Steel HT - Simulation
Steel HT - Simulation
CIMEC - INTEC
(UNL / Conicet)
Gemes 3450, (3000) Santa Fe, Argentina
e-mail: acardona@intec.unl.edu.ar
2443
1 INTRODUCTION
Heat treatment of metallic alloys is a complex thermomechanical process involving solid
state metallurgical transformations that change both thermal and mechanical properties of
materials.
This process is widely used in industrial applications, to release internal stresses, reduce
fragility, improve machinability or modify properties like hardness or strength, to satisfy the
requirements of a definite application. However, heat treatment can cause, during or after
them, undesirable strains and stresses, and also cracking of parts. This fact must be taken into
account when designing the heating and cooling sequence of the process.
Numerical simulation of heat treatment was subject of many research works. Some of them
focused on thermal and mechanical analysis of the process 3,4,12, while the others focused on
aspects of material modeling 1,8,9,13.
Material models with the ability to account for variations in thermal and mechanical
properties due to temperature and metallurgical structure changes are a key point to simulate
accurately the thermomechanical evolution of parts subjected to heat treatment. Several
models describe material properties as functions of alloy composition, temperature and
microstructure. Other models describe the evolution of microstructure as a function of alloy
composition and cooling time, reproducing TTT or CCT diagrams for different alloys and
chemical compositions.
Even though advanced material models are very useful to cover a broad range of chemical
compositions for some well-known alloys (carbon steels, low-alloy steels, austenitic stainless
steels), their efficacy for other alloys is very limited because of the difficulty to obtain all the
coefficients needed to represent accurately their behavior (e.g. for high-alloy white iron).
In this work, we follow an alternative simplified way to simulate heat treatment, using
material properties defined as functions of time and temperature, obtained by merging
TTT/CCT diagrams with curves of dependency of thermomechanical properties with
temperature. Simulations were done using the commercial finite element software SAMCEF.
Section 2.1 describes the numerical model used to simulate thermomechanical processes in
heat treatment of ferrous metals. In section 2.2, a description of the proposed material model
is given. Section 3 presents an application of this model to simulation of heat treatment of
centrifugally cast, three-layer, Hi-Chrome work rolls, commonly used in steel mills.
2 NUMERICAL MODEL
2.1 Thermomechanical model
Numerical analysis of heat treatment processes can be made by modeling the time
evolution of two coupled problems:
1) a thermal one, that involves heating and cooling of parts and must take into account
variations in material properties (thermal conductivity and enthalpy) due to
temperature and microstructural transformations, and also heat releasing/absorption
2444
(T )
H (T ,m )
t
( k ( T , m ) T ) = 0
(1)
where is the density, H the enthalpy, t is the time, k the conductivity, T the temperature and
m accounts for the dependency of material parameters on microstructure.
Convective boundary conditions are applied on the external surface of the roll:
k (T ,m ) T n = h (T Tb )
(2)
where n is the outer normal vector to the external surface, h is the film coefficient and Tb is
the bulk (ambient) temperature.
The relationship between capacity and enthalpy in the presence of phase change is given
by the following expression (assuming the same density for both phases) :
B
epc
(T ) H (T ,m ) = (T ) ceff (T ,m ) dT =
1(T )
(3)
spc
where ceff is the effective (apparent) heat capacity, c1 and c2 are the specific heats for
different microstructures, 1 and 2 are proportions of initial and final microstructural
components, Lpc is the latent heat necessary for a phase change, spc is the initial temperature
of phase change and epc the temperature at the end of phase change.
The dependence of the mechanical properties on material microstructure is simulated using
material models that account for variations in metallurgical constituents with time. The
thermal dependence of the mechanical problem is modeled using a thermal field calculated in
2445
the thermal simulation and given as input to evaluate the mechanical properties for the
mechanical analysis and to compute the strains.
In the mechanical simulation, we use an elastoplastic model in which the stresses are
calculated as:
= C (T , p ,m ) e = C (T , p ,m ) ( p t )
(4)
where and are stress and strain vectors, C is the constitutive tensor, e is the elastic
strain, p is the plastic strain and t is the thermal strain.
The thermal strain is calculated as:
(T )
(T )
t = (T ) (T Tref )
0
0
(5)
where is the thermal dilatation coefficient and Tref is the reference (zero strain) temperature.
The stress field must satisfy a yield criterion (in this case, the isotropic Von Mises
criterion):
eq
( )
1
( x y )2 + ( y z )2 + ( z y )2 + 6 xy2 + yz2 + zx2 Y(T ,eq )
2
(6)
Since the limit stress Y is a function of the equivalent plastic strain (eq) and temperature,
the model can take into account different material hardening behaviors at different
temperatures.
2.2 Material model
Most material properties are a function of temperature and microstructure, and therefore
they have indirect dependency on variables defining material microstructure, like time and
maximal heating/cooling temperature .
In austenitizing processes (heating), the microstructure is modeled as a function of
temperature. In quenching processes (cooling), the microstructure is a function of temperature
and time. In tempering processes (heating and cooling), the microstructure is considered as a
function of only temperature.
Then, for the whole process, we can formulate any general property , as dependent on
temperature and time, as follows:
2446
(T , m(T ,t ) )
(7)
(T , t )
Temperature
Austenitic
Pearlite + Austenite
Pearlitic
Pearlitic
Pearlitic or
Martensitic
Martensite +
Austenite
Time
Austenizing Heating
Quenching Cooling
Tempering Heating
Tempering Cooling
21
Pearlitic
Heating transition
20
Cooling transition
19
Austenitic
0
400
800
Temp.
[C]
Then, for every definite material microstructure (i.e.: austenitic, pearlitic, martensitic), we
define the value of every property as a function of temperature. Figure 2 is an example for
conductivity during the heating process.
Finally, by combining the map of microstructure as a function of temperature and time,
with the curves of temperature dependence of the considered property, and by using the rule
2447
of mixtures for regions with mixed structure (e.g. austenite + pearlite), a map of the property
as a function of temperature and time (continuous piecewise linear approximation) can be
built, as shown in figure 3.
Conductivity
[W/mC]
22
20
Temperature
[C]
800
600
800
400
Time
[Hs]
600
200
0
400
-200
200
0
In this work, we used the commercial finite element code SAMCEF V10.1, in which
material properties can be defined as functions of temperature and time. Using a standard
material model, special care must be taken to adapt the map of microstructure as a function of
temperature and time to the real quenching cooling process, thus avoiding reversions in
austenite-pearlite and austenite-martensite transformations. An improvement will be the
development of a model with the ability to track microstructural evolution and avoid
numerical reversion of physically irreversible phase changes. This work is in progress and
being implemented in our object-oriented finite element code OOFELIE.
2448
Austenizing
Heating
Quenching
Cooling
Austenite
Temperature
[C]
Pearlite
1000
Conductivity
[W/mC]
800
Pearlite
22
600
400
20
200
Martensite
160
0
170
180
-200
190 Time
[Hs]
Figure 4: Conductivity vs. time and temperature diagram Detail in the quenching cooling zone
In the thermal analysis, the material parameters are enthalpy and thermal conductivity. In
the mechanical analysis, the material parameters are Young (elastic) modulus, Poisson
coefficient, thermal expansion coefficient, and yield stress. All of them are modeled following
the above mentioned procedure.
3 APPLICATION CASE
3.1 Problem description
Hi-Chrome work rolls have an exterior layer (shell) of Hi-Chrome white iron, an
intermediate layer of low-alloy iron, and a core of spheroidal graphite iron. The first two
layers are casted centrifugally, and then the core is poured statically. Typical as-cast
dimensions are shown in figure 5.
2449
Mesh Detail
After casting, rolls are heated to 1020 C (with complete austenitization), then they are
quenched cooled in air to room temperature, and finally subjected to two tempering processes
at 480 C and 540 C with air cooling to room temperature. This process requires special care
to avoid defects as cracking near the edges of the barrel.
By ignoring small circumferential temperature differences, and small bending stresses
generated by the horizontal mounting of cylinders, the problem can be modeled as an
axisymmetrical one. The finite element mesh is shown in figure 5.
Convective boundary conditions were modeled on the external surface of the roll. Film
coefficients ranging from 5 [W/(m2C)] to 40 [W/(m2C)] were used in the different stages of
heating and cooling, depending on the agitation of surrounding air.
3.2 Shell material data (Hi-Chrome white cast iron)
A TTT diagram for a similar alloy was used to determine position of pearlitic nose 5. The
cooling behavior of the shell material was determined using a CCT diagram for a Hi-Cr white
iron without Ni 7, with a correction of the transformation time to take into account the
influence of Ni and Mo, as suggested by Laird et al. 10. A plot of the used diagram is shown in
figure 6.
Thermal conductivity as a function of temperature and microstructure was taken from data
2450
600
Pearlite
500
Austenite
400
300
25%
200
100
50% 75%
Austenite +
Martensite
0
1000
10000
100000
Time.
[sec]
25%
600
Pearlite
50%
500
Austenite
75%
400
300
200
Austenite +
Martensite
100
0
100
1000
10000
Time.
[sec]
Yield stress dependence on temperature was taken from tables for SG iron of similar
composition 11.
3.4 Results
In figure 8, a comparison between calculated (continuous line) and measured (dashed line)
temperatures at the midpoint of the barrel surface is shown. The agreement between curves is
found quite acceptable for the purposes of the analysis.
The largest differences between temperatures in different points of the roll are found at the
initial stage of quenching cooling. Figure 9 shows a detail of these differences between points
located in the surface of the barrel (P1, P4) and others situated in the zones of transition
between layers of different materials.
In figures 10 to 17, the evolution of temperatures, principal stresses and equivalent plastic
deformations near the corner are shown. The time instants of these plots along the complete
heat treatment are referred by letters (A) to (H) in figure 8.
Temperature plots show very small differences between temperatures in different sectors
of the roll during heating processes. Even during cooling processes with very high variations
in time, the maximum temperature differences at a given time instant are below 200C. The
computed temperature space gradients (radial) were also relatively small.
During the austenitizing heating, axial stresses in the shell are tensile stresses. However, as
2452
plastic deformations occur during heating, in the quenching cooling and in the tempering
processes axial stresses in the shell are compressive.
Temperature [C]
900
700
G
E
500
300
Measured
D
100
Calculated
0
110
220
330
440
550
Time [Hs]
660
P4
P3
700.
P1
P2
P3
P2
600.
500.
P1
400.
300.
P4
200.
0
11
22
33
44
55
Time [Hs]
2453
TRACTION
COMPRESSION.
Temp
VALUE 1.E-3
C
706.2
4.719
702.1
4.248
697.9
3.776
693.7
3.304
689.5
2.832
685.3
2.36
681.1
1.888
677
1.416
672.8
0.944
668.6
0.472
664.4
Principal stresses
TRACTION
COMPRE.
Temp
VALUE 1.E-3
C
1027
8.797
1027
7.918
1026
7.039
1026
6.159
1025
5.279
1025
4.399
1024
3.519
1024
2.639
1.76
1023
0.88
1023
1022
Principal stresses
TRACTION
COMPRE.
Temp
C
462.4
VALUE 1.E-3
8.797
431.1
7.918
399.7
7.039
368.4
6.159
337
5.279
305.7
4.399
274.3
3.519
242.9
2.639
211.6
1.76
180.2
0.88
148.9
2454
TRACTION
COMPRE.
Temp
C
32.44
VALUE 1.E-3
9.254
32.19
8.33
31.94
7.404
31.7
6.479
31.45
5.553
31.2
4.628
30.95
3.702
30.7
2.777
30.46
1.851
0.926
30.21
29.96
Principal stresses
TRACTION
COMPRE.
Temp
VALUE 1.E-3
9.258
C
501.1
498.4
8.333
495.8
7.407
493.1
6.481
490.5
5.555
487.8
4.629
485.2
3.703
482.5
2.778
479.9
1.852
477.2
0.926
474.6
Principal stresses
TRACTION
COMPRE.
Temp
VALUE 1.E-3
C
58.81
9.258
58.57
8.333
58.33
7.407
58.09
6.481
57.85
5.555
57.61
4.629
57.36
3.703
57.12
2.778
56.88
1.852
56.64
0.926
56.4
2455
TRACTION
COMPRE.
Temp
VALUE 1.E-3
C
556.5
9.258
555.9
8.333
555.3
7.407
554.7
6.481
554.2
5.555
553.6
4.629
553
3.703
552.4
2.778
551.8
1.852
551.3
0.926
550.7
Principal stresses
TRACTION
COMPRE.
Temp
VALUE 1.E-3
C
107.6
9.258
104.6
8.333
101.6
7.407
98.6
6.481
95.7
5.555
92.7
4.629
89.7
3.703
86.7
2.778
83.7
1.852
80.7
0.926
77.8
Figure 18 shows the evolution of equivalent plastic strains in different points near the
barrel corner.
Most of the inelastic deformations occur during the second half of the austenitizing heating
process and during the quenching cooling, as shown in equivalent plastic strain plots. No
evident increment of plastic strains is found during tempering processes. However, this result
could be caused by the lack of information about percentages of retained austenite after
quenching cooling and transformations of retained austenite during tempering.
The magnitude of inelastic strains found near the interface between shell and intermediate
layers, close to 1%, are significative, specially taking into account the brittle nature of HiChrome White iron.
2456
P1
0.006
P1
P2
0.005
0.004
P2
0.003
0.002
0.001
55
110
165
Austenizing Heating
220
275
Time [Hs]
Quenching Cooling
2457
5 REFERENCES
[1] Alberg H. Material Modeling for Simulation of Heat Treatment. Licentiate Thesis,
Lulea University of Technology (2003).
[2] Belyakova P.E. Thermophysical Properties of Wear-Resisting Cast Irons.
Metallovediene I Termicheskaya Obrabotka Metallov. (1975).
[3] Bergheau J.M. Contribution of Numerical Simulation to the Analysis of Heat Treatment
and Surface Hardening Processes. Proc.of ASM Heat Treatment Conference98 (1998).
[4] Berglund D. Simulation of Welding and Stress Relief Heat Treatment in Development of
Aerospace Components. Licentiate Thesis, Lulea University of Technology (2001).
[5] Betts W. Personal communication (2004).
[6] Boyer H.(ed.) Atlas of Isothermal Transformations and Cooling Transformations
Diagrams. American Society for Metals (1977).
[7] Gundlach R. and Doane D. Alloy Cast Irons. in ASM Metals Handbook. American
Society for Metals (2003).
[8] Heine R.W. A Model for Specific Volume and Expansion and Contraction Behavior of
Solidifying Cooling Ductile and Gray Iron. AFS Transactions (1987).
[9] Kirkaldy J., Thomson B. and Baganis E. Hardenability Concepts with Applications to
Steel. in J.Kirkaldy and D.Doane (eds.) AIME Transactions (1978).
[10] Laird G., Gundlach R. and Rrig K. Heat Treatment of High-Alloy AR Cast Irons.
Abrasion Resistant Cast Iron Handbook. American Society for Metals (1996).
[11] Lynch C.T. (ed.). CRC Handbook of Materials Science- Vol.2:Metals, Composites and
Refractory Materials . CRC Press (1975).
[12] Sanchez Sarmiento G., X. Chen, J. Vega, G. Totten, R. Reynoldson, L. Huynh and L.
Meekiso A Comparison On Cooling Curve Analysis Using Inc-Phatran and Winprobe.
Proc. of 20th ASM Heat Treating Society Confererence, (2000)
[13] Saunders N. The Calculation of TTT and CCT diagrams for General Steels.
Thermotech Ltd., Surrey Technology Centre, The Surrey Research Park, U.K. (2004).
[14] Stefanescue D. Physical Properties of Cast Iron (Chapter 8). in Iron Castings
Engineering Handbook. American Society for Metals (1995).
[15] Wang D. Thermophysical Property Data. Auburn Solidification Design Center.
http://metalcasting.auburn.edu/data/data.html (2001).
2458