You are on page 1of 13

Materials Science and Engineering A336 (2002) 177 189

www.elsevier.com/locate/msea

Analytical model of pass-by-pass strain in rod (or bar) rolling and


its applications to prediction of austenite grain size
Y. Lee a,*, S. Choi a, P.D. Hodgson b
a
Plate, Rod and Welding Group, POSCO Technical Research Laboratories, P.O. Box 36, Pohang 790 -785, South Korea
b
School of Engineering and Technology, Deakin Uni6ersity, Pigdons Road, Geelong, Victoria 3217, Australia

Received 22 August 2001; received in revised form 19 November 2001

Abstract

Rod rolling is a process where the deformation state of the workpiece between the work rolls is quite different from the strip
rolling process. However, in most microstructure evolution models, the simple area strains (natural logarithm of the area
reduction ratio) multiplied by a constant have been used to compute pass-by-pass evolution of austenite grain size (AGS) in rod
(or bar) rolling, without any verification. The strains at a given pass play a crucial role in determining the recrystallization
behavior (static or dynamic). In this study, an analytical model that calculates the pass-by-pass strain and strain rate in rod rolling
has been developed and verified by conducting four-pass (oval round) bar and plate rolling experiments. Numerical simulations
have then been carried out for the four-pass rolling sequence using the area strain model and the new analytical model, focusing
on the effect of the method for calculating the strain on the recrystallization behavior and evolution of AGS. The AGS predicted
was compared with those obtained from hot torsion tests. It is shown that the analytical model developed in this study is more
appropriate in the analysis of bar (or rod) rolling. It was found that the recrystallization behavior and evolution of AGS during
this process were influenced significantly by the calculation method for the deformation parameters (strain and strain rate). The
pass-by-pass strain obtained from the simple area strain model is inadequate to be used as an input to the equations for
recrystallization and AGS evolution under these rolling conditions. 2002 Elsevier Science B.V. All rights reserved.

Keywords: Rod rolling process; Numerical simulation; Deformation parameters; Austenite grain size (AGS)

1. Introduction sis of the microstructure evolution in hot bar rolling


coupled with the laboratory scale experiment-based
Many studies have been conducted to predict the mircostructural evolution model. However, considering
pass-by-pass evolution of the austenite grain size (AGS) the computational time used by this approach for bar
of low carbon steels during hot rolling [1 3]. The (or rod) mills that comprise a large number of passes
changes of AGS at a pass are considered to be deter- and complicated mechanical/thermal boundary condi-
mined by the recrystallization (dynamic or static) and tions (the friction condition at the roll/material inter-
grain growth behavior during and after hot rolling, and face and heat transfer coefficients depend on the
have been formulated as a function of temperature, temperature and contact pressure), a mathematical
strain, strain rate at a given pass and the initial grain model that calculates the pass-by-pass plastic deforma-
size. Most work has concentrated on the strip and plate tion (strain and strain rate) and the temperature evolu-
rolling processes, with little research carried out on the tion of the material during rolling within a very short
rod rolling process. time (say less than 1.0 s) still remains a useful tool.
Recently, Karhausen et al. [4] and Yanagimoto et al. There has been a strong demand to develop a mathe-
[5] presented the three-dimensional finite element analy- matical model in a simple form with reliable accuracy
and a non-iterative computational framework, for ob-
* Corresponding author. Tel.: +82-54-220-6058; fax: + 82-54-220-
taining the thermo-mechanical parameters (strain,
6911 strain rate and temperature) associated with the rod (or
E-mail address: pc554162@posco.co.kr (Y. Lee). bar) rolling process. The reason is that these parameters

0921-5093/02/$ - see front matter 2002 Elsevier Science B.V. All rights reserved.
PII: S 0 9 2 1 - 5 0 9 3 ( 0 1 ) 0 1 9 6 4 - 5
178 Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189

are the key elements to predict the AGS in rod rolling Once the analytical model was verified, a numerical
because the recrystallization behavior and the AGS simulation was performed for the four-pass bar rolling
evolution model developed for strip rolling can be sequence. The deformation parameters (strain and
applied directly to rod rolling. There have been several strain rate) obtained by the area strain model and the
attempts [68] to use this approach for rod rolling. analytical model were used as input data for recrystal-
Maccagno et al. [6], in the study of pass-by-pass AGS lization with the AGS evolution model being used in
evolution associated with oval round (or roundoval) hot strip rolling [1 3]. The weakness of the area strain
pass rolling sequence, calculated the strains by the model for the analysis of rod (or bar) rolling was
simple multiplication of a constant factor to the area discussed after this.
strains obtained by taking the natural logarithmic of The calculated evolution of AGS during the four-
the ratio of the fractional reductions in the cross-sec- pass bar rolling employing the analytical model was
tional area through the rolling stands. The strain at a then compared with that obtained from the hot torsion
pass was assumed to be a constant factor of 1.7 and 2.5 test performed under similar conditions (strains, strain
times the area strain for the roughing and finishing rates, temperature and interpass times) to those ex-
stands, respectively. They simply attributed these fac- pected in the four-pass bar rolling. Two reheat condi-
tors to represent the ratio of the nominal strain to tions were used to produce either fine (38 mm) or coarse
redundant strain related with the profile changes of the (92 mm) initial AGSs prior to deformation.
workpiece during oval round and/or round oval pass
rolling.
Kemp [7] proposed that the strain per pass should be
factors of 1.5 2 times the area strains in the roughing 2. Analytical model of the strain and strain rate in a
stands and factors of 2 3 times the area strains in all pass
subsequent stands. In the work of Maccagno et al. [6]
The strain in a pass is defined as the maximum mean
and Kemp [7], however, a mathematical rationale for
effective plastic strain in a given pass. The strain rate in
the use of the constant factors was missing and a
a given pass will be defined later. The following two
modeling approach for the other thermo-mechanical
sections give the pre-requisite steps for the calculation
parameters (strain rate and temperature evolution of
of the strain and strain rate per pass.
the material), which is obviously critical for determin-
ing the recrystallization behavior, was not given. Leh-
nert and Cuong [8] proposed a model that calculates
2.1. Prediction of cross-sectional shape
the strain and strain rate in rod rolling, based on the
assumption of a plane strain state of deformation. The
To determine the pass-by-pass strain and strain rate,
three-dimensional deformation zone was subdivided it is first necessary to predict the material profile of the
into longitudinal strips of equal width in the roll gap stress free surface that does not contact the roll directly
direction and each strip was analyzed separately. No at the exit of roll gap. Recently, Lee et al. [9] developed
experimental verification, however, was given to the an analytical model, which is robust and non-iterative
assumption introduced in the formulation of the model. in computation, to predict the surface profile of the exit
This approach might be applicable once the exit cross- cross-section of a workpiece for the oval round (or
sectional shape at a pass can be correctly predicted. round oval) pass sequence, the most widely used pass
In this paper, an analytical model is proposed which sequence in the present rod (or bar) mills. The advan-
calculates the strain and strain rate at a pass associated tage of this model is that only the geometric consider-
with the rod rolling process. The validity of the analyt- ation is required in the formulation. Hence, it simplifies
ical model proposed was examined by using a four-pass the problem of obtaining the final rolled shape greatly.
warm plate and bar rolling experiments, and by observ- Fig. 1 shows the predicted and measured cross-sec-
ing whether the mechanical behaviors of specimens tional shape of the workpiece after four-pass bar
obtained from the plate rolling experiment are consis- rolling. A bar of 28 mm diameter was passed through
tent with those obtained from the bar rolling experi- four passes to produce a round shape of approximately
ment. The pass schedule for bar rolling has been 19 mm diameter. Small differences are noted between
designed using the proposed analytical model and equa- the cross-sectional shapes predicted and those mea-
tions employed in the plate rolling theory such that the sured. The roll gap at each pass was set up slightly
material (workpiece) during bar and plate rolling expe- larger than expected, which could be due to the surface
riences the same amount of strain at each pass. The layers being oxidized by air and scaled off after air-
purpose of warm rolling (650 C) is to make the mate- cooling the hot workpiece. Overall, the predicted cross-
rial free from the phase transformation during rolling sectional shapes of the workpiece are in good
and/or the interpass time. agreement with those measured.
Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189 179

2.2. Determination of equi6alent rectangular


cross-section

Since the cross-sectional shape of the deformed


workpiece at a pass is not rectilinear, one encounters
difficulties in formulating the equation for the calcula-
tion of the strain within a frame of Cartesian coordi-
nates. To overcome this problem, the equivalent
rectangle approximation method that transforms the
non-rectangular cross-section into a rectilinear one was
employed.
There are many ways of achieving this objective
while the net cross-sectional area is maintained. Main-
taining the cross-sectional area is presumed to be a
necessary, but not a sufficient requirement for equiva-
lence. Fig. 2 describes the following three methods used
to establish equivalent rectangular sections: method of
Fig. 1. Comparison of the cross-sectional shapes of deformed work- maximum height; method of maximum width; and the
piece predicted by the surface profile prediction model and those method of width height ratio. The differences among
obtained by the experiment for the four-pass bar rolling sequence.
these methods are evident from their particular names.

Fig. 2. Three methods of establishing an equivalent rectangular section for the oval pass (a) and round pass (b). (h) Method of width height ratio;
(i) method of maximum width; and (k) method of maximum height.
180 Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189

Fig. 3. (a) Schematic description of round oval pass rolling (front view). (b) Application of equivalent rectangle approximations method to
round oval pass rolling and effective roll radius, Reff, and equivalent cross-sectional heights, H( i and H( p, and equivalent cross-sectional widths,
W( i and W( p.

In the case of the method of width height ratio, the 2.3. Strain at a pass (pass-by-pass strain)
definition of width and height might be arbitrary. Thus,
this method was discarded. The method of maximum The strain defined as the maximum average effective
height might not be acceptable from the point of view plastic strain at a given pass can then be calculated
of the treatment of a spread change of the workpiece from the rectilinear shape transformed. The calculation,
during rolling and the calculation method of the rolling however, should include the non-linear change of
speed at a pass. When the cross-sections are approxi- draught, spread and elongation of the material being
mated into rectangles, the method of maximum height deformed. The assumption introduced to overcome this
indirectly reflects the spread increment (or decrement) problem is the hypothesis of parallelepiped deformation
of the outgoing workpiece, due to the size change of an [10]. According to this hypothesis, a cube of material
incoming workpiece. In the meantime, the method of subjected to a load will change to a rectangular prism
maximum width can directly reflect this. In addition, and its angle and sides will remain orthogonal with this
the effective roll radius necessary for calculating the prior to deformation. It should be noted that the
rolling speed at a pass has been calculated from the calculation is focused on the bulk deformation of the
method of maximum width [10]. Therefore, the cross- workpiece.
sections are approximated into rectangles using the With the elastic and shear strain components ne-
method of maximum width. glected, the incremental plastic deformation along each
Fig. 3 shows an example of the method of maximum principal axis can be assumed proportional, i.e.
width applied to the round oval pass rolling in which dm1:dm2:dm3 = m1:m2:m3. (1)
an incoming stock is of a round shape and a stock with
an oval shape is produced. Sections A C correspond to From the volume constancy condition,
the positions where a section of the undeformed work- m3 = m1 m2. (2)
piece is about to be rolled, being deformed and that of
Then, the strain at a given pass (i.e. section C in Fig. 3),
the deformed workpiece leaving the roll groove, respec-
mp, is expressed as
tively. The equivalent rectangle approximation de-
scribed in Fig. 3 can also be applied to other type of
mp =
 2 2
(m + m 22 + m 23)
n
1/2
=
2     n
m2 1+
m1 2
+
m1 1/2
,
passes, such as oval round, box oval and diamond 3 1
3 m2 m2
square pass. (3)
Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189 181

where 3. Experimental procedure for validating the analytical


model
m1 = ln(W( i/W( p) and m2 =ln(H( i/H( p). (4)
Eq. (3) is a closed form solution for the strain, based 3.1. Rolling equipment and specimen preparation
on the assumption of a parallelepiped uniform defor-
mation of the material. Eq. (4) is valid under the A two-high laboratory mill with a single stand driven
assumption that the principal plastic strains in the three by 75 kW constant torque DC motor was employed.
principal axes are independent of each other. The posi- DCI (Ductile Casting Iron) rolls were used, with 310
tive sign is defined when the workpiece contracts. Thus, mm maximum diameter and 320 mm face width. The
the sign of m2 is always positive due to the contraction rolling speed was 34 rpm. A box type furnace was used
along the y-axis, while the sign of m1 is negative due to to heat the specimens to the desired rolling tempera-
the extension along the x-axis. Eq. (3) illustrates that if ture. Low carbon steel with a chemical composition of
the ratio (m1/m2) is negligibly small, then the deformation Fe 0.1 wt.% C0.45 wt.% Mn0.25 wt.% Si was used.
behavior of the workpiece in the roll groove can be The material was obtained in the form of square as-cast
approximated by the plane strain condition, generally billet with a side length of 160 mm. The specimens to
used in plate rolling. be rolled were cut and machined into a bar with 28 mm
diameter and 300 mm length for the bar rolling experi-
2.4. Strain rate at a pass (pass-by-pass strain rate) ment and a plate of 28 mm thickness, 140 mm width
and 300 mm length for the plate rolling experiment.
As in drawing and forging processes, the strain rate
in bar (or rod) rolling changes at various stages of 3.2. Experimental procedure
deformation. The strain rate is a maximum at the
entrance to the roll (or in its vicinity) and decreases The specimens were soaked at 10 C above the
along the roll bite, finally becoming zero at the outlet. rolling temperatures for 60 min to ensure a homoge-
Therefore, it is necessary to introduce an average nous temperature distribution. They were rolled when
strain rate for a given pass. the center temperature of the specimens reached the
The average strain rate can be defined as the strain desired rolling temperature (650 C). For each subse-
over a time interval, which can be calculated from quent pass, the specimens were reheated for 20 min
before being rolled. There are two reasons for choosing
m; p = mp/tp, (5) this temperature: (i) to avoid any possible phase trans-
formation during the experiment; and (ii) to consider
where tp represents the time interval taken for section A the power limit of the pilot rolling mill, i.e. the rolling
to pass through to section C (Fig. 3). Hereafter the experiment below 600 C cannot withstand the pilot
average strain rate is referred to as the strain rate. The mill power. The primary reason for 20 min reheating
time interval can be expressed as between passes is to provide the specimen some soften-
60L( ing between passes. A thermocouple (Type K) of 1.6
tp = [s] (6) mm diameter was embedded in 50 mm deep holes
2pNReff
drilled in the tail ends of the specimen to measure the
where Reff, L( and N are, the effective roll radius, the rolling temperature. After rolling, the specimens were
effective projected contact length of the grooved roll air cooled to room temperature.
and workpiece, and roll rpm at a given pass, respec-
tively. The effective roll radius at a given pass is 3.2.1. Plate rolling
calculated by the method of maximum width [10]. The Table 1 shows the pass schedule designed for the
effective projected contact length is expressed as plate rolling experiment. The strains and strain rates

L( =
'! Rmax

H( p G " (H( i H( p). (7)
are computed from the equations used in the plate
rolling theory [10].
2
3.2.2. Bar rolling
Table 1
Pass schedule for the plate rolling experiment
Table 2 shows the pass schedule for the bar rolling
experiments. A bar with 28 mm diameter was passed
Pass no. Roll gap (mm) Strain mp Strain rate m; p (s1) through four passes to produce a round shape of
approximately 19 mm diameter. As shown in Fig. 1, a
1 19.50 0.418 5.496 specimen was first rolled into the oval pass at the
2 14.02 0.381 6.244
3 10.12 0.377 7.323
desired temperature. Then, the workpiece was rolled
4 7.41 0.360 8.393 into the round pass after it was rotated 90 along the
rolling direction. The loading sequence in the bar
182 Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189

Table 2 For comparison purpose, the stressstrain curve of


Pass schedule for the bar rolling experiment
as-rolled material was also obtained. The profiles of
Pass type and Roll gap (mm) Strain mp Strain rate
the stressstrain curves at the four passes are slightly
no. m; p (s1) different. At each pass, however, the constitutive
relation of the plate-rolled specimen is similar to the
Oval (1) 4.0 0.418 6.524 bar-rolled specimen. Ultimate tensile stresses obtained
Round (2) 4.0 0.381 5.529 from the bar-rolled specimen are somewhat larger
Oval (3) 4.0 0.377 6.866
Round (4) 4.0 0.360 5.937
than those acquired from the plate-rolled sample,
but the differences are small. Overall, Fig. 6 shows
that both specimens gave very similar results, prior
rolling experiment can be characterized as a kind of to necking. It is interesting that, after necking, the
cross-loading, while that in plate rolling is monotonic elongation of the bar-rolled sample is larger than that
loading. It should be noted that the strain rates in plate of the plate-rolled sample. This might be attributed to
rolling are somewhat different from those in bar rolling the difference of grain shapes (layered or polygonal) in
because of the roll groove in bar rolling. It is believed, the samples and/or a different deformation manner
however, that such a small difference in the low strain arising at the different cross-sectional shapes (cylindri-
rate range has little effect on the experimental results. cal or rectangular) of the samples. These factors may
lead to a different failure strain and consequently a
3.3. Sample preparation dissimilar behavior of plastic instabilities after necking.
From this it is proposed that prior to necking, the
For mechanical tests, samples with a cylindrical specimens (plate shape and bar shape) have experi-
shape (6 mm in diameter and 50 mm in gauge length) enced almost the same amount of plastic work at each
and a plate shape (25 mm in width, 7.41 mm in pass.
thickness and 50 mm in gauge length) were machined We can also see that although the microstructures
from the center section of the rolled materials along the obtained by plate and bar rolling are different, they all
direction of rolling. The samples were machined ac- result in essentially the same room temperature tensile
cording to ASTM E8 Full Size [11]. properties. This implies that even though the grain
shapes of microstructures obtained by plate and bar
rolling are different, the effect of difference of grain
3.4. Experimental obser6ations shape on the tensile properties may diminish by soften-
ing during rolling and cooling to room temperature.
3.4.1. Microstructures From this, it can be deduced that almost the same
Figs. 4 and 5 show the optical micrographs of mi- amount of strain was imposed in both the plate and
crostructures at the mid-section of the specimen normal bar rolling experiments. This suggests that the ratio-
to the rolling direction for plate and bar rolling, respec- nale for the analytical model developed in this study is
tively. The arrows represent the loading direction ap- supported by the experiments.
plied to the specimens at each pass. The grain shape
illustrates the qualitative features of the difference of
strain path (loading direction) attributed to monotonic
compression (plate rolling) and cross-compression (bar 4. Numerical simulation of recrystallization behavior
rolling). and AGS evolution
It is apparent that the microstructures of the material
after plate rolling are composed of the elongated grains The deformation parameters (strain and strain rate)
oriented along the roll axis (gap) direction (Fig. 4), for the four-pass rolling sequence (Fig. 1) are computed
while some of those obtained by bar rolling (Fig. 5) by Eqs. (3) and (6). The temperature variation during
have a more equi-axed shape after the second pass. This rolling was assumed isothermal, because the primary
indicates that the orthogonal changes in the deforma- aim of this simulation is to see the effect of the method
tion (loading) path during bar rolling effectively lead to for calculating the strain and strain rate per pass on the
more equi-axed grain shapes. recrystallization behavior and evolution of AGS during
rolling. Table 3 describes the deformation parameters,
3.4.2. Stress strain cur6e interpass time and temperature of the workpiece used in
Fig. 6 illustrates the tensile stress strain curve of the numerical simulation.
samples produced by the plate rolling and bar rolling A series of numerical simulations was carried out
experiments. The tensile test was carried out on an using the recrystallization and AGS evolution model
Instron testing machine with the cross-head speed of being used in hot strip (or plate) rolling. The equations
0.016 mm s 1. used in this study are listed in Appendix A.
Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189 183

5. Results and discussion model proposed in this study. The strains calculated
by the area strain model show some difference to the
5.1. Deformation parameters area strains multiplied by constants. This is because
the area strain model does not consider the cross-sec-
Fig. 7(a) and (b) shows the strains and strain rates tional shape change of the workpiece and is merely
calculated by the area strain model and the analytical associated with a constant factor which compensates

Table 3
Pass schedule for the simulation of AGS during bar rolling when the finishing rolling speed is 1.0 m s1

Pass type and no. Strain mp Strain rate m; p (s1) Inter standard distance (mm) Temperature (K)
Oval (1) 0.418 6.524 1173
Round (2) 0.381 5.529 1500 1173
Oval (3) 0.376 6.866 1500 1173
Round (4) 0.360 5.937 1500 1173
Pass type and no. Rolling speed (m s1) Inter. pass time (s) Sectional area (mm2)

Oval (1) 0.582 2.577 491


Round (2) 0.723 2.075 407
Oval (3) 0.859 1.746 333
Round (4) 1.000 1.500 286

Fig. 4. Optical micrographs of microstructures at the middle section normal to the rolling direction for plate rolling. Arrows represent the loading
direction.
184 Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189

Fig. 5. Optical micrographs of microstructures at the middle section normal to the rolling direction for bar rolling. Arrows represent the loading
direction.

the profile changes of the workpiece in groove rolling, crystallization occurs during a given pass if the strain
regardless of a pass type such as oval round pass or calculated at this pass exceeds a critical strain
round oval pass.
mc = 5.6 10 4d 0.3
0 Z
0.17
, (8)
The strains calculated from the analytical model at
the round pass (no. 2) are very close to the area strain where d0 is initial grain size (mm) and Z is ZenerHol-
multiplied by 2.0 but those at the round pass (no. 4) lomon parameter (s 1) that is the function of the
approaches the area strains multiplied by 2.5. strain rate and temperature (see Appendix A). This
Meanwhile those at oval pass (nos. 1 and 3) are implies that the deformation parameters and tempera-
between the area strains multiplied by 1.7 and 2.0. A ture at a pass affect the recrystallization behavior sig-
similar pattern was observed for the strain rates (Fig. nificantly and this, in turn, influences the AGS during
7(b)). rolling.
Fig. 8 illustrates the strains calculated by the analyti-
5.2. Recrystallization beha6ior cal and the area strain model, and the critical strains at
a given pass when the finishing rolling speed is 1
According to Maccagno et al. [6], metadynamic re- m s 1. The recrystallization behavior at a given pass is
Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189 185

considered by comparing the strains calculated with the


critical strain. As shown in Fig. 8(a), e.g. static recrys-
tallization takes place at pass no. 1, while metadynamic
recrystallization occurs at the passes 2 4, respectively.
As a constant factor multiplied to the area strain
increases, the number of passes where the calculated
strains becomes larger than the critical strain increases.
In Fig. 8(d), metadynamic recrystallization occurs at
every pass. Fig. 8(a) (analytical model) and (c) (area
strain*1.7) show a similar pattern of recrystallization
behavior at all pass numbers except at pass no. 4.
Fig. 9 shows that the calculated strains and critical
strains at a given pass when the finishing rolling speed
is increased to 20 m s 1. The critical strains increase
with the rolling speed increment (i.e. the strain rate)
because the Zener Hollomon parameter, Z, increases.
Fig. 9(a) (analytical model) and Fig. 9(b) and (c) (area
strain*1.7 and 2.0) show an analogous pattern at all
passes. This implies that using the area strain model
multiplied by a constant factor might have problems
when the rolling speed varies for the same process.
Therefore, it can be deduced that the area strain model
for the prediction of pass-by-pass strain does not have
a mathematical base to be used in the prediction of the
recrystallization behavior and the AGS evolution in
rod (or bar) rolling.

Fig. 7. Comparison of deformation parameters for the four-pass bar


rolling sequence calculated by the analytic model and the area strain
models. (a) Strains. (b) Strain rates.

5.3. E6olution of AGS

An additional aspect of this study is to compute the


microstructure evolution of austenite during bar (or
rod) rolling and then to compare with that predicted
using the analytical model under similar conditions
(temperature, strains, strain rates and interpass times)
to those expected in bar rolling. It is, however, difficult
to measure the AGS of the material directly during hot
bar rolling. Hence, a hot torsion simulation was used
instead. The torsion test specimens with a gauge section
of 20 mm length and 6.7 mm radius were machined.
Two reheat conditions were used to produce either fine
(38.0 mm) or coarse (92.0 mm) initial AGSs prior to
Fig. 6. Tensile stress strain behaviors of low carbon steel after plate deformation. Tests were conducted in the strain rate of
rolling and bar rolling experiment. 6 s 1. The water-quenched specimens after tests were
186 Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189

polished with 1 mm diamond paste and then etched Differences are noted at pass no. 1 for the fine initial
chemically with a solution of saturated picric acid and grain size and the third pass for the coarse initial grain
Teepol to reveal the former austenite microstructure on size. After pass no. 1, however, the predicted AGS is
cross-sectional planes just below the surface of the close to that measured although generally finer. There-
deformed specimens. fore, it can be deduced that the proposed analytical
Tables 4 and 5 illustrate the AGS calculated from the model for computing the pass-by-pass strain, coupled
analytical model at low rolling speed (1 m s 1) and the with the recrystallization behavior and AGS evolution
measurements from the hot torsion simulation. It shows models, might be applicable for rod (or bar) rolling.
the pass-by-pass AGS calculated and measured and the This is one of the first times that a model of this type
deformation conditions used in the hot torsion simula- has been critically compared with multi-deformation
tion where the same strains and interpass times, but a data; particularly with data involving short interpass
constant strain rate of 6 s 1 were maintained. The times. This is an extremely sensitive test of the models.
grain size was determined from specimens quenched The interesting point to note is that the microstructure
either immediately before a given pass or l.3 s after the is essentially the same for the two different starting
last pass. A reticule of five parallel lines was used to conditions by the time the steel enters the final pass and
measure the number of intercepts on ten random fields. 1.3 s after the last pass. However, the model predicts
The total length of the lines was divided by the number that the same microstructure should be obtained by the
of intersections, as described in the ASTM standard entry to the third pass whereas different grain sizes
E112. On considering the confidence limits of the mea- were actually observed at this pass.
surements, we measured the grain sizes to one decimal It has been suggested by a number of workers, but
place. Measuring the grain size to two decimal places is particularly in the pioneering work by Sellars and co-
not necessary unless the gain size is less than 1.0 mm. workers [1], that the effect of the initial grain size is

Fig. 8. Comparison of critical strains and strains calculated for the four-pass bar rolling sequence. (a) Analytic model. (b) (d) Area strain model
when finishing rolling speed is 1.0 m s 1.
Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189 187

Fig. 9. Comparison of critical strains and strains calculated for the four-pass bar rolling sequence. (a) Analytic model. (b) (d) Area strain model
when finishing rolling speed is 20.0 m s 1.

removed quickly in multi-pass rolling. The current Table 4


work suggests that this may not be the case to a large Evolution of AGS when the finishing rolling speed is 1.0 m s1
initial grain size and more work is required to explore
Pass no. Measured Predicted
this under a wider range of deformation scenarios.
It should also be noted that the comparison of AGS Before first pass 38.0 38.0
at each pass, as shown in Tables 4 and 5, is indirect Before second pass 20.8 25.3
since the strain, strain rate and temperature used in hot Before third pass 16.7 16.9
torsion test was assumed to be equivalent to those in Before fourth pass 19.6 16.3
After fourth pass 19.6 16.2
rod rolling. Therefore, further studies are necessary to
measure the AGS of the material directly during Initial grain size: 38 mm.
rolling.

Table 5
6. Concluding remarks Evolution of AGS when the finishing rolling speed is 1.0 m s1

Pass no. Measured Predicted


The aim of this work was to develop an analytical
model to compute the pass-by-pass strain associated Before first pass 92.0 92.0
with the rod (or bar) rolling process because the occur- Before second pass 42.6 39.1
rence of dynamic recrystallization is purely determined Before third pass 24.7 16.9
Before fourth pass 20.0 16.3
by comparing a pre-determined critical strain for the After fourth pass 19.8 16.2
initiation of dynamic recrystallization with the strain
calculated at a given pass. The validity of the analytical Initial grain size: 92 mm.
188 Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189

model was examined using a four-pass plate and bar amount of the strain was given to the specimen at
rolling experiments, and comparing the constitutive each pass. Consequently, it can be deduced that the
relation of specimens acquired from the late rolling analytical model for the prediction of pass-by-pass
experiment with those obtained from bar rolling. strain is appropriate for the analysis of rod (or bar)
A numerical simulation was then performed using rolling. (2) The area strain model multiplied by a
the area strain model and the analytical model devel- constant factor lacks mathematical rationale to be
oped, focusing on the effect of the method to calculate used as input for recrystallization model because it
the strain on the recrystallization behavior and AGS does not adequately model the change of roll groove
evolution for the four-pass oval round (or round profile (oval-to-round or round-to-oval). (3) Without
oval) rolling sequence. The predicted AGS was also an adequate model of the strain, and thereby strain
compared with that obtained from the hot torsion test. rate, it is not possible to accurately model the recrys-
The conclusions are summarized as follows. (1) Both tallization behavior and the resultant AGS in rod (or
rolling experiments demonstrated that almost the same bar) rolling.

Appendix A. Equations and constants used in numerical simulation [6]

Model Equation Parameters

Critical strain mc =5.6104d 0.3


0 Z
0.17
d0 = 92 or 38 mm

Static recrystallization 
x = 1exp 0.693
 n t n
N= 1
t0.5

t0.5 =t0mpd q0 exp


 
QRXN
t0 = 2.31015; p= 2.5;
q= 1; QRXN = 230

 
RT kJ mol1
ZenerHollomon parameter Qdef
Z = m; exp Qdef = 300 kJ mol1;

  n
RT R= 8.31
1.5
Metadynamic recrystallization t

 
x = 1exp 0.693 ; kmd = 1.1; nmd = 0.8;
t0.5
Qmd Qmd = 230 kJ mol1
t0.5 =kmdZ nmd exp
RT
Recrystallized grain size
-Static
 
dSRX =Amad b0 exp
45000
A = 343; a = 0.5; b=0.4
8.31T
-Metadynamic dMDRX =kZ0.23 k= 2.6104

Grain growth
static
(If tip\1 s)
d m =d m
SRX+k(tip4.32t0.5) exp

400000  m= 7; k= 1.51027

 
8.31T
400000
d m =d m
MDRX+k(tip2.65t0.5) exp m= 7; k= 8.21025
8.31T
-Static (If tipB1 s)
d m =d m
SRX+k(tip4.32t0.5) exp
113000   m= 2; k= 4.0107

-Metadynamic d m =d m
MDRX+k(tip2.65t0.5) exp
8.31T
113000   m= 2; k= 1.2107
8.31T
Partial RXN d0i+1 =X 4/3 2
i dRXi+(1Xi ) d0i ;
mai+1 =mi+1+(1Xi )mi
Y. Lee et al. / Materials Science and Engineering A336 (2002) 177189 189

References [6] T.M. Maccagno, J.J. Jonas, P.D. Hodgson, ISIJ Int. 36 (1996)
720 728.
[7] I.P. Kemp, J. Iron Making Steel Making 17 (1990) 139 143.
[1] C.M. Sellars, Mater. Sci. Technol. 6 (1990) 1072 1081. [8] W. Lehnert, N.D. Cuong, ISIJ Int. 32 (1995) 1100 1108.
[2] T. Sakai, J.J. Jonas, Acta Metall. 32 (1984) 189 209. [9] Y. Lee, S. Choi, Y.H. Kim, J. Mater. Proc. Tech. 81 (2000)
[3] P.D. Hodgson, R.K. Gibbs, ISIJ Int. 32 (1992) 1329 1338. 87 96.
[4] K. Karhausen, R. Kopp, M.M. De Souza, Scan. J. Metall. 20 [10] Z. Wusatowski, Fundamentals of Rolling, Pergamon Press, Lon-
(1991) 351 363. don, 1969, pp. 107 109.
[5] J. Yanagimoto, T. Ito, J. Liu, ISIJ Int. 40 (2000) 65 70. [11] ASTM standard E8 full size, 1978

You might also like