You are on page 1of 27

Applied Mathematical Modelling 47 (2017) 501–527

Contents lists available at ScienceDirect

Applied Mathematical Modelling


journal homepage: www.elsevier.com/locate/apm

Analysis of eccentric unbonded bimetal rod in ECAP based on


different arrangements of soft and hard metals
Wang Hongyu∗, Sun Jie, Wan Zhenting, Wang Qinglong, Zhao Dewen,
Zhang Dianhua
State Key Laboratory of Rolling and Automation, Northeastern University, Shenyang, Liaoning 110819, PR China

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, an analytical model of ECAP process with an eccentric unbonded bimetal rod
Received 5 May 2016 is proposed by slab method. Based on the discussion, the effects of the eccentricity ratio
Revised 4 March 2017
and the diameter of inner rod as well as the geometrical parameters of the mold on the
Accepted 20 March 2017
bonding process are obtained. Also this is the first time that different material arrange-
Available online 23 March 2017
ments are both taken into the consideration in ECAP. Two kinds of eccentric bimetal rods
Keywords: are researched, when outer pipe is hard and inner core rod is soft, the eccentricity ratio
ECAP will be relieved. When the outer is soft and inner is hard, the eccentricity ratio will be
Eccentric bimetal rod enlarged. These changes can be more than 10% in some situations. In order to verify these
Material arrangement conclusions, some experiments with aluminum and copper are designed and conducted.
Bond experiment In these experiments, 3 different molds and 10 different eccentricity ratios as well as 5
Slab method different inner diameters are discussed with two kinds of eccentric bimetal rods in ECAP.
The measured results are in agreement with those calculated by analytical model. Most of
the errors between the measurements and calculations are less than 6%.
© 2017 Elsevier Inc. All rights reserved.

Introduction

In order to produce the ultra-fine grain materials, many approaches were used in industries and laboratories. In most
of them, the samples were been moving along a single direction during the process. When it was the common pressing,
the samples were pressed along vertical direction. Though the severe plastic deformations were imposed to the sample,
the grains were still coarse. When it was the common rolling process, the samples were rolled along horizontal direction.
Though several independent plates would be bonded, and even the atoms could be transmitted on interface, the sizes of the
grains were still need to be fined. The asymmetrical rolling would give a different movement direction, due to the different
speeds of the rolls. The samples were not rolled along the horizontal anymore; the shear stresses could refine the sizes
of the grains to some extent. Wang et al. researched the asymmetrical rolling in slab method and also put this method in
the study about asymmetrical snake rolling [1–2]. But during this process, the angle between the initial direction and the
final one is limited. So ECAP which is short for equal channel angular pressing is proposed to produce the refined grain
materials. In order to refine the size of the grains, Valiev et al. [3] found this angle between initial and final direction had to
be large enough. Then the ultra fine-grained structures can be realized by the large deformation. Horita et al. [4] also found


Corresponding author.
E-mail addresses: wanghongyusci@yeah.net (W. Hongyu), sunthird@163.com (S. Jie), 1399520984@qq.com (W. Zhenting), 634203511@qq.com
(W. Qinglong), zhaodw@ral.neu.edu.cn (Z. Dewen), zdhua@mail.neu.edu.cn (Z. Dianhua).

http://dx.doi.org/10.1016/j.apm.2017.03.041
0307-904X/© 2017 Elsevier Inc. All rights reserved.
502 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Nomenclature

s–h outer soft pipe and inner hard rod


h−s outer hard pipe and inner soft rod
p vertical stress
σ horizontal stress
τ = mk shear friction
m1 frictional factor between channel and pipe
m2 frictional factor between pipe and rod
β angle in Fig. 1ࢬAOC
α angle in Fig. 1ࢬAOD
R initial longitudinal diameter
r0 radius of sector BHGF O in Fig. 1
r1 ,r2 ,r3 thickness of each layer
r −r
e = 3R 1 eccentricity ratio of the bimetal rod
ϕ angle from the entrance of the channel
ϕ I(s − h) , ϕ II(s − h) , ϕ III(s − h) and ϕ IV(s − h) boundaries of each zone in s–h
ϕ I(h − s) , ϕ II(h − s) , ϕ III(h − s) , ϕ IV(h − s) , ϕ V(h − s) boundaries of each zone in h−s
k12 , k23 , k123 shear yield strength of bonded layers weighted average by their thicknesses
r1f r2f r3f final thicknesses at the exit of the channel.
r1ϕI r2ϕI r3ϕI thicknesses at ϕ = ϕ I
r1ϕII r2ϕII r3ϕII thicknesses at ϕ = ϕ II
e(s − h) . e(h − s) final eccentricity ratio at the exit of the channel.
ks shear yield strength of the soft metal
kh shear yield strength of the hard metal

that after ECAP, the samples in experiments showed the refinements in both microstructure and the subsequent mechanical
properties. Segal [5] discussed the engineering and commercialization of this advance technology, the optimal choice of
processing route and numbers of passes were shown in his work. These materials can be attractive for use in a range of
applications from biomedical to aerospace industries [6]. Kim et al. [7] in his paper also pointed out that, comparing with
other severe plastic deformation methods, the ECAP can be seen as the best way to produce fully dense samples containing
ultrafine grain size.
The composite materials are also widely used in many fields such as communication, automobile and aerospace. Com-
paring with the single metal, a composite material can not only combine the advantages of each material but also avoid
their own weaknesses. Al/Cu composite material can be used as wires which not only can have a good conductivity but also
can be cheap and light. For another example, Al/Steel material can be applied in some vehicles, it can be not only lighter

Fig. 1. Schematic diagrams of corner in mold and analytical element.


W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 503

than pure steel but also stronger than using pure Aluminum alone. As many earlier studies about ECAP process was focused
on the microstructure evolution characteristics, Eivani and Karimi [8] proposed a new method to produce bimetallic rods.
The advantages of this method were the bimetallic rods could have a larger bonding strength and also it was possible that
the process could be performed in more than one pass. Narooei and Karimi [9] proposed a model for ECAP process with
Bezier formulation. Based on the experiments and analysis, the accuracy of calculation by the new model was verified. Then
Narooei and Karimi [10] extended this single metal ECAP models to research the deformation behavior of bimetal circular
billet in some process. Based on the calculations by the model, the strain distributions and extrusion loads were obtained
in both inner rod and outer layer. They provided the nature of the deformations in the bimetal rods ECAP. Since the ECAP is
an asymmetrical process, the inner rod would have an eccentric distance especially when the unbonded rods have an initial
eccentric distance. Also when the eccentric directions are different, the finial eccentricity can be more complex.
For the purpose of describing the ECAP process, some mathematical models are proposed. They can be divided into two
kinds. One is numerical models based on the finite element method. Kim and Kim [11] used FVM (finite volume method)
to yield the distributions of stress and strain in the sample, he emphasized that the strain distributions in vertical and
transverse direction were different with each other. The largest inhomogeneous strain distribution developed along vertical
direction while a relatively uniform one appeared along the transverse direction. Wei et al. [12] researched the ECAP process
with 2D software and the friction on the interface of the die and workpiece could improve the homogeneity of the stress
and strain distribution. Also a 3-D model was used in Jiang et al. [13] paper. Not only the structure field but also temperature
field were considered and researched in that paper. The distributions of the temperature could be also different based on
the different extrusion conditions. The other kind is analytical methods. Segal [14] first proved that there was an extra-
large, strictly uniform and unidirectional deformation during ECAP process. In this paper, the ECAP was seen as a simple
shear process, and all the deformations appeared in one direction. Reihanian et al. [15] used linear and rotational velocity
fields to analyze the ECAP process. The plastic deformation zone could be optimized for any deforming material undergoing
plastic deformation in plane strain condition. Eivani and Karimi [16] proposed a model based on upper bound approach.
During the ECAP process, the extrusion pressure decreased with increasing both the die angle and the outer curved corner,
and increased with increasing the friction coefficient were completely proved. But since the upper bound approach was
based on the velocity fields, the effect of the punch speed was serious and the results might be inaccuracy in high speed.
In this paper based on the slab method a 2-D model is proposed to research the ECAP process. This model can be used
to calculate the finial thicknesses of each layer after the pressing process. The eccentricity ratios in bimetal rods are firstly
to be studied and also it is the first time, both two kinds of material arrangements are considered together. The differences
and relationship between different material arrangements bimetal rods can be obtained by the models. When it is sandwich
rolling, these different material arrangements can also affect the final thickness ratio [17].

Modeling

Assumption

In order to simplify the formulation, the following assumptions are listed:

1. The channel is assumed to be rigid and the rods are regarded as rigid-plastic materials with no strain-hardening effects.
2. All the deformation appears in x–z plane or the plane parallel to x–z plane. No spread in transverse direction during the
pressing process.
3. The average stresses are distributed uniformly in elements. The vertical stress p and horizontal stress σ are regarded as
principal stress [18].
4. The normal stresses on sides of each element are uniform distribution along the sides, while the frictions between the
channel and pipe, pipe and rod are both constant values. But the friction factors may be different because of the different
materials. The shear friction is assumed as τ = mk, m1 is the friction factor between channel and pipe while m2 is for
the surfaces between pipe and rod [19,20].

As it is shown in Fig. 1 the equal channel is shown by lines ACDEF GHB. In the zone ACHB the longitudinal diameter of
R+r
the channel is increased from R the initial longitudinal diameter to ( cos β0 − r0 ) .In zone CDGH the longitudinal diameter of
R+r
the channel remains as a constant value ( cos β0 − r0 ). Finally when it is in zone DEF G the longitudinal diameter is reduced to
R. r0 is the radius of the sector BHGF O. Angle ࢬAOC is equal to β and ࢬAOD is α . So α + β is equal to π2 . When β is equal
to 0, points A and C are overlap, while points D and E are also the same points. At that time, the channel will be AEF GHB,
the longitudinal diameter of the channel will be R. When β = π4 , points C and D can be merged together. ϕ is the angle
from the initial entrance of the channel. So the range of the ϕ can be 0 ≤ ϕ ≤ π2 . When an infinitesimal angle dϕ is taken
into account, the bimetal rod can be seen as a rectangle element. From side view of the 2 dimension figure, the bimetal rod
can be seen as three layers of the metals. From the inner corner to outer corner the three layers are defined as 1 and 2 as
well as 3. The width of each layers can be r1 and r2 as well as r3 (Fig. 2).
504 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. 2. Changes of longitudinal diameter and transverse diameter during the pressing process.

Fig. 3. Outer soft pipe and inner hard rod bimetal rod (s−h).

Plane strain deformation

During the pressing process, the longitudinal diameter of the channel is changing only by side view in 2-D and in its
transverse direction, the transverse diameter of the channel is always R. So the assumption can be used here, there is no
spread in transverse direction during the pressing process.
Then the two kinds of the bimetal rods are analyzed respectively. One is an aluminum pipe and a copper rod which can
stand for a soft outer pipe with a hard inner rod. The other one is a copper pipe and an aluminum rod. This kind of rod can
be seen as a hard outer pipe with a soft inner rod. In order to enhance the applicability of the models presented, eccentric
outer pipe is considered in this paper. So that r1 is not equal to r3 and the eccentricity ratio can be expressed as:
r3 − r1
e= , R = r1 + r2 + r3 . (1)
R
The different arrangement of the materials will affect the final bonded rod in eccentricity ratios very seriously.

Soft pipe-hard rod

When a soft pipe and a hard rod are used to be pressed in ECAP, the outer pipe will be firstly yielded, from the Fig. 3
Since both the pressures in layers 1 and 3 can make the soft layers yielded, the outer layers will be pressed together.

Layer 1 : p(s−h)0 − σ = 2ks , (2)


W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 505

Fig. 4. Stresses acting on element in zone IV (s−h).

Layer 3 : p(s−h)0 − σ = 2ks . (3)


The total pressing process can be divided into four zones based on characteristics of the deformation.
When ϕ is in the range between 0 and ϕ I , it is zone I. In this zone, only the outer soft pipe is yielded and the deforma-
tion appears in both layers 1 and 3. The layer 2 (hard inner rod) remains r2 in this zone, while r1 and r3 are both increased.
So the ratios of the three thicknesses are changed. When the bimetal rod comes to the zone II ϕ I ≤ ϕ ≤ ϕ II = β , the three
layers are bonded together and the longitudinal diameter of the channel increases to the maximum. Also when the three
layers are bonded together the ratios of the three layers remain as a constant value to the end. In zone III β = ϕ II ≤ ϕ ≤
ϕ III = α the longitudinal diameter of the channel remains as a constant value. Finally the bonded rod comes to the zone IV
α = ϕIII ≤ ϕ ≤ ϕIV = π2 , in this zone the longitudinal diameter is reduced to the initial R. After this ϕ IV is passed, the ECAP
process can be finished.
As ϕ is small the longitudinal diameter of the channel can be shown as
ϕ2
cos ϕ = 1 − , (4)
2
⎧ R+r0
⎪r(s−h) I , r(s−h) II = − r0

⎨ 1− ϕ2
2

R+r
r(s−h) III = cos β
0
− r0 . (5)

⎪  
⎩r = R+r0
1 − (ϕ −2α )
2
− r0
(s−h ) IV cos β

Since at the exit of the channel (ϕ = ϕ IV ) there is a boundary condition p(s − h)IV123 = 0, the analytical process will follow
the reverse order. The zone IV is selected as objective in the first place.
In the zone IV the bimetal rod is pressed together and the longitudinal diameter of the channel is reduced.
From the Fig. 4 the Eq. (6) can be obtained.
d ( pr )
( R+r0 )
+ σ1 tan θ + τ3 + τ1 = 0, (6)
cos β

where σ = σ 3 = σ 1 − τ 1 tan θ .
dp dr
r ( R+r ) + p ( R+r ) + σ tan θ + τ1 tan2 θ + τ1 + τ3 = 0. (7)
0
cos β
dϕ 0

cos β

( R+r0 )
When the Eq. (7) is simplified the followings are noticed tan θ = ϕ − α , dr

=− cos β
(ϕ − α ), p123 − σ = 2k123 . k123 =
r2 r(s−h) −r2
r(s−h) kh + I
r(s−h) ks .
I I

dp
r ( R+r ) − 2k123 (ϕ − α ) + τ1 (ϕ − α )2 + τ1 + τ3 = 0, (8)
0
cos β

−τ1 (ϕ − α )2 + 2k123 (ϕ − α ) − τ1 − τ3 (R + r0 )
dp = dϕ . (9)
r(s−h) IV cos β
506 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. 5. Stresses acting on element in zone III (s–h).

Then the pressure in this zone can be expressed as



2 ( R + r0 ) B(s−h)IV123 2F(s−h)IV123
p(s−h)IV123 = −A(s−h)IV123 (ϕ − α ) − ln 2 + − (ϕ − α )2
D(s−h)IV123 cos β 2 D(s−h)IV123
  ⎛ ⎞⎤
2F(s−h )IV123
1+ D
F(s−h )IV123
A(s−h)IV123 +
E(s−h )IV123
⎜ 2 + D(s−h )IV123
+ ϕ − α ⎟⎥
2
ln ⎜ ⎟⎥, +C(s−h)IV123
(s−h )IV123
+ (10)
2F(s−h )IV123
⎝ 2F(s−h )IV123
⎠⎦
2+ D 2+ D −ϕ+α
(s−h )IV123 (s−h )IV123

where
R + r0
A(s−h)IV123 = τ1 = −m1 · k123 , B(s−h)IV123 = 2k123 , D(s−h)IV123 = , F(s−h)IV123 = −r0 ,
cos β

R + r0 (ϕ − α )2
E(s−h)IV = −τ1 − τ3 = −m1 k123 − m1 k123 . r(s−h) IV = 1− − r0 .
cos β 2
Considering the boundary conditions
When ϕ = π2 , p(s − h)IV = 0 then C(s − h)IV-123 can be obtained.
 π  B
2 ( R + r0 ) 2F(s−h)IV123  π 2
−α − −α
(s−h )IV123
C(s−h)IV123 = − −A(s−h)IV123 ln 2 + −
D(s−h)IV123 cos β 2 2 D(s−h)IV123 2
  ⎛ ⎞ ⎤
2F
F E
1 + D(s−h)IV123 A(s−h)IV123 + (s−h2)IV123 2 + D (s−h)IV123 + π2 − α
⎜ (s−h )IV123 ⎟⎥
ln ⎜ ⎟⎥.
(s−h )IV123
+ (11)
2F(s−h )IV123
⎝ 2F(s−h )IV123
⎠⎦
2+ D 2+ D π
− 2 +α
(s−h )IV123 (s−h )IV123

When it is in the zone III the bimetal rod is pressed in the equal channel, the longitudinal diameter of the channel
R+r
remains as cos β0 − r0 .
From the Fig. 5 then the Eq. (12) is received.
d ( pr )
− σ3 tan θ + σ1 tan θ + τ3 + τ1 = 0, (12)
(R + r0 )dϕ
where σ = σ 3 + τ 3 tan θ = σ 1 − τ 1 tan θ . (R+dr(r ))dϕ = 0
0
The Eq. (12) can be simplified as
dp
r + τ3 tan2 θ + τ1 tan2 θ + τ1 + τ3 = 0. (13)
(R + r0 )dϕ
After integral calculation the pressure in this zone can be yielded as
 3 
(τ1 + τ3 ) ϕ3 + ϕ
p(s−h )III123 = −(R + r0 ) + C(s−h )III123 , (14)
r(s−h) III
R+r
where r(s−h) III = cos β0 − r0 , τ 1 = m1 k123 , τ 3 = m1 k123
Noticing that when ϕ = ϕ III
p(s−h)III123 = p(s−h)IV123 (15)
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 507

Fig. 6. Stresses acting on element in zone II (s−h).

So the integration constant C(s − h)III123 can be yielded.



2 ( R + r0 ) B(s−h)IV123 2F(s−h)IV123
C(s−h )III123 = −A(s−h)IV123 (ϕIII − α ) − ln 2 + − (ϕIII − α )2
D(s−h)IV123 cos β 2 D(s−h)IV123
  ⎛ ⎞⎤
2F(s−h )IV123
1+ D
F(s−h )IV123
A(s−h)IV123 +
E(s−h )IV123
⎜ 2 + D(s−h )IV123
+ ϕ III − α ⎟⎥
2
ln ⎜ ⎟⎥
(s−h )IV123
+
2F(s−h )IV123
⎝ 2F(s−h )IV123
⎠⎦
2+ D 2+ D − ϕIII + α
(s−h )IV123 (s−h )IV123
 3 
ϕIII
(τ1 + τ3 ) 3 + ϕIII
+C(s−h)IV123 + (R + r0 ) . (16)
r(s−h) III
Then it comes to the zone II
From the Fig. 6 the stress analyses are shown and the Eq. (17) is obtained.
d ( pr )
− σ1 tan θ + τ3 + τ1 = 0, (17)
(R + r0 )dϕ
where σ = σ 3 = σ 1 + τ 1 tan θ
Noticing that in this zone the longitudinal diameter is increasingtan θ = ϕ , ddrϕ = (R + r0 )ϕ , p123 − σ = 2k123 .
Then the Eq. (17) can be changed into
d p123
r(s−h)II + 2k123 tan θ + τ1 tan2 θ + τ1 + τ3 = 0. (18)
(R + r0 )dϕ
Then the pressure expression can be changed as


1 ϕ3 ϕ2
p(s−h )II123 = (R + r0 ) A(s−h )II123 + B(s−h )II123
F(s−h )II123 3 2
   
2 D(s−h )II123 + F(s−h )II123
− 2A(s−h )II123 − E(s−h )II123 − A(s−h )II123 ϕ
F(s−h )II123
     
D(s−h )II123 + F(s−h )II123 2 D(s−h )II123 + F(s−h )II123
− 1− B(s−h )II−123 ln ϕ 2

F(s−h )II123 F(s−h )II123
   
2
D(s−h )II123 +F(s−h )II123
2A(s−h )II−123 − E(s−h )II123 −
(
2 D(s−h )II123 +F(s−h )II123 )A + E(s−h )II123
F(s−h )II123 F(s−h )II123 (s−h )II123
− arctan

(
2 D(s−h )II123 +F(s−h )II123 )
F(s−h )II123
⎡ ⎤#
ϕ
×⎣ ⎦ + C(s−h)II123 , (19)

(
2 D(s−h )II123 +F(s−h )II123 )
F(s−h )II123

where A(s − h)II123 = m1 k123 , B(s − h)II123 = −2k123 , E(s − h)II123 = −m1 k123 − m1 k123 , D(s − h)II123 = R + r0 , F(s − h)II123 = −r0 , r(s−h)II =
R+r0
ϕ2
− r0 .
1− 2
508 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. 7. Stresses acting on element in zone I (s−h).

Also noticing the boundary condition when it is at ϕ II .


p(s−h)II123 = p(s−h)III123 . (20)
Then the integration constant C(s − h)II123 is yielded

 3 
(τ1 + τ3 ) ϕ3II + ϕII
C(s−h )II123 = −(R + r0 ) + C(s−h )III123
r(s−h) III

1 ϕII 3 ϕII 2
− ( R + r0 ) A + B(s−h )II123
F(s−h )II123 (s−h )II123 3 2
   
2 D(s−h )II123 + F(s−h )II123
− 2A(s−h )II123 − E(s−h )II123 − A(s−h )II123 ϕII
F(s−h )II123
     
D(s−h )II123 + F(s−h )II123 2 D(s−h )II123 + F(s−h )II123
− 1− B(s−h )II−123 ln ϕII 2

F(s−h )II123 F(s−h )II123
   
2
D(s−h )II123 +F(s−h )II123
2A(s−h )II−123 − E(s−h )II123 −
(
2 D(s−h )II123 +F(s−h )II123 )A + E(s−h )II123
F(s−h )II123 F(s−h )II123 (s−h )II123
− arctan

(
2 D(s−h )II123 +F(s−h )II123 )
F(s−h )II123
⎡ ⎤#
ϕII
×⎣ ⎦ . (21)

(
2 D(s−h )II123 +F(s−h )II123 )
F(s−h )II123

Finally it comes to the zone I, in this zone there is no deformation in the inner hard rod. So r2 remains its initial value.
The soft outer pipe layers 1 and 3 are pressed together. And r1 and r3 are increased.
From the Fig. 7, the Eqs. (22)–(24) are obtained.

Layer 1
dp r
− σ1 tan θ + τ1 − τ12 + σ12 tan δ = 0,
1 1
(22)
( R + r 0 )d ϕ
where σ = σ 1 + τ 1 tan θ = σ 12 + τ 12 tan δ
Layer 2
dp r
− σ12 tan δ + σ23 tan δ + τ12 − τ23 = 0,
2 2
(23)
( R + r 0 )d ϕ
where σ = σ 12 + τ 12 tan δ = σ 23 + τ 23 tan δ
Layer 3
dp r
− σ23 tan δ + τ3 + τ23 = 0,
3 3
(24)
( R + r 0 )d ϕ
where σ = σ 3 = σ 23 + τ 23 tan δ

Considering layers 1 and 3, from Eqs. (22) and (24)


d p1 r1
− σ tan θ + τ1 tan θ 2 + σ tan δ − τ12 tan2 δ + τ1 − τ12 = 0, (25)
(R + r0 )dϕ
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 509

d p3 r3
− σ tan δ + τ23 tan2 δ + τ3 + τ23 = 0. (26)
(R + r0 )dϕ

= (R + r0 )ϕ , tan θ = ϕ ,p13 − σ = 2ks .


dr13
Adding Eqs. (25) and (26), at the same time noticing that dϕ

dp
( r1 + r3 ) + 2ks tan θ + τ1 + τ3 + τ1 tan2 θ = 0. (27)
(R + r0 )dϕ
Then Eq. (27) can be changed as


1 ϕ3 ϕ2
p(s−h )I13 = (R + r0 ) A(s−h )I13 + B(s−h )I13
F(s−h )I13 3 2
   
2 D(s−h )I13 + F(s−h )I13
− 2A(s−h )I13 − E(s−h )I13 − A(s−h )I13 ϕ
F(s−h )I13
     
D(s−h )I13 + F(s−h )I13 2 D(s−h )I13 + F(s−h )I13
− 1− B(s−h )I13 ln ϕ 2

F(s−h )I13 F(s−h )I13
   
2
D(s−h )I13 +F(s−h )I13
2A(s−h )I13 − E(s−h )I13 −
(
2 D(s−h )I13 +F(s−h )I13 )A + E(s−h )I13
F(s−h )I13 F (s−h )I13
− arctan

(
2 D(s−h )I13 +F(s−h )I13 )
F(s−h )I13
⎛ ⎞#
ϕ
×⎝ ⎠ + C(s−h)I13 , (28)

(
2 D(s−h )I13 +F(s−h )I13 )
F(s−h )I13

R+r0
where A(s − h)I13 = −m1 ks , B(s − h)I13 = −2ks , E(s − h)I13 = −m1 ks − m1 ks , D(s − h)I13 = R + r0 , F(s − h)I13 = −r0 − r2 .r(s−h)I = ϕ2
− r0 , r1 +
1− 2
R+r0
r3 = ϕ2
− r0 − r2 .
1− 2
d (r )
Then from the Layer 2, Eq. (23) can be changed as following with noticing (R+r 2)dϕ = 0.
0

R + r0
p(s−h )I−2 = ϕ + C(s−h)I2 . (29)
r2
Based on the boundary condition, when ϕ = ϕ I

p(s−h)II123 − σ = 2k123 , (30)

p(s−h)I2 − σ = 2kh . (31)

In zone II the three layers are pressed together so yield criterion Eq. (30) is satisfied. Also at this ϕ , the hard inner rod
layer 2 is yielded Eq. (31).
Based on the above Eqs.

p(s−h)II123 − 2k123 + 2kh = p(s−h)I2 . (32)

Then C(s − h)I2 is yielded.


1 ϕI 3 ϕI 2
C(s−h )I2 = (R + r0 ) A(s−h )II123 + B(s−h )II123
F(s−h )II123 3 2
   
2 D(s−h )II123 + F(s−h )II123
− 2A(s−h )II123 − E(s−h )II123 − A(s−h )II123 ϕI
F(s−h )II123
     
D(s−h )II123 + F(s−h )II123 2 D(s−h )II123 + F(s−h )II123
− 1− B(s−h )II−123 ln ϕI 2 −
F(s−h )II123 F(s−h )II123
510 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. 8. Schematic diagram of metal flow when β = 0.

Fig. 9. Effects of r0 on e, ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) as well as r2f


R
.

   
2
D(s−h )II123 +F(s−h )II123
2A(s−h )II−123 − E(s−h )II123 −
(
2 D(s−h )II123 +F(s−h )II123 )A + E(s−h )II123
F(s−h )II123 F(s−h )II123 (s−h )II123
− arctan

(
2 D(s−h )II123 +F(s−h )II123 )
F(s−h )II123
⎡ ⎤#
ϕI ⎦ + C(s−h)II123 . − 2k123 + 2kh − R + r0 ϕI
×⎣ (33)

(
2 D(s−h )II123 +F(s−h )II123 ) r2
F(s−h )II123
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 511

Fig. 10. Effects of r1 on e, ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) as well as r2f


R
.

Since in zone I the layers 1 and 3 are pressed and they are yielded together, at ϕ = ϕ I
p(s−h)I13 − σ = 2ks . (34)
With Eq. (30) Then constant value C(s − h)I13 is yielded

    
1 ϕI 3 ϕI 2 2 D(s−h )II123 + F(s−h )II123
C(s−h )I13 = (R + r0 ) A(s−h )II123 + B(s−h )II123 − 2A(s−h )II123 − E(s−h )II123 − A(s−h )II123 ϕI
F(s−h )II123 3 2 F(s−h )II123
     
D(s−h )II123 + F(s−h )II123 2 D(s−h )II123 + F(s−h )II123
− 1− B(s−h )II−123 ln ϕI 2 −
F(s−h )II123 F(s−h )II123
    ⎡ ⎤#
2
D(s−h )II123 +F(s−h )II123
2A(s−h )II−123 − E(s−h )II123 −
(
2 D(s−h )II123 +F(s−h )II123 )A + E(s−h )II123
F(s−h )II123 F(s−h )II123 (s−h )II123 ϕI
− arctan ⎣ ⎦

(
2 D(s−h )II123 +F(s−h )II123 ) −
(
2 D(s−h )II123 +F(s−h )II123 )
F(s−h )II123 F(s−h )II123

+C(s−h )II123 − 2k123 + 2ks


    
1 ϕI 3 ϕI 2 2 D(s−h )I13 + F(s−h )I13
− ( R + r0 ) A(s−h )I13 + B(s−h )I13 − 2A(s−h )I13 − E(s−h )I13 − A(s−h )I13 ϕI
F(s−h )I13 3 2 F(s−h )I13
     
D(s−h )I13 + F(s−h )I13 2 D(s−h )I13 + F(s−h )I13
− 1− B(s−h )I13 ln ϕI 2 −
F(s−h )I13 F(s−h )I13
    ⎛ ⎞#
2
D(s−h )I13 +F(s−h )I13
2A(s−h )I13 − E(s−h )I13 −
(
2 D(s−h )I13 +F(s−h )I13 )A + E(s−h )I13
F(s−h )I13 F (s−h )I13 ϕI
− arctan ⎝ ⎠ . (35)

(
2 D(s−h )I13 +F(s−h )I13 ) −
(
2 D(s−h )I13 +F(s−h )I13 )
F(s−h )I13 F(s−h )I13

So only ϕ I is unknown, then the final equation can be shown as follow:


512 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. 11. Effects of r2 on e, ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) as well as r2f


R
.

When ϕ = 0
p(s−h)I2 = p(s−h)I13 = p(s−h) 0 . (36)

Restriction of analysis

Besides the ordinary situations some special ones are also discussed here. As it is shown in Fig. 8 when β = 0, during the
pressing process the longitudinal diameter of the channel is always R, when the bimetal rod enters the channel and all the
three layers will be pressed together with their initial r1 and r2 as well as r3 .
Since
R = r1 + r2 + r3 . (37)
if soft layers are pressed while the hard ones are not
$
rs ↓ +rh + rs ↓
R = . (38)
rh + rs ↓ +rh

The longitudinal diameter will be changed which is incorrect. So when it is in the equal longitudinal diameter zone,
For s–h
ϕII = β ≤ ϕ ≤ α = ϕIII . (39)
For h−s
ϕIII = β ≤ ϕ ≤ α = ϕIV . (40)
The thickness of each layer will be kept as a constant value. When β is too small
For s–h
ϕI = ϕII = β , (41)
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 513

Fig. 12. Effects of kh on e, ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) as well as r2f


R
.

and for h−s


ϕI ≤ ϕII = ϕIII = β , ϕI = ϕII = ϕIII = β . (42)
That means when the metal is pressed to ϕ = β , the channel will meet an inflection point where the layers will remain
their thickness ratio at that time to the exit. ϕ I(s − h) and ϕ I(h − s) as well as ϕ II(h − s) may be equal to β . A small β channel
will make the thickness ratio of the layers changed a little but a big one will make each thickness seriously different.

The final thicknesses

In order to research the final thicknesses of each layer, here the equations are obtained based on above analysis.
As for s–h from 0 ≤ ϕ ≤ ϕ I the layer 2 remains its thickness. The thicknesses of layers 1 and 3 are both added. Since
they are pressed together, the thickness ratios of them remain as the initial one. When it is the point ϕ = ϕ I , the thickness
can be as following

R + r0 r1 R + r0 r3
r 1ϕI = ϕI 2
− r0 − r2 , r 2ϕI = r 2 , r 3ϕI = ϕI 2
− r0 − r2 . (43)
1− r1 + r3 1− r1 + r3
2 2

After this point the three layers are all pressed and the thickness ratio can be remained as it is at ϕ = ϕ I
So the final thicknesses can be expressed as
r 1ϕI r 2ϕI r 3ϕI
r1f = R , r2f = R , r3f = R . (44)
r 1ϕI + r 2ϕI + r 3ϕI r 1ϕI + r 2ϕI + r 3ϕI r 1ϕI + r 2ϕI + r 3ϕI
For h−s, make the situation of r1 > r3 as an example
In 0 ≤ ϕ ≤ ϕ I only inner soft layer 2 is pressed. The layers 1 and 3 remain their thicknesses r1 and r3 . At the point
ϕ = ϕ I , the layers thicknesses can be
R + r0
r 1ϕI = r 1 , r 2ϕI = − r0 − r1 − r3 , r 3ϕI = r 3 . (45)
1 − ϕ2I
2
514 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. 13. Effects of β on e, ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) as well as r2f


R
.

10mm

40mm

120mm

z
180mm

50mm

Fig. 14. Experiment for ECAP.


W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 515

Fig. 15. Effects of β on r1f , r2f and r3f as well as e.

When it comes to the zone II ϕ I ≤ ϕ ≤ ϕ II , layers 1 and 2 are both pressed and their thickness ratio will remain as a
constant. Only layer 3 is still r3
R + r0 r 1ϕI R + r0 r 2ϕI
r1 ϕII = − r0 − r3 , r2 ϕII = − r0 − r3 , r3ϕII = r3 . (46)
1 − ϕ2II r 1ϕI + r 2ϕI 1 − ϕ2II r 1ϕI + r 2ϕI
2 2

After this ϕ II , all the three layers are pressed and the ratios among them will be kept. Then the final thicknesses can be
obtained as
r1 ϕII r2 ϕII r3 ϕII
r1f =R , r2f =R , r3f =R . (47)
r1 ϕII + r2 ϕII + r3 ϕII r1 ϕII + r2 ϕII + r3 ϕII r1 ϕII + r2 ϕII + r3 ϕII

Discussions

Here some figures are received through the analytical solutions above. Based on them the effects of r0 , r1 , r2 , kh and β
r
on e, ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) as well as R2f are shown.
r −r
As it is shown in the Fig. 9, when the r0 is increased from 4 mm to 8 mm, einitial is a constant value 3 R 1 for both
r3f −r1f
s–h and h−s. When the bimetal rods is are made of a soft outer pipe and a hard inner rod, e(s−h ) = R is reduced a
r −r
little while when it is hard outer and soft inner, e(h−s ) = 3f R 1f will be increased. Both the two eccentricity ratio e(s − h)
and e(h − s) are closer to the initial one with increasing r0 . This situation can be expressed by the Fig. 17(b), when the r0 is
increasing, ϕ I(s − h) and ϕ I(h − s) as well as ϕ II(h − s) will be reduced which means there is less time for the thickness ratios to
be changed. And a bigger r0 can make the deformation to appear in all the three layers more quickly. Since r0 is bigger,
r r
a more uniform deformation will also occur during the pressing process. As for the last Fig. 9(c) both R2f (s−h) and R2f (h−s)
will be increased a little. It can be easily explained by Eqs. (44) and (47), when ϕ I(s − h) and ϕ I(h − s) as well as ϕ II(h − s) are all
reduced, r1ϕI + r2ϕI + r3ϕI and r1 ϕII + r2 ϕII + r3 ϕII will be both reduced.
516 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. 16. Effects of r1 on r1f , r2f and r3f as well as e.

Fig. 17. Effects of r2 on r1f , r2f and r3f as well as e.

In the Fig. 10, r1 is increased from 2.0 mm to 2.8 mm and r3 is also reduced from 3 mm to 2.2 mm. From the figures all
r2 is 3 mm and the R is 8 mm. The eccentricity ratio is from a positive value to a negative one. When einitial is a positive
value, e(s − h) can be bigger than the initial one but e(h − s) can be less than it. When it comes to the negative value (r1 > r3 ),
e(s − h) is the smallest one and e(h − s) becomes the biggest. This can be expressed by |e(h − s) | ≤ |einitial | ≤ |e(s − h) |. So when the
eccentricity occurs, a soft outer pipe and hard inner rod may let this eccentricity become more serious. However, when
it is a hard outer and soft inner, this eccentricity may be less after the pressing process. When r1 is increased and r3 is
reduced, ϕ I(s − h) will remains as a constant value. This is because both layers 1 and 3 will be pressed together. But when it
is h−s, ϕ I(h − s) is the first to be increased and then reduced while ϕ II(h − s) is reduced at first and then goes up. The center of
r2f
the symmetry will be r1 = r3 = 2.5 mm. R (s−h ) remains a fixed value based on both Eqs. (43) and (44). Due the changes of
r2f
ϕ I(h − s) and ϕ II(h − s) , slightly goes up and then goes down.
R (h−s )
As it is shown in the Fig. 11, when r2 is increased from 2 mm to 6 mm, the R is also increased from 7 mm to 11 mm. All
r1 is 2.4 mm and r3 is 2.6 mm. The r0 is fixed as 7 mm. Both e(s − h) and e(h − s) will be reduced. But when r2 is beyond 4 mm,
e(h − s) will be sharply decreased. This situation can be explained by the Fig. 11(b). When r2 is increased more than 4 mm,
ϕ I(h − s) and ϕ II(h − s) will come close to β . Based on the Eq. (42), they will be equal to β . For this reason, the eccentricity ratio
will suddenly changes. When r2 is increased the final r2 /R will also be increased in both two situations s–h and h–s.
From the Fig. 12 when shear yield strength of the hard metal is increased from 3.4 MPa to 4.2 MPa, e(s − h) will softly go
up while e(h − s) will be reduced. The reason for this situation can be described in Fig. 12(b), when the hard metal is harder,
much more time is needed to finish the bonding process. ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) are all increased when kh is increased.
Also when the kh is beyond 4.2 MPa, ϕ I(h − s) and ϕ II(h − s) are equal to β . This makes e(h − s) to be changed suddenly. As it is
r r
shown in the Fig. 12(c), the inner R2f (h−s) will be increased while R2f (s−h) will be reduced with increasing kh . This is because
the harder metal will make the inner soft to be amplified. But when the outer is a soft one, the same amplification will
appear in the outer pipe. Then the inner hard rod will become smaller than initial one.
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 517

z z
z
y
y
x
z
z
x

y o
Transverse
z
Direction z z
y
Longitudinal y y
z
Direction
y

Fig. 18. Bonding effects along y, z directions.

Table 1
Mold sizes in experiments.

Number R (mm) r0 (mm) β


1 10 2 0
2 10 2 0.3489
3 10 2 0.6105

Table 2
Materials initial sizes in experiments.

Group Materials arrangement |r1 − r3 | (mm) r2 (mm)

A Al–Cu 0.2–1 6
B Cu–Al 0.2–1 6
C Al–Cu 0.6 4–8
D Cu–Al 0.6 4–8

Table 3
Materials shear yield strength.

Materials Chemical compositions Shear yield strength

Al 99.5% Al k = 49 MPa
Cu 99.99% Cu k = 90.5 MPa

r
When β is increasing from 0 to 29 π , all the changes of e, ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) as well as R2f are shown in the Fig.
13. When β = 0,e(h − s) = einitial = e(s − h) . When β is increased, e(h − s) will be decreased and e(s − h) will be increased. From the
Fig. 13(b), before β = π6 , all ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) are equal to β . After β is big enough, these parameters start to have
their independent value respectively. Also the variation trends of e(h − s) and e(s − h) are eased off slowly. Similar situations are
r r
shown in the Fig. 13(c) and the same reasons can be explained the changes of R2f (h−s) and R2f (s−h) .
All the Figs. 8–13 shows that |e(h − s) | ≤ |einitial | ≤ |e(s − h) |. This means during this ECAP process, if the outer pipe is soft,
the initial eccentricity ratio can be enlarged and if it is a hard pipe outside, the ratio can be relieved by this process.

Experiments

In order to verify the results calculated by the models proposed above, bimetal rods with outer aluminum pipe inner
copper rod and outer copper pipe inner aluminum rod were used in the experiments. They were designed with the same
initial sizes. Also there were three molds with different β (Tables 1 and 2).
The materials were designed with different |r1 − r3 | 0.2, 0.4, 0.6, 0.8 and 1 mm. These samples were used to research both
r1 − r3 ≤ 0 and r1 − r3 ≥ 0. r2 was also researched respectively with different 4, 5, 6, 7 and 8 mm (Fig. 14).
Also in Table 3 the material shear yield strengths are shown.
All the experiments were conducted with the same machine. After the experiments were finished, the mold could be
opened easily. The bonded bimetal rods were then observed under microscope. Then the thicknesses of each layer were
518 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

measured. The values measured were used to compare with these calculated by the models. The results are shown by the
following figures.
From the Fig. 15 different molds are used in the experiments. In (a) and (b), the outer pipes are aluminum and the inner
rods are hard copper. But in the (c) and (d) the outer hard layers are copper and inner ones are soft aluminum. Also in
both (a) and (c), r1 is 1.7 mm and r3 is 2.3 mm, r1 − r3 = −0.6 mm. But in the (b) and (d), r1 = 2.3 mm and r3 is 1.7 mm,
r1 − r3 = 0.6 mm. Based on the figures when β is increasing, in (a) and (b) r2f is decreasing. r1f andr3f are also increasing with
this change. Since the outer pipe is aluminum, when β is increasing, the outer parts will be larger. However, in both (c) and
(d), when β is increasing, r2f is increasing, r1f and r3f are decreasing. Also because of the different material arrangements,
the inner soft rods will be larger with this change. In these four figures, the results calculated are all close to these measured
in the experiments. The eccentricity ratios calculated are also close to the measured ones in all the figures.
From the Fig. 16 when r1 is increasing from 1.5 mm to 1.9 mm, r3 is decreasing from 2.5 mm to 2.1 mm. With a fixed
r2 = 6 mm, r3 − r1 is decreasing from 1 mm to 0.2 mm. The same mold is used here with β = π9 . Also the Figs. 16(a) and
(b) are used to represent different material arrangements, for (a) the outer is soft aluminum while for (b) the outer is hard
copper. From the figures, all r2f remains as a constant value due to the same initial r2 and fixed β = π9 . But r3f will be
reduced and r1f will be increased with the changes. When r1 is increasing, r1f and r3f in (b) will be closer than that in (a).
Also the eccentricity ratios are accurately predicted by the models.
As it is shown in the Fig. 17, when r2 is increasing from 4 mm to 8 mm, r3 −r1 is a constant value 0.6 mm. In (a) the
materials are aluminum pipe and copper rod while in (b) the materials are reverse with outer copper and inner aluminum.
β = π9 is also fixed during these experiments. When r2 is increasing, r2f is also increasing obviously. And both r1f andr3f will
be decreasing with this change. Though both in (a) and (b) r1f and r3f will be reduced, in (b) the two parameters will be
closer to each other than that in (a).
During all the experiments the errors are small enough. When the outer is copper pipe the errors may be a little larger
than these with outer aluminum pipe. This may be because the soft inner rod may entirely receive the pressures from the
channel. The pressure delivery through the hard outer pipe may be difficult than a soft one. And when outer is hard copper
and inner is soft aluminum, the r1f , r3f will be closer to each other than that in different materials arrangements. After the
calculation, all the eccentricity ratios satisfy |e(h − s) | ≤ |einitial | ≤ |e(s − h) |and that means r1f will be closer to r3f during the
ECAP when the outer is hard, but when the outer is soft the difference between r1f and r3f will be larger than initial one.
As it is shown in Fig. 18, during ECAP, the deformation appears in the x–z plane or the plane parallel to the x–z plane.
When the bimetal rods are bonded after ECAP process, the bonding effects can be different along different directions. When
it is in y direction, some bonding defects can be shown because of no transverse spread. When it is in z direction, the
bonding effects can be better than that in y direction.

5. Conclusions

1. Based on the slab method unbond bimetal rods are researched with ECAP process. Both two kinds of material arrange-
ments are considered in this paper. One is that the outer is soft pipe, while the inner is hard rod. The other one is
outer hard and inner soft.
r
2. The effects of r0 , r1 , r2 , kh and β on e, ϕ I(s − h) , ϕ I(h − s) and ϕ II(h − s) as well as R2f are researched. All the results are shown
in the figures. After the process, because of |e(h − s) | ≤ |einitial | ≤ |e(s − h) |, when the outer is soft pipe, the eccentricity is
more serious. But when the pipe is hard this eccentricity is relieved.
3. The aluminum and copper are used as the materials to be conducted in the ECAP experiments. And all the results
calculated by the proposed model can be close to those measured in the experiments.

Appendix

A.1. Hard pipe-soft rod

The situation will be a little complex when the bimetal rod is composed with a hard pipe and a soft inner rod. When it
is in the zone I, the inner soft rod is yielded and the outer hard pipe remains to its initial thickness.

Layer 2 p(h−s)0 − σ = 2ks . (A.1)

Layers 1 and 3 p(h−s)0 − σ < 2kh . (A.2)

When it comes to the next zone, since the pressures delivered in layers 1 and 3 are not equal to each other,

p(h−s)1 = p(h−s)3 . (A.3)


So there are two kinds of situations need to be discussed.
When e < 0, r1 > r3 , in zone II the deformation will appear in layer 1;
When e > 0, r3 > r1 , in zone II there will be deformation in layer 3.
The above conclusion will be proved particularly by derivations in zone I.
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 519

In Fig. A.1 based on the characteristics of the deformation during the pressing process, five zones are divided in these
two situations. In zone I 0 ≤ ϕ ≤ ϕ I , only layer 2 the soft inner rod is yielded and r2 is increased. Both the layers 1 and 3
remain at their initial r1 and r3 . Then it is zone II ϕ I ≤ ϕ ≤ ϕ II . Only one of layers 1 and 3 is yielded and it will bonded
with layer 2, the other one will remain at its thickness in zone II. When it comes into zone III ϕ II ≤ ϕ ≤ ϕ III = β the three
layers will be pressed together. And the ratios of the thickness will remain as constant values to the end. In zone IV β = ϕ III
≤ ϕ ≤ ϕ IV = α , the longitudinal diameter will be a constant value while in zone V α = ϕIV ≤ ϕ ≤ ϕV = π2 , the longitudinal
diameter will be reduced to initial R.
The longitudinal diameter in these situations can be expressed as following

⎪r ,r ,r = R+r0 − r0
⎨ (h−s) I (h−s) II (h−s) III 1− ϕ22
R+r0
r(h−s) IV = cos β− r0 . (A.4)
⎪ 
⎩r = R+r0
1 − ( ϕ −α ) − r
2

(h−s ) V cos β 2 0

Also as in the final zone, Zone V, when it is at ϕ = π2 there is a boundary condition p(h−s)V123 = 0
From the final zone V in Fig. A.2, the following equation can be obtained as
d ( p123 r123 )
( R+r0 )
+ σ1 tan θ + τ1 + τ3 = 0, (A.5)
cos β

Fig. A.1. Outer hard pipe and inner soft rod bimetal rod (h−s) left r1 − r3 ≥ 0, right r1 − r3 ≤ 0.

Fig. A.2. Stresses acting on element in zone V (h−s).


520 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. A.3. Stresses acting on element in zone IV (h−s).

( R+r0 )
where σ = σ 1 − τ 1 tan θ = σ 3 also noticing that tan θ = ϕ − α , (ϕ − α ),p123 − σ = 2k123 ,
dr r3

=− cos β
k123 = r(h−s) kh +
II
r(h−s) −r3
II
r(h−s) k12 .
II
Then Eq. (A.5) can be changed as
2k123 (ϕ − α ) − τ1 (ϕ − α )2 − τ1 − τ3 (R + r0 )
dp = dϕ , (A.6)
r(h−s)V cos β

2 ( R + r0 ) B(h−s)V123 2F(h−s)V123
p(h−s)V123 = −A(h−s)V123 (ϕ − α ) − ln 2 + − (ϕ − α )2
D(h−s)V123 cos β 2 D(h−s)V123
  ⎛ ⎞⎤
2F
F E
1 + D(h−s)V123 A(h−s)V123 + (h−s2)V123 2 + D (h−s)V123 + ϕ − α
⎜ (h−s )V123 ⎟⎥
ln ⎜ ⎟⎥ + C(h−s)V123 ,
(h−s )V123
+ (A.7)
2F(h−s )V123
⎝ 2F(h−s )V123
⎠⎦
2+ D 2+ D −ϕ+α
(h−s )V123 (h−s )V123

R+r0
where A(h − s)V123 = −m1 k123 , B(h − s)V123 = 2k123 , D(h−s)V123 = cos β
, E(h − s)V123 = −m1 k123 − m1 k123 , F(h − s)V123 = −r0 , r(h−s) V =
R+r0 (ϕ −α ) 2

cos β
( 1 − 2 ) − r0 .
Considering the boundary condition as when it is at ϕ = π2 , p(h−s)V123 = 0. C(h−s)V123 can be expressed as
 π  B
2 ( R + r0 ) 2F(h−s)V123  π 2
−α − −α
(h−s )V123
C(h−s)V123 =− −A(h−s)V123 ln 2 + −
D(h−s)V123 cos β 2 2 D(h−s)V123 2
  ⎛ ⎞ ⎤
2F
F E
1 + D(h−s)V123 A(h−s)V123 + (h−s2)V123 2 + D (h−s)V123 + π2 − α
⎜ (h−s )V123 ⎟⎥
ln ⎜ ⎟⎥.
(h−s )V123
+ (A.8)
2F(h−s )V123
⎝ 2F(h−s )V123
⎠⎦
2+ D 2+ D π
− 2 +α
(h−s )V123 (h−s )V123

Then it comes to the zone IV


When it is in the zone IV, as it is shown in Fig. A.3, the bimetal rod are pressed together and the longitudinal diameter
of the channel remains as a constant value. From the Fig. A.3 the following equation can be yielded.
d ( p123 r123 )
− σ1 tan θ + σ3 tan θ + τ1 + τ3 = 0, (A.9)
(R + r0 )dϕ
where σ = σ 1 + τ 1 tan θ = σ 3 − τ 3 tan θ At the same time noticing that in this zone, the longitudinal diameter of the channel
d (r )
is a constant value. So (R+(hr−s))dIVϕ = 0
0
 
−τ1 tan2 θ − τ3 tan2 θ − τ1 − τ3
d ( p123 ) = (R + r0 )dϕ . (A.10)
r(h−s)IV
So p(h − s)IV123 can be described as
 
−(τ1 + τ3 ) ϕ3 + ϕ
3

p(h−s)IV123 = (R + r0 ) + C(h−s)IV123 , (A.11)


r(h−s)IV
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 521

Fig. A.4. Stresses acting on element in zone III (h−s).

R+r
where r(h−s) IV = cos β0 − r0 , τ 1 = m1 k123 , τ 3 = m1 k123 .
When it is at ϕ = ϕ IV .

p(h−s)IV123 = p(h−s)V123 . (A.12)


Then C(h − s)IV123 can be yielded as

2 ( R + r0 ) B(h−s)V123 2F(h−s)V123
C(h−s)IV123 = −A(h−s)V123 (ϕIV − α ) − ln 2 + − (ϕIV − α )2
D(h−s)V123 cos β 2 D(h−s)V123
  ⎛ ⎞⎤
2F(h−s )V123
1+ D
F(h−s )V123
A(h−s)V123 +
E(h−s )V123
⎜ 2 + D(h−s )V123
+ ϕ IV − α ⎟⎥
2
ln ⎜ ⎟⎥
(h−s )V123
+
2F(h−s )V123
⎝ 2F(h−s )V123
⎠⎦
2+ D 2+ D − ϕIV + α
(h−s )V123 (h−s )V123
 3 
(τ1 + τ3 ) ϕIV3 + ϕIV
+C(h−s)V123 + (R + r0 ). (A.13)
r123
In Fig. A.4 then it comes to the zone III where the bimetal rod is firstly bonded together.
d ( p123 r123 )
− σ1 tan θ + τ1 + τ3 = 0, (A.14)
(R + r0 )dϕ
d (r123 )
where σ = σ 1 + τ 1 tan θ = σ 3 also noticing that tan θ = ϕ , dϕ
= (R + r0 )ϕ , p123 − σ = 2k123
The Eq. (A.14) can be changed as

−2k123 ϕ − τ1 tan2 θ − τ1 − τ3
d p123 = . (A.15)
r123
Then the pressure in this zone III can be shown as


1 ϕ3 ϕ2
p(h−s)III123 = (R + r0 ) A(h−s)III123 + B(h−s)III123
F(h−s)III123 3 2
   
2 D(h−s)III123 + F(h−s)III123
− 2A(h−s)III123 − E(h−s)III123 − A(h−s)III123 ϕ
F(h−s)III123
     
D(h−s)III123 + F(h−s)III123 2 D(h−s)III123 + F(h−s)III123
− 1− B(h−s)III123 ln ϕ2 −
F(h−s)III123 F(h−s)III123
   
2
D(h−s)III123 +F(h−s)III123
2A(h−s)III123 − E(h−s)III123 −
(
2 D(h−s)III123 +F(h−s)III123 )A + E(h−s)III123
F(h−s)III123 F(h−s)III123 (h−s)III123
− arctan

(
2 D(h−s)III123 +F(h−s)III123 )
F(h−s)III123
⎛ ⎞#
ϕ
×⎝ ⎠ + C(h−s)III123 , (A.16)

(
2 D(h−s)III123 +F(h−s)III123 )
F(h−s)III123
522 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. A.5. Stresses acting on element in zone II (h−s) r1 − r3 ≥ 0.

where A(h − s)III123 = −τ 1 = −m1 k123 , B(h − s)III123 = −2k123 , E(h − s)III123 = −τ 1 − τ 3 = −m1 k123 − m1 k123 , D(h − s)III123 = R + r0 ,
F(h − s)III123 = −r0 .
When it is at ϕ = ϕ III .

p(h−s)III123 = p(h−s)IV123 . (A.17)

Then the integration constant C(h − s)III123 can be received.


 
−(τ1 + τ3 ) ϕ3III + ϕIII
3

1 ϕIII 3 ϕIII 2
C(h−s)III123 = (R + r0 ) + C(h−s)IV123 − (R + r0 ) A(h−s)III123 + B(h−s)III123
r123 F(h−s)III123 3 2
   
2 D(h−s)III123 + F(h−s)III123
− 2A(h−s)III123 − E(h−s)III123 − A(h−s)III123 ϕIII
F(h−s)III123
     
D(h−s)III123 + F(h−s)III123 2 D(h−s)III123 + F(h−s)III123
− 1− B(h−s)III123 ln ϕIII 2

F(h−s)III123 F(h−s)III123
   
2
D(h−s)III123 +F(h−s)III123
2A(h−s)III123 − E(h−s)III123 −
(
2 D(h−s)III123 +F(h−s)III123 )A + E(h−s)III123
F(h−s)III123 F(h−s)III123 (h−s)III123
− arctan

(
2 D(h−s)III123 +F(h−s)III123 )
F(h−s)III123
⎛ ⎞#
ϕIII
×⎝ ⎠ . (A.18)

(
2 D(h−s)III123 +F(h−s)III123 )
F(h−s)III123

When r1 > r3 the layer 1 can be firstly pressed with inner layer 2, From the Fig. A.5 the following equation can be yielded
as
d ( p12 r12 )
− σ1 tan θ + τ1 + τ23 = 0, (A.19)
(R + r0 )dϕ
d (r12 )
where σ = σ 23 = σ 1 + τ 1 tan θ , tan θ = ϕ , dϕ
= (R + r0 )ϕ , p12 − σ = 2k12 .
Then the Eq. (A.19) can be changed as

−2k12 ϕ − τ1 ϕ 2 − τ1 − τ23
d p12 = (R + r0 ) dϕ . (A.20)
r12
So the pressure in this zone II can be expressed as
    
1 ϕ3 ϕ2 2 D(h−s)II12 + F(h−s)II12
p(h−s)I12 = (R + r0 ) A(h−s)II12 + B(h−s)II12 − 2A(h−s)II12 − E(h−s)II12 − A(h−s)II12 ϕ
F(h−s)II12 3 2 F(h−s)II12
     
D(h−s)II12 + F(h−s)II12 2 D(h−s)II12 + F(h−s)II12
− 1− B(h−s)II12 ln ϕ2 −
F(h−s)II12 F(h−s)II12
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 523

   
2
D(h−s)II12 +F(h−s)II12
2A(h−s)II12 − E(h−s)II12 −
(
2 D(h−s)II12 +F(h−s)II12 )A + E(h−s)II12
F(h−s)II12 F(h−s)II12 (h−s )II12
− arctan

(
2 D(h−s)II12 +F(h−s)II12 )
F(h−s)II12
⎛ ⎞#
ϕ
×⎝ ⎠ + C(h−s)II12 , (A.21)

(
2 D(h−s)II12 +F(h−s)II12 )
F(h−s)II12

where A(h − s)II12 = −m1 k12 , B(h − s)II12 = −2k12 , E(h − s)II12 = −m1 k12 − m2 k12 , D(h − s)II12 = R + r0 , F(h − s)II12 = −r0 − r3 , r12 = r(h−s)II −
R+r0 r r
r3 = ϕ2
− r0 − r3 . k12 = r 1 −r kh + r 2 −r ks .
1− 2 (h−s) I 3 (h−s) I 3

From layer 3

d ( p3 r3 )
+ τ3 − τ23 = 0, (A.23)
(R + r0 )dϕ
d (r )
where σ = σ 3 = σ 23 , (R+r 3)dϕ = 0.
0
Then the Eq. (A.23) can be changed as

(τ23 − τ3 )
p(h−s)II3 = ϕ (R + r0 ) + C(h−s)II3 , (A.24)
r3
where τ 23 = m2 k12 , τ 3 = m1 kh .
Noticing the boundary condition when it is at ϕ = ϕ II

p(h−s)II12 − σ = 2k12 , (A.25)

p(h−s)III123 − σ = 2k123 , (A.26)

So C(h − s)II12 can be yielded

p(h−s)III123 − 2k123 = p(h−s)II12 − 2k12 . (A.27)

The Eq. (A.27) can be changed as



1 ϕII 3 ϕII 2
C(h−s)II12 = (R + r0 ) A(h−s)III123 + B(h−s)III123
F(h−s)III123 3 2
   
2 D(h−s)III123 + F(h−s)III123
− 2A(h−s)III123 − E(h−s)III123 − A(h−s)III123 ϕII
F(h−s)III123
     
D(h−s)III123 + F(h−s)III123 2 D(h−s)III123 + F(h−s)III123
− 1− B(h−s)III123 ln ϕII − 2
F(h−s)III123 F(h−s)III123
   
2
D(h−s)III123 +F(h−s)III123
2A(h−s)III123 − E(h−s)III123 −
(
2 D(h−s)III123 +F(h−s)III123 )A + E(h−s)III123
F(h−s)III123 F(h−s)III123 (h−s)III123
− arctan

(
2 D(h−s)III123 +F(h−s)III123 )
F(h−s)III123
⎛ ⎞#
ϕII
×⎝ ⎠ + C(h−s)III123

(
2 D(h−s)III123 +F(h−s)III123 )
F(h−s)III123

1 ϕII 3 ϕII 2
− ( R + r0 ) A(h−s)II12 + B(h−s)II12
F(h−s)II12 3 2
   
2 D(h−s)II12 + F(h−s)II12
− 2A(h−s)II12 − E(h−s)II12 − A(h−s)II12 ϕII
F(h−s)II12
     
D(h−s)II12 + F(h−s)II12 2 D(h−s)II12 + F(h−s)II12
− 1− B(h−s)II12 ln ϕII 2 −
F(h−s)II12 F(h−s)II12
524 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

Fig. A.6. Stresses acting on element in zone I (h−s).

   
2
D(h−s)II12 +F(h−s)II12
2A(h−s)II12 − E(h−s)II12 −
(
2 D(h−s)II12 +F(h−s)II12 )A + E(h−s)II12
F(h−s)II12 F(h−s)II12 (h−s)II12
− arctan

(
2 D(h−s)II12 +F(h−s)II12 )
F(h−s)II12
⎛ ⎞#
ϕII
×⎝ ⎠ + 2k12 − 2k123 . (A.28)

(
2 D(h−s)II12 +F(h−s)II12 )
F(h−s)II12

As when it is at ϕ = ϕ II
p(h−s)II3 − σ = 2kh , (A.29)

p(h−s)III123 − σ = 2k123 . (A.30)


Then the integration constant C(h − s)II3 can be received

1 ϕII 3 ϕII 2
C(h−s)II3 = (R + r0 ) A(h−s)III123 + B(h−s)III123
F(h−s)III123 3 2
   
2 D(h−s)III123 + F(h−s)III123
− 2A(h−s)III123 − E(h−s)III123 − A(h−s)III123 ϕII
F(h−s)III123
     
D(h−s)III123 + F(h−s)III123 2 D(h−s)III123 + F(h−s)III123
− 1− B(h−s)III123 ln ϕII 2

F(h−s)III123 F(h−s)III123
   
2
D(h−s)III123 +F(h−s)III123
2A(h−s)III123 − E(h−s)III123 −
(
2 D(h−s)III123 +F(h−s)III123 )A + E(h−s)III123
F(h−s)III123 F(h−s)III123 (h−s)III123
− arctan

(
2 D(h−s)III123 +F(h−s)III123 )
F(h−s)III123
⎛ ⎞#
ϕII
×⎝ ⎠

(
2 D(h−s)III123 +F(h−s)III123 )
F(h−s)III123

(τ23 − τ3 )
+C(h−s)III123 − 2k123 − 2kh − ϕII (R + r0 ). (A.31)
r3
Then the first zone is considered. In this zone only the inner soft layer 2 is deformed and the layers 1 and 3 remains at
r1 and r3 .
From the Fig. A.6 following are obtained

Layer 1
d ( p1 r1 )
− σ1 tan θ + σ12 tan θ + τ1 − τ12 = 0, (A.32)
( R + r 0 )d ϕ
where σ = σ 1 + τ 1 tan θ = σ 12 + τ 12 tan θ
Layer 2
d ( p2 r2 )
− σ12 tan θ + τ12 + τ23 = 0, (A.33)
( R + r 0 )d ϕ
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 525

Fig. A.7. Stresses acting on element in zone II (h−s) r1 − r3 ≤ 0.

where σ = σ 12 + τ 12 tan θ = σ 23
Layer 3
d ( p3 r3 )
+ τ3 − τ23 = 0, (A.34)
(R + r0 )dϕ
where σ = σ 3 = σ 23

From the layer 2, the Eq. (A.33) can be changed as


d p2 d r2
r2 + p2 − σ tan θ + τ12 tan2 θ + τ12 + τ23 = 0. (A.35)
( R + r 0 )d ϕ ( R + r 0 )d ϕ
= (R + r0 )ϕ , tan θ = ϕ . p2 − σ = 2ks .
d r2
Noticing that dϕ

−2ks ϕ − τ12 ϕ 2 − τ12 − τ23


dp = R+r0
(R + r0 )dϕ . (A.36)
− r0 − r1 − r3
1− ϕ2
2

Then it yields p(h − s)I2


    
1 ϕ3 ϕ2 2 D(h−s)I2 + F(h−s)I2
p(h−s)I2 = (R + r0 ) A(h−s)I2 + B(h−s)I2 − 2A(h−s)I2 − E(h−s)I2 − A(h−s)I2 ϕ
F(h−s)I2 3 2 F(h−s)I2
     
D(h−s)I2 + F(h−s)I2 2 D(h−s)I2 + F(h−s)I2
− 1− B(h−s)I2 ln ϕ2 −
F(h−s)I2 F(h−s)I2
    ⎛ ⎞#
2
D(h−s)I2 +F(h−s)I2
2A(h−s)I2 − E(h−s)I2 −
(
2 D(h−s)I2 +F(h−s)I2 )A + E(h−s)I2
F(h−s)I2 F(h−s)I2 (h−s)I2 ϕ
− arctan ⎝ ⎠

(
2 D(h−s)I2 +F(h−s)I2 ) −
(
2 D(h−s)I2 +F(h−s)I2 )
F(h−s)I2 F(h−s)I2

+C(h−s)I2 . (A.37)
where A(h − s)I2 = −m2 ks , B(h − s)I2 = −2ks , E(h − s)I2 = −m2 ks − m2 ks , D(h − s)I2 = R + r0 , F(h − s)I2 = −r0 − r1 − r3 .
When ϕ = ϕ I ,
p(h−s)I2 − 2ks = p(h−s)II12 − 2k12 . (A.38)
Then C(h − s)I2 can be obtained.
    
1 ϕI 3 ϕI 2 2 D(h−s)II12 + F(h−s)II12
C(h−s)I2 = (R + r0 ) A(h−s)II12 + B(h−s)II12 − 2A(h−s)II12 − E(h−s)II12 − A(h−s)II12 ϕI
F(h−s)II12 3 2 F(h−s)II12
     
D(h−s)II12 + F(h−s)II12 2 D(h−s)II12 + F(h−s)II12
− 1− B(h−s)II12 ln ϕI 2

F(h−s)II12 F(h−s)II12
   
2
D(h−s)II12 +F(h−s)II12
2A(h−s)II12 − E(h−s)II12 −
(
2 D(h−s)II12 +F(h−s)II12 )A + E(h−s)II12
F(h−s)II12 F(h−s)II12 (h−s)II12
− arctan

(
2 D(h−s)II12 +F(h−s)II12 )
F(h−s)II12
⎛ ⎞#
ϕI
×⎝ ⎠ + C(h−s)II12 − 2k123 + 2ks

(
2 D(h−s)II12 +F(h−s)II12 )
F(h−s)II12
526 W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527

    
1 ϕI 3 ϕI 2 2 D(h−s)I2 + F(h−s)I2
− ( R + r0 ) A(h−s)I2 + B(h−s)I2 − 2A(h−s)I2 − E(h−s)I2 − A(h−s)I2 ϕI
F(h−s)I2 3 2 F(h−s)I2
     
D(h−s)I2 + F(h−s)I2 2 D(h−s)I2 + F(h−s)I2
− 1− B(h−s)I2 ln ϕI 2

F(h−s)I2 F(h−s)I2
   
2
D(h−s)I2 +F(h−s)I2
2A(h−s)I2 − E(h−s)I2 −
(
2 D(h−s)I2 +F(h−s)I2 )A + E(h−s)I2
F(h−s)I2 F(h−s)I2 (h−s)I2
− arctan

(
2 D(h−s)I2 +F(h−s)I2 )
F(h−s)I2
⎛ ⎞#
ϕI
×⎝ ⎠ . (A.39)

(
2 D(h−s)I2 +F(h−s)I2 )
F(h−s)I2

Also at ϕ = ϕ I
p(h−s)I1 − 2kh = p(h−s)I12 − 2k12 , (A.40)

p(h−s)I3 = p(h−s)II3 . (A.41)


From the layer 1

(τ12 − τ1 ) ϕ 3
p(h−s)I1 = + ϕ (R + r0 ) + C(h−s)I1 . (A.42)
r1 3
So from the Eq. (A.42)
    
1 ϕI 3 ϕI 2 2 D(h−s)I2 + F(h−s)I2
C(h−s)I1 = (R + r0 ) A(h−s)I2 + B(h−s)I2 − 2A(h−s)I2 − E(h−s)I2 − A(h−s)I2 ϕI
F(h−s)I2 3 2 F(h−s)I2
     
D(h−s)I2 + F(h−s)I2 2 D(h−s)I2 + F(h−s)I2
− 1− B(h−s)I2 ln ϕI 2 −
F(h−s)I2 F(h−s)I2
    ⎛ ⎞#
2
D(h−s)I2 +F(h−s)I2
2A(h−s)I2 − E(h−s)I2 −
(
2 D(h−s)I2 +F(h−s)I2 )A + E(h−s)I2
F(h−s)I2 F(h−s)I2 (h−s)I2 ϕI
− arctan ⎝ ⎠

(
2 D(h−s)I2 +F(h−s)I2 ) −
(
2 D(h−s)I2 +F(h−s)I2 )
F(h−s)I2 F(h−s)I2

(τ12 − τ1 ) ϕI 3
+C(h−s)I2 − 2k12 + 2kh − + ϕI (R + r0 ). (A.43)
r1 3
Based on the Eq. (A.43)
(τ23 − τ3 ) (τ23 − τ3 )
C(h−s)I3 = ϕI (R + r0 ) + C(h−s)II3 − ϕI (R + r0 ). (A.44)
r3 r3
Then all the parameters are obtained except ϕ I and ϕ II . they can be received by Eq. (A.45)
When it is at ϕ = 0.
p(h−s)I1 = p(h−s)I2 = p(h−s)I3 = p(h−s)0 . (A.45)
Since r1 = r3 the pressures in layers 1 and 3 are not equal to each other.
From the Eqs. (A.32) and (A.34). noticing that both the two parts of the outer hard pipe are undeformed. So the two
layers are shown:

(τ12 − τ1 ) ϕ 3
p(h−s) 1 = + ϕ (R + r0 ) + C1 , (A.46)
r1 3

(τ23 − τ3 )
p(h−s) 3 = ϕ (R + r0 ) + C3 . (A.47)
r3
Since ϕ is very small so we simplify p(h − s)1 based on the Eq. (A.46)

(τ12 − τ1 ) ϕ 3 (τ12 − τ1 )
p(h−s) 1 = + ϕ (R + r0 ) + C1 ≈ ϕ (R + r0 ) + C1 , (A.48)
r1 3 r1
W. Hongyu et al. / Applied Mathematical Modelling 47 (2017) 501–527 527

where τ 12 − τ 1 = m2 ks − m1 kh ≤ 0 and τ 23 − τ 3 = m2 ks − m1 kh ≤ 0
Only if the layers 1 and 3 satisfy the yield criterion, the deformation can appear in the hard layers
p − σ = 2kh . (A.49)
When r1 > r3 , the thicker layer 1 will be first pressed.
p(h−s) 1 − σ > p(h−s) 3 − σ . (A.50)
When r1 < r3 , the deformation will first appear in thicker layer 3.
p(h−s) 1 − σ < p(h−s) 3 − σ . (A.51)
When r3 > r1
In Fig. A.7 the analytical process will be almost the same with r3 < r1 , so the previous parts will not be repeated here.
Based on the Eq. (A.45), when r3 > r1 , ϕ I and ϕ II will be obtained.

References

[1] H.Y. Wang, D.H. Zhang, D.W. Zhao, Analysis of asymmetrical rolling of unbonded clad sheet by slab method considering vertical shear stress, ISIJ Int
55 (2015) 1058–1066.
[2] H.Y. Wang, Z.H. Wang, D.H. Zhang, D.W. Zhao, Analysis of snake rolling force and torque with changes of thickness depending on unequal roll radii
based on pure aluminum experiments, Proc. Inst. Mech. Eng. Part C:J. Mech. Eng. Sci 231 (2017) 161–174.
[3] R.Z. Valiev, R.K. Islamgaliev, I.V. Alexandrov, Bulk nanostructured materials from severe plastic deformation, Prog. Mater. Sci 45 (20 0 0) 103–189.
[4] Z. Horita, T. Fujinami, T.G. Langdon, The potential for scaling ECAP: effect of sample size on grain refinement and mechanical properties, Mater. Sci.
Eng. A 318 (2001) 34–41.
[5] V.M. Segal, Engineering and commercialization of equal channel angular extrusion (ECAE), Mater. Sci. Eng. A 386 (2004) 269–276.
[6] Y.T. Zhu, T.C. Lowe, T.G. Langdon, Performance and applications of nanostructured materials produced by severe plastic deformation, Scr. Mater 51
(2004) 825–830.
[7] H.S. Kim, M.H. Seo, S.I. Hong, Plastic deformation analysis of metals during equal channel angular extrusion, J. Mater. Process Technol 113 (2001)
622–626.
[8] A.R.Eivani, A. Karimi Taheri, A new method for producing bimetallic rods, Mater. Lett. 61 (2007) 4110–4113.
[9] K. Narooei, T.A. Karimi, A new model for prediction the strain field and extrusion pressure in ECAE process of circular cross section, Appl. Math. Model.
34 (2010) 1901–1917.
[10] K. Narooei, T.A. Karimi, Strain field and extrusion load in ECAE process of bi-metal circular cross section, Appl. Math. Model. 36 (2012) 2128–2141.
[11] J.K. Kim, W.J. Kim, Analysis of deformation behavior in 3D during equal channel angular extrusion, J. Mater. Process. Technol. 176 (2006) 260–267.
[12] W Wei, W. Zhang, K.X. Wei, Y. Zhong, G. Cheng, J. Hu, Finite element analysis of deformation behavior in continuous ECAP process, J. Mater. Sci. Eng.
A 516 (2009) 111–118.
[13] H. Jiang, Z.G. Fan, C.Y. Xie, Finite element analysis of temperature rise in CP–Ti during equal channel angular extrusion, J. Mater. Sci. Eng. A 513–514
(2009) 109–114.
[14] V.M. Segal, Materials processing by simple shear, Mater. Sci. Eng. A 197 (1995) 157–164.
[15] M. Reihanian, R. Ebrahimi, M.M. Moshksar, Upper-bound analysis of equal channel angular extrusion using linear and rotational velocity fields, Mater.
Des. 30 (2009) 28–34.
[16] A.R. Eivani, T.A. Karimi, An upper bound solution of ECAE process with outer curved corner, J. Mater. Process. Technol. 182 (2007) 555–563.
[17] H.Y. Wang, X. Li, J. Sun, Z.H. Wang, D.W. Zhao, D.H. Zhang, Analysis of sandwich rolling with two different thicknesses outer layers based on slab
method, Int. J. Mech. Sci 106 (2016) 194–208.
[18] X. Li, H.Y. Wang, J.G. Ding, J.J. Xu, D.H. Zhang, Analysis and prediction of fishtail during V-H hot rolling process, J. Cent. South Univ 22 (2015) 1184–1190.
[19] H.Y. Wang, Y.K. Hu, F. Gao, D.W. Zhao, D.H. Zhang, Nominal friction coefficient in spread formulas based on the lead rolling experiments, Trans.
Nonferrous. Met. Soc. China 25 (2015) 2693–2700.
[20] H.Y. Wang, J.G. Ding, X. Lu, D.H. Zhang, D.W. Zhao, Analysis of ultra-heavy plate rolling force based on thickness temperature gradient elements and
experiment simulations, Proc. Inst. Mech. Eng. Part C:J. Mech. Eng. Sci 231 (2017) 599–615.

You might also like