Research Article

Advances in Mechanical Engineering

2017, Vol. 9(10) 1–19
Ó The Author(s) 2017
Three-dimensional metal particle flow DOI: 10.1177/1687814017720879
journals.sagepub.com/home/ade
simulation during the whole rolling
process for 60 kg/m heavy rail

Cheng-gang Pan1, Chuan-xiang Zhao1,

Qing-ming Chang1, Jia-lin Zhou1 and Qiong Wu2

Abstract
Three-dimensional metal particle flow simulation during the whole rolling process for 60 kg/m heavy rail was accom-
plished by explicit dynamic finite element method and modified updating geometric method. A finite element model,
physical parameters of U71Mn, and parameter setting of simulation were introduced in detail. The metal flow rules of
typical metal particles in rolled stock cross-section during the whole heavy rail rolling were obtained. Meanwhile, the
continuous casting slab with artificial surface cracks, referred to as ‘‘circle’’ holes, was rolled in actual production line of
Wuhan Iron and Steel (Group) Company. The finite element method calculated results of cross-sectional metal particles
for U71Mn heavy rail are in good agreement with the experimental ones. It is shown that the explicit dynamic finite ele-
ment method and modified updating geometric method can be used effectively to analyze the metal particle flow during
the whole heavy rail rolling process.

Keywords
Particle flow, modified updating geometric method, finite element method, heavy rail, U71Mn

Date received: 19 December 2016; accepted: 20 June 2017

Handling Editor: Kai Bao

Introduction the process of producing heavy rail. Physical simula-

With the rapid development of high-speed railway
transportation, the world’s railway technology is devel-
oping towards automatic traction, high-speed and
heavy load.1 Meanwhile, surface defects originating
from steel making, continuous casting, and/or heavy
rail rolling are often present in heavy rail final prod-
ucts. The market demand for heavy rail without any
defects has increased.2–4 Therefore, understanding the
metal flow behavior in the process of heavy rail rolling
becomes very important. Heavy rail rolling has charac-
teristics of a lot of rolling passes and section grooves,
complex cross-sectional shape, large velocity, difference
of roll, and concurrence of direct pressure and lateral
pressure in groove. So, heavy rail rolling is extremely
non-uniform plastic deformation. Theoretical analysis
is very difficult to accurately find out the metal flow in
tion, such as trial rolling, is time-consuming, high cost,
and hard to do quantitative analysis and get mathemat-
ical rules.5 With the rapid development of computer
technology, numerical simulation method has become
an important tool for rolling theory research. Finite
element method (FEM) was widely applied in solving
hot rolling problems. For example, an explicit dynamic
1
The State Key Laboratory of Refractories and Metallurgy, Wuhan
University of Science and Technology, Wuhan, China
2
The Huaihe River Commission of the Ministry of Water Resources,
Bengbu, China
Corresponding author:
Cheng-gang Pan, The State Key Laboratory of Refractories and
Metallurgy, Wuhan University of Science and Technology, 430081 Wuhan,
China.
Email: pcg2000@163.com

Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 Advances in Mechanical Engineering
FEM and updating geometric method was used by Yu
et al.6 to investigate slab deformation behavior during
the multi-pass vertical-horizontal (V-H) rolling process.
M Glowacki7,8 and Guerrero et al.9 used generalized
plane-strain method and two-dimensional (2D) finite
element formulation to simulate the rolling process of
rail and shape steel in order to predict the metal flow
and heat transfer and analyze the roll-pass design. A
three-dimensional (3D) finite element model of U75V
heavy rail in UF finishing rolling of universal rail roll-
ing line was established by Tan and Jia10 to study the
characteristics of metal flow and deformation of the
U75V rolled stock in the X and Y direction. Pei et al.1
built a 3D thermo-mechanical coupled model of the
whole rolling process for 60 kg/m heavy rail by
ANSYS/LS-DYNA software and analyzed workpiece
deformation, metal flow, stress and strain, temperature,
and rolling force.
All these studies and the literatures above mentioned
provide valuable information; however, they mainly
focused on stress, strain, heat transfer, and geometrical
shape of rolled stock and so on during the rolling.
There is rare research on metal flow or movement
behavior of rolled stock during the whole hot rolling
process for 60 kg/m heavy rail, especially for tracing
the metal particle of rolled stock during the whole
heavy rail rolling process.
In this article, 3D metal particle flow behavior of the
whole rolling process for 60 kg/m heavy rail was simu-
lated by explicit dynamic finite element software LS-
DYNA and modified updating geometric method.
Moreover, the metal particle flow regularity of cross-
section of 60 kg/m U71Mn heavy rail during the whole
rolling process was analyzed. At the same time, the
Figure 1. Cross-sectional schematic diagram of heavy rail.
continuous casting slab with artificial surface defects,
referred to as ‘‘circle’’ holes, was rolled in actual pro-
duction line of Wuhan Iron and Steel (Group)
Company. The FEM calculated results of cross-section
were compared with the experimental findings.
Heavy rail rolling process
Figure 1 shows the cross-sectional schematic diagram of
heavy rail. The up and down is the height direction of
heavy rail, while the left and right is the width direction
of heavy rail. The direction perpendicular to cross-section
of heavy rail is the length direction. The corner radii R1
and R3 are equal to the R2 and R4 in value, respectively.
Continuous casting slabs with the dimension of 7950 mm
3 380 mm 3 280 mm are adopted as billets (U71Mn)
whose main chemical composition is C 0.71, Si 0.25, Mn
1.30, S 0.013, and P 0.019 in wt%.11–15 In the heavy rail
rolling process, continuous casting slabs are charged into
the furnace for reheating and then discharged at a tem-
perature of 1240°C. In order to clear off the scale layer
formed during the continuous casting and reheating peri-
ods, the billet at first passes through the scale cleaning
unit that applies water with 220 bar pressure, then goes
through the BD1 unit where the symmetric rolling is car-
ried out, and box grooves are used, next through the
BD2 unit where the asymmetric rolling is conducted, and
grooves (calibers) of varying geometrical shapes are
used,16 and finally through the universal rolling unit
which comprises of three stands, that is, universal rough-
ing rolling mill (UR), edger roughing mill (ER)/edger fin-
ishing rolling mill (EF), and universal finishing rolling
mill (UF).
Pan et al. 3

Figure 2. Schematic drawing of heavy rail rolling process.

Figure 3. Roll distribution drawing of BD2 unit for 60 kg/m heavy rail.

Figure 4. Schematic views of heavy rail rolling in universal rolling unit: (a) UR1, UR2, and UR3; (b) ER; (c) EF; and (d) UF.

Figure 2 shows the schematic drawing of heavy rail
rolling process. There are five passes in BD1, five passes
in BD2, and six passes in universal unit. The grooves of
BD1 unit are symmetric box grooves, while groove of
each pass in BD2 and universal unit is the section
groove. Figure 3 shows the roll distribution drawing of
BD2 unit for 60 kg/m heavy rail. The schematic views
of heavy rail rolling in universal rolling unit are given
in Figure 4. Table 1 lists the rolling schedule table of
60 kg/m heavy rail.
Experimental materials and procedures
In order to trail the metal particle of rolled stock, artifi-
cial round holes with a dimension of j12 mm 3 20 mm
were drilled in the trial billet which was the dimension
4 Advances in Mechanical Engineering

Table 1. Rolling schedule table.

Unit Pass Groove Dimension, mm 3 mm Rolling reduction, mm Roll gap, mm Angular velocity of roll, rad/s

BD1 1 B2 290 3 325 55 193 3.71

2 B2 305 3 270 55 138 4.12
3 B2 280 3 252 53 120 4.33
4 B3 265 3 210 70 58 4.54
5 B4 230 3 220 35 78 4.79
BD2 6 E 210 55 13 5.96
7 D 200.5 45.5 13 6.36
8 C 73 95 13 6.47
9 B 53 20 13 6.83
10 A 43 10 13 7.29
Universal 11 UR1 32 11 32 7.59
12 ER 32 0 12 9.64
13 UR2 24 8 24 8.58
14 UR3 18.3 5.7 18.3 5.48
15 EF 18.3 0 12 6.90
16 UF 16.7 1.6 5 6.22

Figure 5. Topography of round hole drilled in the trial billet: (a) original photo and (b) magnification photo.

of 5000 mm (length) 3 280 mm (thickness) 3 380 mm round holes was oxidized in the process of reheating in
(width). The flow behavior of metal particle on the sur- furnace, so a large number of round holes were very
face of rolled stock can be represented by artificial difficult to heal by rolling reduction due to hard and
round hole because of the little dimension of round brittle oxide scale, and appeared in the form of black
hole and large rolling reduction during the heavy rail oxide scale.
rolling. Topography of round hole drilled in the trial
billet is shown in Figure 5. Figures 6 and 7 show the
wireframe and shade figures of the locations of round Basic theory
holes in the trial billet before numerical and physical
The basic equation of explicit dynamic analysis is
simulation, respectively.
In this study, the artificial round holes changed into  
long black lines on the surface of heavy rail finished M u +Cu +F =R ð1Þ
product after whole rolling. The locations of black lines
where M is the mass matrix, C is the damping matrix,
were measured by ruler and human eyes. Table 2 shows
F is the vector of inner force, R is the vector of outer
the photos of black lines founded in the heavy rail final  
loads, u is the nodes’ velocity, and u is the nodes’
products after 16 passes rolling. The locations of round 
acceleration. Based on the central difference method, u
holes before and after rolling are shown in Figure 8. 
and u can be expressed as
Due to large coefficient of elongation in the length or
rolling direction, artificial round holes were elongated  1
in the length direction after rolling and appeared in the u= (un + 1  un1 ) ð2Þ
2Dt
form of long lines. At the same time, the surface of
Pan et al. 5

Figure 6. Wireframe figure of the locations of round holes drilled in the trial billet before rolling.

Figure 7. Shade figure of the locations of round holes drilled in the trial billet before rolling.

 1 where Wmax is the max natural frequency of system; l is

u = (un + 1  2un + un1 ) ð3Þ
Dt2 the element length; c, v, E, and r are the air speed,
The discrete-time recurrence equation can be deter- Poisson ratio, Young’s modulus, and density of mate-
mined by substituting equations (2) and (3) into equa- rial, respectively.
tion (1)
  Finite element model
Dt 1
un + 1 = M + C Because the artificial round holes have very small size
2
    before rolling, in the meantime, it is the main objective
2 Dt
Dt (Rn  Fn ) + 2Mun  M  C un1 ð4Þ of just tracking the locations of artificial round holes in
2
the whole heavy rail rolling process. So, it is not neces-
In order to stably solve the central difference equa- sary to make artificial round holes on the finite element
tion of 3D problem, the time step Dt must be lower model. In this simulation, continuous casting slab with
than one critical value Dtmin which can be defined by a size of 5000 mm 3 380 mm 3 280 mm was adopted.
Courant–Friedrichs–Lewy stability criterion The flow stress s at high temperature is related to strain
_ and deformation temperature T.
value e, stain rate e,
2 l l During this simulation, the flow stresses were computed
Dtmin = = = qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð5Þ
Wmax c (1n)E as follows
(1 + n)(12n)r
6 Advances in Mechanical Engineering

Table 2. Photos of black lines in heavy rail.

Hole number Photo of black line Location/refer to Figure 1

Distant view Close shot

1 About in the middle of rail waist

2 On the left top surface of rail base near

the R3

3 At the location of 18 mm of the left top

surface of rail base from left side surface

4 On the left top surface of rail base near

the R1

5 About in the middle of the left side surface

of rail base

6 At the location of 8 mm of the bottom

surface of rail base from the left side surface

7 At the location of 23 mm of the bottom

surface of rail base from the left side surface

8 At the location of 41 mm of the bottom

surface of rail base from the left side surface

9 At the location of 59 mm of the bottom

surface of rail base from the left side surface

10 About in the middle of the bottom surface

of rail base

s = AeB e_C exp (D=T ) ð6Þ
where A, B, C, D are constant. A, B, C, and D are
5.103, 0.266, 0.116, and 4.464, respectively, according
to thermal simulation test results of U71Mn using
Gleeble-1500 thermal simulation experiment. The
bilinear isotropic hardening material model (Figure 9)
related to temperature was employed for the rolled
stock. The strain rate e_ is different during the whole
heavy rail rolling, so each rolling simulation time cell
adopted stress–strain curve with eight kinds of tempera-
tures at a certain strain rate according to actual produc-
tion line. The yield stress and tangent modulus under
different temperatures could be determined by 0.2%
residual plastic strain and tangent method, respectively.
The rolls were considered as rigid because rolling
Pan et al. 7

temperature was high and whose deformation could

be neglected. The physical parameters of rolls
adopted constants. Physical parameters of U71Mn
are plotted in Figure 10. Because the density,
Poisson’s ratio and thermal expansion coefficient of
rolled stock vary little with temperature, they were
set as constants which were 7850 kg/m3, 0.3 and
0.118E-04, respectively.
Because the rolling line of 60 kg/m heavy rail was
very long, and rolling passes were up to 16 passes, it
was difficult to simulate it at a time. As a result, the
simulation of the whole rolling process for 60 kg/m
heavy rail was divided into 12 time cells as shown
Figure 8. Locations of round holes before and after rolling.
in Table 3. Due to symmetry of rolled stock and rolls in
BD1 unit, only a quarter of rolled stock and rolls were
employed in the geometric model, and the continuous
rolling simulation model of BD1 unit was established at
a time, as shown in Figure 11. No. 1 and No. 2 rolling
passes were carried out in No. 1 groove of Figure 11,
and other grooves were used only once. Rolled stock
was tilted after No. 2, No. 3, and No. 4 rolling passes.
Rolls and rolled stock were meshed with eight-noded
hexahedral elements. After different mass scaling simu-
lations, 100 times mass scaling with minor error was
adopted. Reduction integration and hourglass control
in the simulation setup were employed to save compu-
tational time and avoid the zero-energy mode caused
by roll deformation, respectively. The hourglass energy
is less than 10% of total internal energy. Figures 12 and
13 show the FE model of rolls of BD2 and universal
unit, respectively. Figure 14 shows the FEM of rolled
stock and rolls in UR1 groove rolling pass of universal
unit. The Coulomb friction was adopted, and dynamic
friction coefficient between rolls and rolled stock was
set to 0.36. The effective coefficient of plastic convert-
Figure 9. Bilinear isotropic hardening material model.
ing into heat was set to 0.9. The initial temperature of

Figure 10. Physical parameters of U71Mn with temperature: (a) specific heat and (b) thermal conductivity.
8 Advances in Mechanical Engineering

continuous casting slab and rolls was set to 1180°C and

a constant value of 200°C.
As shown in Table 3, the numerical simulation of the
whole rolling process for 60 kg/m heavy rail was divided
into 12 time cells, which included rolling period and
cooling period. The simulation results of previous time
cell, including temperature and structure results, were
both as the initial condition of next time cell. Because
work hardening and recrystallization were existed in
hot rolling process at the same time, the effect of work
hardening could be basically eliminated. Therefore,
Figure 11. Geometric model of BD1 unit.
residual stress of previous time cell was not inherited by

Figure 12. FEM of rolls in BD2 unit: (a) Groove E, (b) Groove D, (c) Groove C, (d) Groove B, and (e) Groove A.

Figure 13. FEM of rolls in universal unit: (a) UR1, UR2, and UR3; (b) ER; (c) EF; and (d) UF.

Table 3. Simulation time cells of the whole rolling process for 60 kg/m heavy rail.

Time cell 1 2 3 4 5 6

Explanation BD1/cooling BD2-E/cooling BD2-D/cooling BD2-C/cooling BD2-B/cooling BD2-A/cooling

Time cell 7 8 9 10 11 12
Explanation Universal- Universal- Universal- Universal- Universal- Universal-UF
UR1/cooling ER/cooling UR2/cooling UR3/cooling EF/cooling
Pan et al.
Figure 14. FEM of rolled stock and rolls in UR1 groove rolling
pass of universal unit.
the next time cell in this article. A full size FE model of
rolled stock of No. 2 time cell could be obtained by
mapping the 1/4 FE model of rolled stock after No. 1
time cell along the plane of symmetry, as shown in
Figure 15.
Because initial mesh of rolled stock distorted seri-
ously after multi-pass rolling simulation, calculation
accuracy decreased directly and the FE model of rolled
stock needed to rebuild at appropriate time. Yu et al.5,6
presented an updating geometric method for simulating
V-H rolling process, but this method is only suitable
for less rolling passes or small mesh distortion of rolled
stock. It is necessary to mesh new gird for rolled stock
Figure 15. Finite element model of rolled stock after No. 1 time cell: (a) 1/4 model and (b) full size model.
9
when rolled stock go through a lot of rolling passes or
mesh distortion. The model rebuilding method applied
by Pei et al.1 is not suitable for trailing metal particle
during the whole rolling process for 60 kg/m heavy rail
because the node positions have changed after remesh-
ing. Figure 16 shows the flow chart of model rebuilding
in this simulation, that is, extracting an element length
body element from the finite element model of rolled
stock after previous pass rolling in the middle position
of the rolled stock length direction, then extracting
nodes and lines on the surface of elements which per-
pendicular to the rolling direction, next generating the
2D surface meshes by connecting the existing lines, and
finally, generating new 3D FEM of rolled stock by
extruding surface meshes generated above along the
rolling direction. After finishing model rebuilding, the
temperature needed to be mapped from the old model
to the new model according to the coordinates. After
that, establishing the FE model of rolls for next pass,
changing the material attribution of rolled stock,
changing boundary conditions and loads, and then
simulating the next pass rolling process, which make
the rolling process simulation continuous. The follow-
ing passes just as above.
Results and discussion
Figure 17 shows the groove filling simulation results of
each pass in BD1 unit. From Figure 17, due to large
section size of rolled stock, plastic deformation cannot
penetrate into the center of rolled stock and then rolled
stocks after No. 1–No. 5 rolling in BD1 unit appear
double drum shapes. The roll gaps of No. 3 and No. 5
passes have not been filled well.
Figure 18 shows the groove filling simulation results
of each pass in BD2 unit. As shown in Figure 18, due to
the smaller roll diameter on the left side of No. 6 pass
groove, and the smaller metal flow resistance on the left
10 Advances in Mechanical Engineering

Figure 16. Flow chart of model rebuilding.

Figure 17. Groove filling results of (a) No. 1 pass, (b) No. 2 pass, (c) No. 3 pass, (d) No. 4 pass, and (e) No. 5 pass in BD1 unit.

Figure 18. Groove filling results of (a) No. 6 pass, (b) No. 7 pass, (c) No. 8 pass, (d) No. 9 pass, and (e) No. 10 pass.

side of groove than that on the right side, the metal the right side of groove. The filling of No. 7 pass groove
from rolling reduction direction is easier to flow to the is good because of the symmetrical profile of groove.
left side of groove, which leads to slightly not full on The groove profile of No. 8 pass has a big change, that
Pan et al.
Figure 19. Groove filling results of (a) No. 11 pass, (b) No. 12 pass, (c) No. 13 pass, (d) No. 14 pass, (e) No. 15 pass, and (f) No.
16 pass.
Figure 20. Isometric view of cross-section in the middle
position of billet length direction.
is, there are the circular arcs with small curvature radii
along the groove, so that the parts of the small arcs are
not filled well. After No. 9 and No. 10 passes, the filling
conditions are gradually improved, but the roll gap at
11
the position of rail head after No. 10 pass rolling is still
underfilled.
Figure 19 shows the groove filling simulation results
of each pass in universal unit. From Figure 19, the main
function of No. 12 and No. 15 passes is to control the
width dimension of heavy rail, especially for the width
of rail base and rail head, so the dimensional accuracy
of the height of heavy rail is not important. The main
mask of No. 11, No. 13, No. 14, and No. 16 passes is to
adjust the width size of rail waist, the height size, and
shape of heavy rail.
In order to observe the metal particle flow of billet
cross-section, isometric view of cross-section in the
middle position of the billet length direction is shown
in Figure 20. Figures 21–23 show 3D metal particle
position results of each pass in BD1, BD2, and univer-
sal unit, respectively. From Figure 21, the positions of
metal particles in BD1 unit are basically in a plane, and
the cross-section of rolled stock is mainly rectangular.
Because plastic deformation cannot go deep into the
center of rolled stock, the flow velocities of metal parti-
cles on subsurface layer are faster than those in the cen-
ter and on the surface, which leads to double drum
shapes in the rolling direction.
From Figure 22, the metal particle positions of
rolled stock in BD2 unit are not hold in a plane and
appear 3D status. Because of different rolling reduc-
tions along the roll length direction, rail waist has the
biggest plastic strain, and rail base has the minimal
plastic strain. Spatial features of metal particles in
12 Advances in Mechanical Engineering

Figure 21. 3D metal particle positions of (a) No. 1 pass, (b) No. 2 pass, (c) No. 3 pass, (d) No. 4 pass, and (e) No. 5 pass in BD1
unit.

Figure 22. 3D metal particle positions of (a) No. 6 pass, (b) No. 7 pass, (c) No. 8 pass, (d) No. 9 pass, and (e) No. 10 pass.

Figure 23. 3D metal particle positions of (a) No. 11 pass, (b) No. 12 pass, (c) No. 13 pass, (d) No. 14 pass, (e) No. 15 pass, and (f)
No. 16 pass.
Pan et al.
Figure 24. Position map of typical points: (a) plane and (b) spatial.
Figure 25. Positions of typical points of rolled stock cross-section after (a) No. 1 pass, (b) No. 2 pass, (c) No. 3 pass, (d) No. 4
pass, and (e) No. 5 pass rolling in BD1 unit.
universal unit are more serious than those in BD2 unit,
as shown in Figure 23.
In order to track the flow behavior of metal parti-
cles of rolled stock cross-section in the whole rolling
process, some typical points of rolled stock cross-
section are selected as shown in Figure 24. Points a,
b, c, and d are located in the corner of cross-section
of continuous casting slab; points e, f, g, and h are the
midpoints of edges; and point i is the geometric center
point.
Figure 25 shows the positions of typical points of
rolled stock cross-section in BD1 unit. From Figure 25,
it is obvious that the positions of typical points of rolled
stock cross-section have not changed basically on the
relative position, just have changed on absolute posi-
tion because of symmetrical box grooves in BD1 unit.
Figure 26 shows the positions of typical points of
rolled stock cross-section in BD2 unit. Points a and b
are still in the corner of rolled stock because points a
and b are on the side surface of rail base where little
rolling reduction happens. Points c and d move to the
13
side surface of rail head by turn over spread because of
large amount of rolling reduction. Points e and g basi-
cally stay in the middle of edges of rolled stock cross-
section due to basic horizontally symmetrical groove in
BD2 unit. Points h and f move to the left of rolled stock
height midpoint, that is, rail base of heavy rail because
rolling reduction at points h and f is the largest, and the
flow resistance of rail base is smaller than that of rail
head. Point i is the same as the points h and f. At the
same time, the grooves of No. 6, No. 8, and No. 9 are
not completely horizontal symmetry, which lead to
closer distance from rail base for point h than that for
point f.
Figure 27 shows the positions of typical points of
rolled stock cross-section in universal unit. From
Figure 27, the horizontal distances between point h and
point f become more and more big. At the same time,
points c and d are getting closer and closer to border
between side surface and bottom surface of rail head.
Compared with BD2 unit, other points are almost not
change.
14 Advances in Mechanical Engineering

Figure 26. Positions of typical points of rolled stock cross-section after (a) No. 6 pass, (b) No. 7 pass, (c) No. 8 pass, (d) No. 9
pass, and (e) No. 10 pass rolling.

Figure 27. Positions of typical points of rolled stock cross-section after (a) No. 11 pass, (b) No. 12 pass, (c) No. 13 pass, (d) No.
14 pass, (e) No. 15 pass, and (f) No. 16 pass rolling.
Pan et al.
Figure 28. Positions of artificial round holes in FEM of continuous casting slab.
Verification of experimental points
Figure 28 shows the positions of artificial round roles
in FEM of continuous casting slab. The experimental
points 540, 1064, 1069, 1074, 1079, 1085, 1090, 1095,
1100, and 1105 shown in Figure 28 are representative
of artificial round holes 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10 in
experimental rolling, respectively.
Figures 29–31 show the positions of 10 experimen-
tal points of rolled stock cross-section in BD1, BD2,
and universal unit, respectively. From Figures 29–31,
we can track the positions of 10 experimental points
in the whole heavy rail rolling and find out their flow
rules.
Figure 32 shows the experimental and simulation
results of the positions of 10 experimental points in the
continuous casting slab and heavy rail final product.
From Figure 32(a), the rectangle with rounded corners
represents the cross-section of continuous casting slab.
Points 1–10 on the side of the rectangle and heavy rail
final product profile are on behalf of the positions of
artificial round roles in the continuous casting slab and
heavy rail final product. The simulation results are in
good agreement with the experimental results by the
naked eye.
Figure 33 shows the experimental and simulation
values of the positions of 10 experimental points in the
heavy rail final product. From Figure 33, the simula-
tion values are in good agreement with the experimen-
tal values. Because the metal at the rail base is free
15
deformation in the width direction of heavy rail during
the last pass rolling (No. 16 pass), the height error is
smaller than the width error.
Conclusion
1. A 3D finite element model of whole heavy rail
rolling for 60 kg/m heavy rail was built using
explicit dynamic software and modified updat-
ing geometric method.
2. The groove filling of each pass in BD1, BD2,
and universal unit was simulated. The groove
filling of BD1 unit is good, and the cross-section
of rolled stock is prone to double drum shape.
The metal easily flows to position of small roll
diameter in BD2 unit, and the position of large
roll diameter and small arc has a slightly worse
filling result. In universal rolling unit, grooves
UR and UF fill well, and grooves ER and EF
are underfilled in the height direction of heavy
rail, especially for rail head part because grooves
ER and EF mainly control the width of heavy
rail rather than height.
3. 3D metal particle positions of each pass in BD1,
BD2, and universal unit were studied. The posi-
tions of metal particles of BD1 unit are basically
in a plane, and the cross-section of rolled stock
is mainly rectangular. In BD2 and universal
16 Advances in Mechanical Engineering

Figure 29. Positions of 10 experimental points of rolled stock cross-section after (a) No. 1 pass, (b) No. 2 pass, (c) No. 3 pass, (d)
No. 4 pass, and (e) No. 5 pass rolling in BD1 unit.

Figure 30. Positions of 10 experimental points of rolled stock cross-section after (a) No. 6 pass, (b) No. 7 pass, (c) No. 8 pass, (d)
No. 9 pass, and (e) No. 10 pass rolling.
Pan et al. 17

Figure 31. Positions of 10 experimental points of rolled stock cross-section after (a) No. 11 pass, (b) No. 12 pass, (c) No. 13 pass,
(d) No. 14 pass, (e) No. 15 pass, and (f) No. 16 pass rolling.

Figure 32. Positions of experimental points in the heavy rail final product: (a) experimental results and (b) simulation results.

unit, the positions of metal particles appear
3D status. Spatial features of metal particles of
rail waist are the most serious, and the metal
particles of rail base have the minimum plastic
strain.
4. The flow rules of typical points of rolled stock
cross-section in BD1, BD2, and universal unit
were studied. The positions of typical points of
rolled stock cross-section in BD1 unit have not
changed basically on the relative position. In
BD2 unit, points c and d move to the side sur-
face of rail head by turn over spread, and points
h, f, and i move to rail base. In universal unit,
the horizontal distances between point h and
point f become more and more big, and points c
and d are getting closer and closer to border
18
Figure 33. Experimental and simulation values of the positions of 10 experimental points in the heavy rail final product.
between side surface and bottom surface of rail
head.
5. 10 artificial round holes were drilled in the con-
tinuous casting slab. The simulation results of
10 experimental points are in good agreement
with the experimental results of 10 artificial
round holes in actual production line of Wuhan
Iron and Steel (Group) Company, which means
that the whole heavy rail rolling process can be
analyzed by means of explicit dynamics FEM
and modified updating geometric method.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with
respect to the research, authorship, and/or publication of this
article.
Funding
The author(s) disclosed receipt of the following financial sup-
port for the research, authorship, and/or publication of this
article: This work was financially supported by the National
Natural Science Foundation of China (No. 51375353), Open
Fund of the State Key Laboratory of Refractories and
Metallurgy of WUST (No. 2014QN03), Open Fund of the
Key Laboratory for Metallurgical Equipment and Control of
Ministry of Education of WUST (No. 2013A17), and Wuhan
Iron and Steel (Group) Company.
Advances in Mechanical Engineering
References
1. Pei N, Zhu G, Li B, et al. 3D thermo-mechanical coupled
simulation of whole rolling process for 60 kg/m heavy
rail. J Iron Steel Res Int 2014; 21: 1104–1110.
2. Filipovic M, Eriksson C and Överstam H. Behaviour of
surface defects in wire rod rolling. Steel Res Int 2006; 77:
439–444.
3. Yu H, Tieu K, Lu C, et al. Occurrence of surface defects
on strips during hot rolling process by FEM. Int J Adv
Manuf Tech 2013; 67: 1161–1170.
4. Chen L and Bi KX. Study on simulation experiment with
universal pass rolling deformation for heavy rail. Adv
Mat Res 2012; 430: 525–529.
5. Yu H, Liu X, Zhao X, et al. Explicit dynamic FEM anal-
ysis of multipass vertical-horizontal rolling. J Iron Steel
Res Int 2006; 13: 26–30.
6. Yu H, Liu X, Zhao X, et al. FEM analysis for V–H roll-
ing process by updating geometric method. J Mater Pro-
cess Tech 2006; 180: 323–327.
7. Glowacki M. Simulation of rail rolling using the general-
ized plane-strain finite-element approach. J Mater Pro-
cess Tech 1996; 62: 229–234.
8. Glowacki M. The mathematical modelling of thermo-
mechanical processing of steel during multi-pass shape
rolling. J Mater Process Tech 2005; 168: 336–343.
9. Guerrero A, Belzunce J, Betegón C, et al. Hot rolling
process simulation: application to UIC-60 rail rolling.
Recent Patent Mech Eng 2010; 3: 65–71.
Pan et al.
10. Tan X and Jia S. The numerical simulation for plastic
deformation of uf finishing rolling of universal rail rolling
line. Metal Int 2013; 18: 16–20.
11. Ma N, Cai Z, Huang H, et al. Investigation of welding
residual stress in flash-butt joint of U71Mn rail steel by
numerical simulation and experiment. Mater Design
2015; 88: 1296–1309.
12. Kang G and Gao Q. Uniaxial and non-proportionally
multiaxial ratcheting of U71Mn rail steel: experiments
and simulations. Mech Mater 2002; 34: 809–820.
13. Kang G, Gao Q and Yang X. Experimental study on the
cyclic deformation and plastic flow of U71Mn rail steel.
Int J Mech Sci 2002; 44: 1647–1663.
19
14. Zhong W, Hu JJ, Li B, et al. A study of rolling contact
fatigue crack growth in U75V and U71Mn rails. Wear
2011; 271: 388–392.
15. Kang GZ, Gao Q, Cai LX, et al. Experimental study on
the non-proportional cyclic plasticity of U71Mn rail steel
at room temperature. J Mater Sci Technol 2002; 18:
13–16.
16. Altınkaya H, Orak IM and Esen I. Artificial neural net-
work application for modeling the rail rolling process.
Expert Syst Appl 2014; 41: 7135–7146.