical shaped AISI 5120 cut cylinder were mea- OH: ASM International; 2003. p. 291.
sured. FEM simulations were performed to predict the [12] Koistinen DP, Marburger RE. Acta Metall 1959;7:59.
microstructure, temperature, distortion and hardness of [13] Kirkaldy JS, Venugopalan D. In: Marder AR, Goldstein JI, editors.
AISI 5120 steel during heat treatment. These simulations Phase transformations in ferrous alloys. Warrendale, PA: TMS-
used the thermal and mechanical properties obtained from AIME; 1984. p. 125.
[14] Zener C. Trans AIME 1946;167:550.
thermodynamic calculations, literature and transformation [15] Hillert M. Jermkont Ann 1957;141:557.
kinetics measured by a dilatometer. The predicted results [16] Lee S-J, Lusk MT, Lee Y-K. Acta Mater 2007;55:875.
were successfully validated with experimentally measured [17] Payson P, Savage CH. Trans ASM 1944;33:261.
and observed results. [18] Steven W, Haynes AG. JISI 1956;183:349.
The effects of transformations on the distortion of the [19] Andrew KW. JISI 1965;203:721.
[20] Capdevila C, Caballero FG, Garcı́a De Andrés C. ISIJ Int
cut cylinder (i.e. the transformation strain, phase-depen- 2002;42:894.
dent thermal expansion coefficients and flow stresses) are [21] Liu C, Zhao Z, Northwood DO, Liu Y. J Mater Proc Tech
clearly verified by comparing the simulated results with/ 2001;113:556.
without phase transformations. The phase transformations [22] Izumiyama M, Tsuchiya M, Imai Y. J Japan Inst Net 1974;34:291.
as well as the thermal contraction of the asymmetrically cut [23] Ackert RJ, Parr JG. JISI 1971;209:912.
[24] Umemoto M, Owen WS. Metall Trans 1974;5:2041.
cylinder upon cooling cause high stress values. The impor- [25] Guimarães JRC, Gomes JC. Acta Metall 1978;26:1591.
tance of an accurate martensite kinetics for better predic- [26] Harris WJ, Cohen M. Trans AIME 1949;180:447.
tion of quenching distortion was verified by using two [27] Skrotzki B. J de Phys IV 1991;1:367.
different martensite kinetic equations: the K–M equation [28] American Society for Metals. Atlas of isothermal transformation and
and the new kinetic equation proposed in this study. The cooling transformation diagrams. Metals park, OH: ASM Interna-
tional; 1977.
final distortion of the quenched asymmetrically cut cylinder [29] Miettinen J. Metall Mater Trans B 1997;28B:281.
shows excellent correlation with the new martensite kinet- [30] Thermo-Calc version P with TCFE3 database, Stockholm, Sweden.
ics equation. [31] Lee KJ. Scripta Mater 1999;40:735.
[32] Cho YG, Im YR, Kim G, Han HN. Solid State Phen 2006;118:343.
Acknowledgment [33] Kunze J, Beyer B. Z Metallkd 2000;91:106.
[34] Jena AK, Chaturvedi MC. Phase transformation in materials. Engle-
wood Cliffs, NJ: Prentice Hall; 1992.
This research was supported by the National Core [35] Lee KO, Kim JM, Chin MH, Kang SS. J Mater Proc Tech
Research Center (NCRC) program from MOST and 2007;182:65.
KOSEF (No. R15-2006-022-01002-0). The authors are [36] Maynier P, Dollet J, Bastien P. In: Doane DV, Kirkaldy JS, editors.
grateful to Professor C.J. Van Tyne at the Colorado School Hardenability concepts with applications to steels. New York: AIME;
1978. p. 163.
of Mines for helpful discussions. [37] T. Inoue, Y. Watanabe, K. Okamura, M. Narazaki, H. Shichino,
D.Y. Ju, et al. In: Proceedings of the 1st international conference on
References distortion engineering. Bremen, Germany; 2005. p. 133.
[38] American Society for Metals. Thermal properties of metals. Metals
[1] Denis S, Gautier E, Simon A, Beck G. Mater Sci Tech 1985;1:805. Park, OH: ASM International; 2002.
[2] Denis S, Sjöström S, Simon A. Metal Trans A 1987;18A:1203. [39] ABAQUS Analysis user’s manual, Version 6.6.

Das könnte Ihnen auch gefallen