Thermal Stress Evolution of The Roll During Rolling and Idling in Hot Strip Rolling Process

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/263474175

Thermal Stress Evolution of the Roll During Rolling and Idling in


Hot Strip Rolling Process

Article  in  Journal of Thermal Stresses · May 2014


DOI: 10.1080/01495739.2014.913418

CITATIONS READS

5 467

3 authors, including:

Youngseog Lee
Chung-Ang University
90 PUBLICATIONS   569 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Roll pass design View project

2012R1A1 A2006433 View project

All content following this page was uploaded by Youngseog Lee on 06 March 2019.

The user has requested enhancement of the downloaded file.


Journal of Thermal Stresses, 37: 981–1001, 2014
Copyright © Taylor & Francis Group, LLC
ISSN: 0149-5739 print/1521-074X online
DOI: 10.1080/01495739.2014.913418

THERMAL STRESS EVOLUTION OF THE ROLL DURING


ROLLING AND IDLING IN HOT STRIP ROLLING PROCESS

Doo-Hyun Na1 , Chang-Ho Moon2 , and Youngseog Lee1


1
Department of Mechanical Engineering, Chung-Ang University, Seoul, Korea
2
Engineering Research Center, Technical Research Laboratories, POSCO,
Korea

This study examined a critical idling time for the roll to attain a target temperature
prior to rolling the next strip in the hot strip rolling process. Thermoelastic finite
element simulations were performed to compute the temperature evolution and resulting
thermal stress variations of the roll during rolling and idling. The results showed that
a circumferential compressive stress was generated in the roll during rolling. However,
circumferential tensile stress on the roll surface was produced very shortly (about
0.76 s) and reached about 68 MPa around 28 s after the beginning of idling. The
temperature distribution of the roll’s 1.5-mm-thick outer layer was homogenized and
reached a steady state approximately 28 s after the onset of idling. Hence, we may cut
down the long idling time (90–120 s), generally adopted in a hot strip rolling process,
to decrease (by 68.9–76.7%) the roll temperature elevated during rolling to a target
temperature.

Keywords: Finite element simulation; Hot strip rolling process; Idling time; Initial roll temperature;
Thermal stress

INTRODUCTION
In the hot strip rolling process, a hot slab (1100–1200 C) is processed into
a sheet with acceptable dimensional tolerance as it passes through the stands; the
slab thickness is progressively reduced by the work rolls. Here, “stand” denotes a
unit machine with work rolls, backup rolls, a screw-down device, housing to contain
these parts, and a drive motor. The slab is called a “strip” once it is rolled into
the roughing stand. Hereafter, the work roll is simply referred to as the “roll” for
convenience.
A roll with multiple revolutions experiences cyclic thermal compressive and
tensile stresses along its circumferential direction since a roll surface gains heat
by contact with the hot strip and loses heat by water and air cooling during a
revolution. The micro cracks on the roll surface open due to the repetitive thermal
tensile stresses. As a result, the surface quality of the rolled product is degraded,

Received 13 August 2013; accepted 7 March 2014.


Address correspondence to Youngseog Lee, Department of Mechanical Engineering, Chung-Ang
University, Seoul 156-756, Korea. E-mail: ysl@cau.ac.kr
Color versions of one or more of the figures in the article can be found online at www.
tandfonline.com/uths.

981
982 D.-H. NA ET AL.

and the rolling operation should be stopped to change the rolls. This decreases
the operating hours (period of time that the rolls are in operation) of the rolls.
Therefore, understanding the temperature evolution and resulting thermal stress
variations in the roll is important to prohibit the growth of micro cracks and thus
increase the roll life.
Thermomechanical phenomena of the roll in the hot rolling process are
influenced mainly by i) water cooling parameters (spray angle, number of nozzles,
nozzle type, water pressure, and distance from a nozzle to the roll surface) and ii)
rolling parameters (strip thickness, reduction ratio, strip temperature, and rolling
speed). Many research groups have investigated the effect of these parameters on
the temperature evolution and resulting thermal stress variations in the roll.
Sun et al. [1] conducted thermomechanical analysis of the roll under the
assumption of steady-state rolling conditions using the finite element method and
examined the effect of water cooling parameters. However, their results are not
applicable to multiple revolutions of a roll. Other researchers [2–5] predicted the
temperature evolution and thermal stress variations in the roll under the assumption
of non-steady-state rolling conditions for multiple rotations using the finite element
method. Saboonchi and Abbaspour [6] employed the 2D finite difference method
to solve heat transfer differential equations of the roll with varying boundary
conditions such as the spray angle, angle of spray impact on the roll, water pressure,
and nozzle type. Serajzadeh [7] combined finite element analysis and boundary
element technique to compute the temperature field in the roll of a single stand
for reversing-type rolling. Other research groups [8–12] solved the heat transfer
differential equation of the roll using analytical methods. They replaced the complex
boundary and initial conditions with simpler ones and significantly reduced the
computation time.
However, these studies [1–12] were limited to predicting the changes in
temperature and/or thermal stress in the roll for a single rotation of the roll and/or
multiple revolutions of the roll during rolling and/or interpass times. Here, a pass
means a single rolling action, i.e., passing a material (strip) through the rolls. The
roughing mill of the hot strip rolling process has a single-stand reversing-type rolling
mill. Hence, there is an interpass time (about 10–15 s; dependent upon the roll
diameter, transportation speed of the roller table, and motor capacity) to reverse the
rotation of the rolls for further thickness reduction of the strip.
The concept of intervals exists for a finishing mill that consists of seven stands
in tandem. There is an interval between the tail of the strip (i) and head of the
strip (i + 1) (see Figure 1). During this interval, the rolls revolve under no loading.
The interval (approximately 90–120 s) depends upon the roll speed and layout of
the hot strip mill. This interval is usually called idling at the production site. The
primary aim of idling is to reduce the roll temperature raised during rolling to a
target temperature before the next strip enters between the rolls. Unless otherwise
stated, idling indicates an interval between two strips for the purposes of this article.
The target temperature indicates the initial roll temperature before rolling begins.
In this study, we have investigated the temperature evolution and resulting
thermal stress variations at different depths from the roll surface toward the roll
center during both rolling and idling operations and examined the critical idling
time for the roll between strips in the hot strip rolling process. A fully coupled
thermoelastic finite element (FE) model of the roll at the last stand (Std. No. 7)
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 983

Figure 1 Schematic of rolling and idling at a stand in the finishing mill. Roll velocity denotes
peripheral velocity of the roll. (a) Rolling ends at a stand and idling begins; (b) idling; and (c) idling
is completed and rolling starts at the stand.

of the finishing mill was established. The FE model was validated by comparing
the roll surface temperature computed in this study with the experimental results
reported by Stevens et al. [13], who conducted a pilot hot strip rolling test. The
element size in the proposed FE model was determined using the mesh test. Based
on the FE simulation results, we investigated whether or not the amount of idling
time (90–120 s) generally adopted in the hot strip rolling process is appropriate.
We suggest a critical idling time that decreases the roll temperature elevated during
rolling to the initial roll temperature. The effect of the initial roll temperature on
the change in the stress state of the roll during rolling and idling was also examined.

PROBLEM FORMULATION
Rolling Conditions of Hot Strip Rolling Process
In the hot strip rolling process, the final products (1.2–12.0 mm in thickness)
are produced throughout the finishing mill, which consists of seven stands in
tandem. In the production site, the rolls at the last stand (Std. No. 7) are replaced
most frequently because the surface quality of the rolled product is noticeably
influenced by the state of the roll surface at Std. No. 7. Therefore, we chose the
roll of Std. No. 7 to examine the temperature evolution and resulting thermal
stress variations. The heat conduction from the hot strip to only the work roll was
984 D.-H. NA ET AL.

Figure 2 (a) Water-cooling nozzles installed around the upper (work) roll and strip being rolled at
Std. No. 7 of the finishing stand unit of an actual hot strip mill. C1–C4 indicate the regions where the
water-impinging jet sprayed from each nozzle comes into contact with the roll surface. (b) Angular
zones into which the roll was divided by heat transfer characteristics. r ∗ denotes the distance from the
roll center to a point; within this distance, heat conduction during rolling is insignificant.

modeled by using a convective-type boundary condition. The backup roll was not
considered because its effect on the roll surface temperature was small [1].
Figure 2(a) shows a set of water-cooling nozzles mounted around the upper
roll and strip being rolled at Std. No. 7. The lower roll and half of the strip were
omitted for convenience. C1 denotes the forced convective cooling zone where water
sprayed from nozzle No. 1 comes in contact with the roll surface. The same is true
for C2–C4. Table 1 provides the locations of the four nozzles, spray angle, amount
of water flux coming out of each nozzle, and distance from a nozzle to the roll
surface. The roll diameter and strip velocity were 700 mm and 11.7 m/s, respectively.
The entry and exit thicknesses of the strip are 1.62 and 1.48 mm, respectively. The
temperatures of the strip and water are 860 and 16 C, respectively.
Figure 2(b) illustrates three angular zones with differing surface cooling
conditions as the roll revolves. The roll surface gains heat by contacting with the
strip and loses heat by convection and radiation to the surrounding air. In the water-
cooling zone, the roll temperature drops rapidly due to water sprayed from the
nozzle. For each revolution, the roll goes through three boundary zones (contact,
air cooling, and water cooling) in sequence. The contact zone is excluded during
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 985

Table 1 Locations of four nozzles, spray angle, water flux, and distance between nozzle and roll
surface at Std. No. 7

Angle (deg.)

Nozzle no. 1 2 Water flux (liter/sec) Distance from the roll surface (mm)

1 262.0 298.5 0.8788 186.2


2 248.4 262.7 0.8788 174.8
3 217.0 248.2 0.8788 217.5
4 91.3 117.0 0.4513 250.3

idling; instead, the roll repeatedly passes through two boundary zones (air cooling
and water cooling).

Thermomechanical Properties of the Roll


A high-speed steel (HSS) roll is installed at Std. No. 7 and divided into two
parts: the shell and core. The core is covered by the shell in the circumferential
direction. A relatively hard and wear-resistant material is used for the shell, i.e., the
outer ring of the roll in contact with the hot strip. Inside the roll, the core is filled
with a relatively ductile and durable material.
Table 2 shows the mechanical properties of the shell and core. The core
diameter is 500 mm, and the shell thickness is 100 mm. To analyze the heat
transfer phenomena of the HSS roll, we measured the thermal conductivity, thermal
expansion coefficient, and specific heat of the shell using a thermal conductivity
tester (Laser Flash Apparatus 457, Netzsch, Germany), dilatometer (DIL 402C,
Netzsch, Germany), and differential scanning calorimeter (DSC 200 F3, NETZSCH,
Germany), respectively. The measured thermal properties of the shell at 20, 100, 200,
300, and 400 C are listed in Figure 3. The thermal properties of the core were not
measured since heat was not transferred into the core during rolling and idling.

Heat Conduction in the Roll


Ignoring heat conduction along the roll axis, i.e., the z-axis, the governing heat
conduction equation for the roll is expressed as follows:

T
kTi i = c (1)
t

Table 2 Mechanical properties of the roll shell and core [1]

Property Density Young’s modulus Poisson’s ratio

Shell 7620 kg/m3 220 GPa 0.287


Core 7100 kg/m3 160 GPa 0.275
986 D.-H. NA ET AL.

Figure 3 Measured thermal properties of the high-speed steel (HSS) roll shell at the given
temperatures: (a) conductivity, (b) thermal expansion coefficient, and (c) specific heat.

where k, , and c are the thermal conductivity, density, and specific heat,
respectively, of the roll. t is the time for rolling and idling. T indicates the roll
temperature. The boundary condition can be expressed by Eqs. (2) and (3):

T 
−k = 0 at r = r ∗ (2)
r 

T 
−k = qt at r = R (3)
r 

where r ∗ denotes the distance from the roll center to a point, which is close to
the roll surface and R indicates the roll radius. Hence, R−r ∗ means the outer layer
thickness of the roll. Heat conduction occurs within the outer layer of the roll
R–r ∗ since the heat flux from the strip into the roll is always slightly larger than the
heat flux from the roll into the cooling sprays and surrounding. qt is the heat flux
on the roll surface, which is divided into the contact, air cooling, and water cooling
zones (See Figure 2(b)). Contact zone indicates the region where the temperature on
roll surface layer is risen due to physical contact between the roll and hot strip. Air
cooling zone implies the area where the temperature on roll surface layer drops by
natural convection.
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 987

Descending temperature in the air cooling zone is smaller than that in the
contact zone. Water cooling zone is the region where the temperature on roll surface
layer falls rapidly by water cooling. In this way, the roll surface layer is subject to
quick temperature change due to the alternating contact with the hot strip and the
cooling water in every rotation. On the other hand, there is no temperature rise in
the roll surface layer during idling since there is no a direct contact between the roll
and hot strip. The heat flux on the three contact zones are described as follows:
In the contact zone,
1
qt = hc T − Ts  − fric t − ¯ t  (4)
2
where fric = −
n gr  (5)

hc and Ts are the interface (contact) heat transfer coefficient and the surface
temperature of the strip. The second term on the right-hand side of Eq. (4) denotes
heat due to the friction between the strip and roll. It is assumed that half of the heat
flows into the roll. t and ¯ t are the tangential component of velocity vector of the
strip and roll, respectively. Hence, vr is the relative tangential velocity between the
strip and roll. fric ,
n and are the tangential (friction) stress, normal stress, and the
Coulomb friction coefficient, respectively. Note when 
n  is greater than the shear
yield stress of the strip, sticking friction occurs. The function gvr  in Eq. (5) was
introduced by Chen and Kobayashi [14] to deal with both the sliding and sticking
friction in the contact zone:
v
gr  = − r (6)
vr 

However, for a given normal stress, the tangential (friction) stress in Eq. (5) has a
step function behavior upon the value of vr . The discontinuity of the value of fric
results in no converged solution.
In order to overcome this difficulty, Chen and Kobayashi modified the
function gvr  in the followings
2 
gr  = − tan−1 r (7)
b
where b is a positive constant that is several orders less than the roll velocity so
that the function gvr  in Eq. (7) is almost identical to that in Eq. (6) for vr 
slightly larger than zero. Note that the magnitude of b has almost no effect on the
magnitude of tangential (friction) stress if the value of b is several orders less than
the roll velocity [1, 14]. The role of very small positive number b is to avoid poor
convergence.
The contact heat transfer coefficient, hc , in Eq. (4) was calculated from the
equations proposed by Hlady et al. [15]:
 
k̄ Pr 1 7
hc = (8)
c1
¯ s

where
kr ks
k̄ = (9)
kr + k s
988 D.-H. NA ET AL.

Pr and
¯ s are the average contact pressure of the roll and flow stress of the strip at
the roll-strip contact surface, respectively. c1 is 0.035 mm for steel [15]. ks and kr are
the conductivity of the strip and roll, respectively.
In the air cooling zone,

qt = ha T − Te  +
T 4 − Te4  (10)

where ha and Te are the convection heat transfer coefficient in air and surrounding
temperature, respectively. In this study, the natural convection heat transfer
coefficient of air against the roll (7 8 Wm−2  C−1 ) was used [16].
and are
Stephan–Boltzmann radiation constant of black body (5 67E-8 Wm−2  C−4 ) and
emissivity of the material at a temperature around 1000 C (0.8), respectively.
In the water cooling zone,

qt = h0 CQ CP C Cd CT T − TW 
= heff T − TW  (11)

where h0 denotes the basic heat transfer coefficient (0 12 Wmm−2  C−1 ). CQ , CP , C ,


CT , and Cd are the correction parameters depending on the water flow rate, water
pressure, angle of water spray, roll surface temperature, and distance between the
roll surface and nozzle, respectively. Thus heff represents the effective convection
heat transfer coefficient reflecting all the correction parameters and the basic heat
transfer coefficient. Tw is the coolant temperature (16 C). Ginzburg et al. [17]
proposed an equation for calculating the correction parameters.
Table 3 shows the values of the correction parameters (CQ , CP , C , Cd ) and
the effective convection heat transfer coefficient hw calculated for four water cooling
zones. Variation of the correction parameter CT as a function of temperature is
shown in Figure 4. Detailed explanation for computing the correction parameters is
given in Ref. [17].

Thermoelastic Deformation of the Roll


The governing differential equation for thermoelastic deformation of the roll
is expressed as follows:


ijj + fi = 0 (12)

Table 3 The values of correction parameters (CQ , CP , C , Cd ) and effective convection heat transfer
coefficient hw for each water cooling zone

Water cooling zone CQ CP C Cd hw (W/mm2 ·  C)

C1 0.9984 0.8849 0.8126 0.9332 0.0689


C2 0.9872 0.8849 0.7144 0.9498 0.0616
C3 0.9923 0.8849 0.7959 0.9040 0.0654
C4 0.9417 0.8849 0.7420 0.8976 0.0605
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 989

Figure 4 Correction parameter CT dependent on roll surface temperature.

fi is the body force of the roll, which is assumed to be negligible. The stresses are
given in the following form if the roll material is assumed to be isotropic:


ij = 2 ij + ij kk − 3 + 2ij T (13)

where
E E 1
=  = and ij = uij + uji  (14)
1 + 1 − 2 21 +  2

E and  represent Young’s modulus and Poisson’s ratio, respectively.  is the


thermal expansion coefficient, and T is the difference between the initial and
current temperatures.

FINITE ELEMENT ANALYSIS


Because the stress evolution in the roll is strongly dependent on the
temperature variation in the roll and vice versa, the temperature evolution and
displacement field should be computed at the same time. The finite element
approximation of Eq. (1) and Eq. (12) can be written in the following form:
    e   e  e 
0 0 U̇ e K  0 U F
+ = (15)
0 C e  Ṫ e e
0 Kcon  Te Qe

K e  and U e  indicate the stiffness matrix and displacement of the element. F e 


is the vector of the load generated by the surface traction. C e  and Kcon
e
 are the
heat capacitance matrix and heat conductance matrix, respectively, of the element.
In Eq. (14), conductivity, thermal expansion coefficient, and specific heat of the roll
are dependent on temperature.
990 D.-H. NA ET AL.

Their relations to the temperature are given in Figure 3. Ṫ e  and T e  are the
vector of temperature change with time and temperature vector, respectively. Qe 
is an array of thermal loads and is given by the sum of the contributions due to the
surface heat flux [18].
This fully coupled temperature–displacement system can be solved by using
the Newton–Raphson method. The commercial FE program, Abaqus Ver. 6.11 [19]
offers an exact implementation of Newton’s method. The implicit time integration
method was employed, and the element type used for the roll was CPE4T
(four-node plane strain thermally coupled quadrilateral, bilinear displacement, and
temperature).

Virtual Rolling Model to Reduce the Run Time


The contact algorithm, which analyzes various characteristics at the interface
of the roll and the strip, requires a considerable computational cost. In addition,
a huge number of elements has to be used to construct the very long strip
(approximately 1000 m) passing through Std. No. 7; thus, the run time (i.e.,
computational cost) is unavoidably greatly increased.
To overcome the problem of the huge computational cost, we used a virtual
rolling approach that utilizes the similarities between hot strip rolling and a disc-pad
in the brake system of an automobile. The disc gains heat from the frictional heat
produced at the interface of disc and pad; the disc cools down by air after being
separated from the pad. Similarly, the roll gains heat generated by contacting the
hot strip during rolling and loses heat by air and water cooling.
Therefore, the roll corresponds to the disc, and the strip in contact with the
roll is equivalent to the pad in contact with the disc. The validity of this approach
was proven by Na and Lee [20]. If this approach is employed in FE analysis, the
entire length of strip, which reaches about 1000 m at Std. No. 7, does not need to be
considered. The strip length used in the virtual rolling model is only 0.0245 m since
the heat conduction from the strip to the roll occurs along the contact length of the
roll and strip. Note that the strip length 0.0245 m is not selected arbitrary.
The strip length 0.0245 m is equal to the length of arc of contact at stand no.
7 and is calculated based on the given values of the roll diameter, the entering strip
thickness and reduction ratio at stand no. 7. When the virtual rolling model was
adopted in FE analysis, the run time was reduced from about 3000 h to 11 h. As a
result, the computational cost was reduced by 96.9%.

Mesh Test
A preliminary FE analysis was performed to determine the element size at
the outer layer of the roll at Std. No. 7 of an actual hot strip mill. First, we
utilized Stevens et al.’s roll surface temperature measurements obtained from the
pilot hot rolling test. Figure 5 compares the predicted roll surface temperatures with
those measured by Stevens et al. [13] when the element height (element size toward
the radial direction of the roll) is fixed to 1.0 mm and the element width (element
size toward the circumferential direction of the roll) varies. For an element width
of 1.7 mm, there are some differences between the measurements and predictions.
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 991

Figure 5 Comparison of measured and predicted roll surface temperatures for different element sizes
when the roll rotates once. Roll surface temperatures were obtained from a pilot hot rolling test [13].
The diameter of the roll used in the pilot hot-rolling mill was 193 mm.

However, the differences were reduced at element widths of 3.4 and 6.8 mm. Note
that the ratio of the perimeter of the roll (=193 mm) used in Stevens et al.’s [13]
rolling test over the element width w (=3 4 mm) was 56 8 . This ratio is used to
determine the element width of the outer layer of the roll at Std. No. 7.
Figure 6 shows the variation in maximum temperature on the roll surface
when the element width W varies for three different element heights H1 , H2 , and H3

Figure 6 Variations in maximum temperature on the roll surface at Std. No. 7 of an actual hot strip
mill when element width W varies for three different element heights: H1 , H2 , and H3 .
992 D.-H. NA ET AL.

Table 4 Temperatures and state of temperature in terms of time increment at three points in the
outer roll layer (R−r∗ ); r is greater than r∗ (see Fig. 2(b))

Temperature ( C) at

Time increment
(second) State of temperature r/R = 1 000 r/R = 0 9986 r/R = 0 9957 Run time (hour)

0.0005 Converge 128.7 298.1 81.9 56


0.0010 Converge 128.7 298.1 81.9 27
0.0020 Converge 128.7 298.1 81.9 11
0.0030 Diverge 201.4 424.3 138.5 8
0.0040 Diverge 287.5 576.8 192.7 6
0.0050 Diverge 395.8 784.2 301.6 4

at Std. No. 7. The variation patterns of the maximum roll surface temperature as a
function of the aspect ratio are similar for H1 , H2 , and H3 , but their magnitudes are
different. In the case of H1 , the maximum temperature on the roll surface converges
to 300 C. Serajzadeh and Mucciardi [21] and Belzunce et al. [22] reported that the
maximum roll surface temperature in the actual hot strip rolling process reached
around 300 C.
Thus, H1 (=0 5 mm) may be a suitable element height H. Finally, element
width W may be determined by using the ratio of 56 8 mentioned above.
W (= 12 32 mm) is then obtained by dividing the perimeter of the roll at Std. No. 7
(700 mm) by the ratio of 56 8 . Thus, the element size used in the outer layer of the
roll is 12.32 mm (W) × 0.5 mm (H). The outer layer toward the roll center has four
elements; therefore, the thickness of the outer layer is 2.0 mm. Coarser elements are
used in the remaining roll region.
Time step, i.e., time increment is dependent on the mesh size in the outer layer
of the roll. Table 4 shows that the state of temperatures in the outer roll layer of

Table 5 List of parameters used in finite element analysis

Parameters Explanation

w1 , w2 , and w3 Element widths along the circumferential direction of the outer layer of
the roll in the pilot hot plate mill

h Element height toward the radial direction of the roll in the pilot hot
plate mill

W Element width along circumferential direction of the outer layer of the


roll of Std. No. 7 in the actual hot plate mill

H1 , H2 , and H3 Element heights toward the radial direction of the roll of Std. No. 7 in
the actual hot plate mill

R Roll radius of Std. No. 7 in the actual hot plate mill

r∗ Distance from the roll center to a point, which is close to the roll surface

R−r ∗ Outer layer thickness of the roll, which is close to the roll surface

r A distance from the roll center to a point where the temperature is


computed
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 993

thickness diverges if time increment is greater than 0.002 s. If time increment is less
than 0.002 s, the state of temperatures at the points converges. Hence, in the light of
the run time, time increment 0.002 s was selected in the FE analysis. Table 5 shows
the list of parameters used in finite element analysis.

RESULTS AND DISCUSSION


Temperature Evolution and Resulting Thermal Stress Variations
During Rolling
Figure 7(a) shows the temperature evolution at three points of the roll along
the radial direction during rolling when the initial temperature of the roll is 30 C.
The number “0” (zero) of the axis of abscissa denotes the onset of rolling at Std.
No. 7: i.e., the time that the strip enters the roll gap of Std. No. 7. The roll rotates
about 5.8 revolutions per second (348 rpm). R denotes the roll radius, and r is the
distance from the roll center to a point where the temperature is computed. As the
ratio (r/R) decreases, the temperature amplitude, i.e., magnitude of the temperature
fluctuation per revolution diminishes radically.

Figure 7 (a) Temperature evolution at different depths in the roll at Std. No. 7 during rolling. The
initial temperature of the roll is 30 C. (b) Circumferential stresses at different radial depths at Std.
No. 7 during rolling.
994 D.-H. NA ET AL.

The temperature amplitude at r/R = 1 0 (roll surface) and r/R = 0 9986


(0.49 mm beneath the roll surface) are around 260 and 60 C, respectively. When
r/R = 0 9957 (1.5 mm beneath the roll surface), the temperature amplitude
decreased by as much as 10 C. This is because heat conduction due to heating and
cooling on the roll surface (r/R = 1 0) is much more significant than that inside the
roll. In particular, the higher temperature amplitude at the roll surface is attributable
to that the roll rotates very quickly.
Even though the rolling time at Std. No. 7 is 70 s, computation results for only
10s are shown in Figure 7(a) because the temperature at the three points already
reached a steady state after 8 s. Therefore, the surface temperature evolution in the
roll is inferred to stabilize at around 300 C for the remaining 60s. This steady-
state roll surface temperature of 300 C during rolling is similar to the roll surface
temperature reported by Serajzadeh and Mucciardi [21] and Belzunce et al. [22]
Figure 7(b) illustrates the resulting circumferential stresses at three points of
the roll during rolling. Hereafter, the circumferential tensile stress and compressive
stress are referred to as the tensile stress and compressive stress, respectively, for
convenience. Note that no tensile stress was generated in the roll during rolling.
This result contradicts previous works [1, 2] that reported that tensile stress was
generated in the roll during rolling. This disagreement can be attributed to the initial
temperature setting of the roll. The initial roll temperature in this study was set to
30 C.
On the other hand, the previous works [1, 2] set the initial temperature
of the roll to be 100 or 200 C. These abnormally high initial temperatures are
attributable to that they [1, 2] did not consider idling. The initial roll temperature
in an actual hot strip rolling process is much lower than 100 and 200 C when idling
is considered. We will explain how idling influences the initial roll temperature. The
rationale for setting the initial roll temperature to 30 C is also explained. We will
also discuss the effects of the initial roll temperature on the change in stress state of
the roll during rolling.
The Peclet number (Pr ) of roll can be measures of relative importance of
convective and conductive heat transfer in the roll [23]. The Peclet number (Pr ) also
indicates the magnitude of thermal boundary layer in the roll, in which considerable
temperature change occurs. The Peclet number for roll is expressed as follows:
 2
1 x1
Pr = = (16)
r /Ur x1 yr

r and Ur are the thermal diffusivity and rolling speed. x1 and yr are the contact
length of the roll-strip interface and the thickness of thermal boundary layer, i.e.,
the outer layer thickness of the roll (= R−r ∗ ). The thermal diffusivity of roll of the
last stand r is 4 7E10−6 m2 /sec.
In the last stand, the contact length between the roll and the strip is 24.5 mm
and rolling speed is 11.7 m/s. Therefore, the computed roll Peclet number is 24765
and yr is 0.1557 mm. If rolling speed is reduced by 20%, Pr = 19812 and yr =
0 1741 mm. The roll Peclet number is 29718 and yr = 0 1421 mm if rolling speed is
increased by 20%. It denotes that the effect of roll rotation rate is very small.
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 995

Because the Peclet numbers are very large, conduction parallel to the interface
is negligible. It indicates two modes of heat transfer (convection parallel to the roll-
strip interface and conduction normal to this direction) prevail during rolling. The
thickness of thermal boundary layer (0.1557 mm) predicted by the Peclet number is
very thin compared with the thermal boundary layer (R−r ∗ = 14 1 mm) predicted
by finite element analysis.
About two orders of magnitude gap can be attributed to the Peclet number
only being applicable to a cold rolling process where the inlet strip temperature and
roll temperature are equal. Thus, the thickness of thermal boundary layer (R−r ∗ =
14 1 mm) predicted by FEA might be reasonable because the initial roll temperature
(30 C) is greatly lower than the strip temperature (about 860 C).

Temperature Evolution During Idling


Figure 8(a) illustrates the temperature evolution at three points in the roll
toward the roll center during idling. Note that idling begins right after rolling is
completed. The number “0”(zero) of the axis of abscissa denotes the onset of idling
at Std. No. 7, i.e., the time that the tail of a strip exits the roll gap of Std. No. 7.
Note that four nozzles spray the roll surface with water during idling just like during
rolling (see Figure 1). When r/R was 1.0 the steady periodic state was attained at
about 6 s after rolling begins. When r/R was 0.9957, the steady state was attained
at about 9 s after rolling starts. Meanwhile, the temperature at r/R = 1 0 and r/R =
0 9957 reached a steady state approximately 42 s and 81 s after the onset of idling.
This result indicates that the idling time (120 s) given at Std. No. 7 is more than
enough. Therefore, we can considerably reduce the idling time that has been adopted
in hot strip mills to reduce the roll temperature.
To examine the temperature evolution at early stages of idling in detail, the
interval from the beginning of idling to 1.0 s is magnified (Figure 8(b)). The roll
surface (r/R = 1 0) temperature drops very quickly from about 300 C to around
80 C within 0.1 s. Afterward, the slope of the temperature drop is very low because
the temperature is already decreased to near the initial roll temperature (30 C) and
ultimately the water temperature (16 C). The temperature fluctuation at the roll
surface is larger than at the other two points (r/R = 0 9986 and r/R = 0 9957)
because the roll surface cools down rapidly by natural convection with air and
forced convection with water, but the two points cool down by conduction only.
Figure 8(c) shows the temperature evolution in the roll at deeper points during
idling. The temperature evolution r/R = 0 9663 (11.8 mm below the roll surface)
remains almost constant after about 80s. The temperature evolution at r/R = 0 9957
(1.5 mm below the roll surface) reaches steady state much earlier than that at
r/R = 0 9663 since the thermal state at 1.5 mm from the roll surface is much more
influenced by forced convection heat transfer than the thermal state at 11.8 mm.
The temperature evolution at r/R = 0 8571 (50 mm below the roll surface)
remains unchanged following the onset of idling. This indicates that no heat
conduction occurred from 50 mm below the roll surface to the center of the roll.
This means that only the outer layer of 50 mm thickness experiences heat conduction
during idling. For three-quarters of the idling time, the temperature distribution
along the radial direction of the roll is around 17–30 C. Thus, there is good
996 D.-H. NA ET AL.

Figure 8 (a) Temperature evolution at different depths in the roll. (b) Temperature evolution in the
roll from the 0 s to 1.0 s of idling is enlarged. (c) Temperature evolution in the roll at deeper points
during idling. No heat conduction occurs in the roll when the ratio r/R reaches 0.8571.

justification for setting the initial roll temperature to 30 C (see Figure 7(a)), and this
temperature (30 C) might actually be a conservative estimate.

Effect of Initial Roll Temperature on the Changes in Temperature and


Stress State of the Roll Surface During Rolling
We examine how the temperature and stress state of the roll surface
during rolling is influenced by the initial roll temperature. Figure 9(a) shows the
temperature variation at a point on the roll surface for a single rotation of the roll
when three different initial roll temperatures were assumed. A and B indicate the
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 997

Figure 9 (a) Variations in the roll surface temperature for three different initial roll temperatures when
the roll rotates once. A and B indicate the starting and end points of the arc of contact. C, D and D,
F denote the starting and end points of the cooling regions. (b) Water-cooling nozzles installed around
the upper (work) roll and strip being rolled. C1–C4 indicate the regions where the water-impinging jet
sprayed from each nozzle came into contact with the roll surface. (c) Change in the stress state for
one revolution of the roll.

starting and end points of the arc of contact of the roll and strip. C, D and E, F
denote the starting and end points of water cooling zones. When sections B–C, D–E,
and F–A are subjected to air cooling, the temperature falls slowly by radiation.
These alphabet characters are fixed in space and not to the roll; they
correspond to those in Figure 9(b). For multiple revolutions of the roll, the
temperature profile of the roll surface underwent the temperature cycle mentioned
here. The roll surface temperature dropped below the initial roll temperature
(100 C) after about 85 C (indicated by the number 1 in Figure 9(a)). Similarly,
the roll surface temperature fell below the initial roll temperature after about 75 C
(indicated by the number 2 in Fig. 9(a)) when the initial roll temperature was set
to 200 C. However, the roll surface temperature was higher than the initial roll
temperature for the whole rotation when the initial roll temperature was set to
998 D.-H. NA ET AL.

30 C. This is because the difference between the water temperature (16 C) and initial
roll temperature (30 C) was small, and water was not sprayed over the whole roll
surface.
Figure 9(c) shows the stress variation on the roll surface for a single rotation of
the roll. In the beginning, compressive stress prevailed on the roll surface, but tensile
stress started to occur when the initial roll temperature was relatively high (100 and
200 C). The numbers 1 and 2 in Figure 9(c) correspond with those in Figure 9(a).
This change in the stress state can be explained by the term T in Eq. (8). Note that
T = initial roll temperature - current surface temperature. T was negative until
the point rotated from zero to the angles indicated by the numbers 1 and 2. Hence,
a compressive stress was produced.
However, T became positive when the point passed through the angles
indicated by the numbers 1 and 2. Tensile stress was then generated. Meanwhile,
when the initial roll temperature was 30 C, T was negative throughout one
revolution. Therefore, compressive stress was generated the entire time. In other
words, tensile stress was not generated on the roll surface during rolling when the
initial roll temperature was less than or equal to 30 C.

Variations in Thermal Stresses During Idling


Figure 10(a) shows the changes in the stresses at the roll surface (r/R = 1 0)
and the other two points in the roll during idling. The ratios r/R = 0 9986 and
r/R = 0 9957 correspond to points 0.49 and 1.5 mm below the roll surface. The
stresses at r/R = 1 0 and r/R = 0 9986 fluctuate sharply, but stresses at r/R =
0 9957 increases smoothly.
This is because the temperatures and resulting thermal stresses at r/R = 1 0
and r/R = 0 9986 are influenced directly by natural convection in air and forced
convection in water during idling. The compressive state of stress at the early stage
of idling alters to the tensile state of stress in a very short time of less than a second.
Overall, the tensile stress prevails at r/R = 1 0 0 9986, and 0.9957 after about 5 s.
This means that the outer layer thickness of 1.5 mm is in the state of tensile stress
for most of the idling.
To examine the transition from compressive stress to tensile stress, the interval
from the beginning of idling to 3.0 s was magnified, as shown in Figure 10(b). At
approximately 0.76 s, which corresponds to the time that the roll surface during
idling first reached the initial toll temperature (30 C), tensile stress started occurring
at the roll surface (r/R = 1 0). At a depth of 1.5 mm (r/R = 0 9957) from the roll
surface, tensile stress began to be generated at about 3.0 s.
This is because the temperature at r/R = 0 9957 decreased below the initial
roll temperature at 3.0 following the onset of idling. The roll’s outer layer of 1.5-
mm thickness experienced tensile stress for a long time, i.e., about 117–119 s during
idling. Therefore, preexisting micro cracks on the roll’s outer layer may begin to
grow during idling, not during rolling. Consequently, we can reduce the idling time
from 120 s to 28 s in order to prevent the growth of micro cracks on the roll surface
layer. This reduction would decrease the operation time of the hot strip rolling
process and decrease water usage by 76.7%.
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 999

Figure 10 (a) Stress variations at different depths of the roll toward the roll center during idling. (b)
Stress variations in the roll from 0 s to 3.0 s of idling are enlarged for convenience.

CONCLUSIONS
We have analyzed the temperature evolutions and resulting stress changes of
the work roll during both rolling and idling at the last stand (Std. No. 7) of a
finishing mill in the actual hot strip rolling process to examine the critical idling
time for the work roll between strips. To reduce the run time, we have used a virtual
rolling approach that utilizes the similarities between hot strip rolling and a disc-pad
brake of an automobile. The finite element model used in this study was validated
by comparing the computed roll surface temperature with that measured by Stevens
et al. [13]. Our conclusions are as follows:
Circumferential tensile stress is not generated in the roll during rolling if the
initial roll temperature is equal to 30 C. However, circumferential tensile stress
(about 68 MPa) was created on the roll surface very shortly after idling began
(about 0.76 s). The temperature at the roll surface increased to 300 C during rolling
decreases of 30 C about 28 s after the onset of idling. Therefore, the idling time
of 120 s that has been adopted at the last stand to reduce the roll temperature
to the initial roll temperature (30 C) prior to rolling the next strip is too much.
We recommend that the idling time be decreased from 120 s to 28 s: i.e., by 76.7%.
1000 D.-H. NA ET AL.

This decrease will significantly reduce the chances of growth of preexisting micro
cracks on the roll surface and consequently increase the roll life and decrease water
usage in the actual hot strip rolling process.

REFERENCES
1. C. G. Sun, C. S. Yun, J. S. Chung, and S. M. Hwang, Investigation of Thermo-
Mechanical Behavior of a Work Roll and of Roll Life in Hot Strip Rolling, Metall.
Mater. Trans. A, vol. 29, pp. 2407–2424, 1998.
2. D. Benasciutti, E. Brusa, and G. Bazzaro, Finite Elements Prediction of Thermal
Stresses in Work Roll of Hot Rolling Mills, Procedia Eng., vol. 2, pp. 707–716, 2010.
3. M. A. Cavaliere, M. B. Goldschmit, and E. N. Dvorkin, Finite Element Simulation
of the Steel Plates Hot Rolling Process, Int. J Numerical Methods in Eng., vol. 52,
pp. 1411–1430, 2001.
4. W. B. Lai, T. C. Chen, and C. I. Weng, Transient Thermal Stresses of Work Roll by
Coupled Thermoelasticity, J. Comp. Mechanics, vol. 9, pp. 55–71, 1991.
5. R. D. Mercado-Solis, J. Talamantes-Silva, J. H. Beynon, and M. A. L. Hernandez-
Rodriguez, Modelling Surface Thermal Damage to Hot Mill Rolls, Wear, vol. 263,
pp. 1560–1567, 2007.
6. A. Saboonchi and M. Abbaspour, Changing the Geometry of Water Spray on Milling
Work Roll and Its Effect on Work Roll Temperature, J. Mater. Proc. Tech., vol. 148,
pp. 35–49, 2004.
7. S. Serajzadeh, Effects of Rolling Parameters on Work-Roll Temperature Distribution
in the Hot Rolling of Steels, Int. J Adv. Manuf. Tech., vol. 35, pp. 859–866, 2008.
8. A. C. Yiannopoulos, N. K. Anifantis, and A. D. Dimarogonas, Thermal Stress
Optimization in Metal Rolling, J. Therm. Stresses, vol. 20, pp. 569–590, 1997.
9. D. Chang, Thermal Stresses in Work Rolls During the Rolling of Metal Strip, J. Mater.
Proc. Tech., vol. 94, pp. 45–51, 1999.
10. P. T. Hsu, Y. T. Yang, and C. K. Chen, A Three-Dimensional Inverse Problem of
Estimating the Surface Thermal Behavior of the Working Roll in Rolling Process, J.
Manuf. Sci. Eng., vol. 122, pp. 76–82, 2000.
11. R. L. Corral, R. Colas, and A. Perez, Modeling the Thermal and Thermoelastic
Responses of Work Rolls Used for Hot Rolling Steel Strip, J. Mater. Proc. Tech., vol.
153–154, pp. 886–893, 2004.
12. J. K. Saha, S. Kundu, S. Chandra, S. K. Sinha, U. Singhal, and A. K. Das,
Mathematical Modelling of Roll Cooling and Roll Surface Stress, ISIJ Int., vol. 45, pp.
1641–1650, 2005.
13. P. G. Stevens, K. P. Ivens, and P. Harper, Increasing Work-Roll Life by Improved Roll
Cooling Practice, J. Iron Steel Instit., vol. 1, pp. 1–11, 1971.
14. C. C. Chen and S. Kobayashi, Rigid-plastic Finite Element Analysis of Ring
Compression, Application of Numerical Methods to Forming Processes, ASME, Applied
Mechanics Division, vol. 28, pp. 163–174, 1978.
15. C. O. Hlady, J. K. Brimacombe, I. V. Samarasekera, and E. B. Hawbolt, Heat Transfer
in the Hot Rolling of Metals, Metall. Mater. Trans. B, vol. 26, pp. 1019–1027, 1995.
16. D. E. Lee, Evaluation of Optimal Residence Time in a Hot Rolled Reheating Furnace,
World Acad. Sci. Eng. Tech., vol. 59, pp. 1180–1184, 2011.
17. V. B. Ginzburg, F. A. Bakhtar, and R. J. Issa, Application of Coolflex Model for
Analysis of Work Roll Thermal Conditions in Hot Strip Mills, Iron Steel Eng., vol. 11,
pp. 38–45, 1997.
18. L. M. Galantucci and L. Tricarico, Thermo-mechanical Simulation of a Rolling Process
with an FEM Approach, J. Mater. Proc. Technol., vol. 92–93, pp. 494–501, 1999.
THERMAL STRESS EVOLUTION DURING ROLLING AND IDLING 1001

19. ABAQUS Version 6.11 Documentation, Dassault Systemes Simulia Corp., Providence,
RI, 2011.
20. D. H. Na and Y. Lee, An Approach Which Reduces Drastically the Run Time in
Computing the Surface Temperature on the Work Roll in Hot Strip Rolling, Proc.
14th International Conference on Metal Forming, Krakow, Poland, pp. 47–50, September
16–19, 2012.
21. S. Serajzadeh and F. Mucciardi, Modeling the Work-Roll Temperature Variation at
Unsteady State Condition, Model. Simul. Mater. Sci. Eng., vol. 11, pp. 179–194, 2003.
22. F. J. Belzunce, A. Ziadi, and C. Rodriguez, Structural Integrity of Hot Strip Mill
Rolling Rolls, Eng. Failure Analysis, vol. 11, pp. 789–797, 2004.
23. W. R. D. Wilson, C. T. Chang, and C. Y. Sa, Interface Temperatures in Cold Rolling,
J. Mater. Shaping Technol., vol. 6, pp. 229–240, 1989.

View publication stats

You might also like