Beruflich Dokumente
Kultur Dokumente
Technical Report
a r t i c l e
i n f o
Article history:
Received 11 April 2014
Accepted 7 June 2014
Available online 25 June 2014
a b s t r a c t
Accuracy of the nite element simulation of sheet metal forming is signicantly dependent on the correctness of input properties and appropriate selection of material models. In this work, anisotropic yield
criteria namely, Hill 1948, Barlat 1989, Barlat 1996, Barlat 2000 and Cazacu Barlat have been implemented for Ti6Al4V alloy at 400 C. Material constants required for the yield criteria have been determined and deformation behavior in deep drawing process has been analyzed in nite element software.
Also, deep drawing experiments on Ti6Al4V alloy have been performed at 400 C to validate nite element simulation results. Further, comparison of yield criteria based on thickness distribution, earing prole, complexity in material parameter identication and computational time has shown Cazacu-Barlat to
be well suited for deep drawing of Ti6Al4V alloy.
2014 Elsevier Ltd. All rights reserved.
1. Introduction
In recent years, Ti6Al4V alloy is increasingly being used
because of their lightweight characteristic and an excellent combination of strength, corrosion resistance and fabricability [1]. The
importance of this alloy is indicated from the fact that currently
it is mostly used alloy accounting for more than 50% of all titanium
tonnage in the world [2]. Sheet metal forming is one of the major
manufacturing processes to fabricate titanium alloy components
which cannot only reduce the cost due to machining but also
enhance the performance of the products [3]. But the formability
of Ti6Al4V alloy is very poor compared with other traditional
metallic materials at room temperature. The main reasons for poor
formability are low ductility at room temperature due to its hexagonal close-packed structure and high degree of springback [4].
Hence, to overcome the difculty in forming of HCP crystal structure alloys, the best choice would be warm forming [5].
Nowadays, nite element simulations are extensively used to
reduce inaccurate and expensive tryouts in sheet metal industries
[6]. However, the trustworthiness of the numerical simulations largely depends on the input material models used and correctness of
the input material data [7]. Particularly, in warm-forming simulation, selection of an appropriate yield criterion is essential because
it provides an accurate prediction of the observed initial and subsequent yield behaviors of a material [5].
Corresponding author. Tel.: +91 9010451444; fax: +91 40 66303998.
E-mail address: nitink@hyderabad.bits-pilani.ac.in (N. Kotkunde).
http://dx.doi.org/10.1016/j.matdes.2014.06.017
0261-3069/ 2014 Elsevier Ltd. All rights reserved.
337
Fig. 1. Experimental test rig and enlarged view of die with induction furnace.
Fr
r211 r222 2
2R 1
2R 1
r11 r22 2
r12
R1
R1
1=2
1
Al
Fe
Ti
Comp (wt.%)
5.56
4.07
0.185
0.022
89.997
rx hry
2
; K2
r
rx hry 2
2
s
R00 R90
a22
1 R00 1 R90
s
R00 1 R90
h
and c 2 a
R90 1 R00
where rh is anisotropy ratio measured for h orientation from rolling
direction p is found out iteratively. The following equation is used
to nd out the value of p iteratively [12].
rh
@/
@ rx
2r m
y
@@/
ry
rh
1
338
Table 2
Material properties for Ti6Al4V alloy.
Temperature (C)
r0 (MPa)
r45 (MPa)
r90 (MPa)
rC (MPa)
rb (MPa)
r0
r45
r90
rb
400
681.0
591.0
691.0
751.0
744.9
0.6
1.26
0.512
1.02
Here, h is taken as 45 to compare the value and to further iterate. The yield locus plotted by Barlat 1989 model is as shown in
Fig. 2(b).
=r90 a
F /x jc3 2c1 ja /y jc3 c1 ja /z0 jc1 2c3 ja 23r
Barlat et al. [18] developed a yield model for modeling anisotropic materials. The yield function is expressed in Eq. (3)
a
a
/ /1 jS2 S3 j /2 jS3 S1 j /3 jS1 S2 j 2r
=rb a
F /x j2c1 c2 ja /y jc1 2c2 ja /z0 jc1 c2 ja 23r
6
@u
@Sx
r90
2
6
6
where L 6
4
@u
@Sx
c2 c3 =3
c3 =3
c2 =3
c3 =3
c3 c1 =3
c1 =3
c2 =3
c1 =3
c1 c2 =3
07
7
7
05
@u
@S
y
r0
a
2r
@u
@Sx
rb
S Lr
a
2r
r0
@u
@S
y
a
2r
@u
@S
y
1
r90
rb
1
1
c4
c1, c2, c3, c4 are material constants and /i1;2;3 depends upon the orientation of frame of the principal value of S and anisotropy axis [8].
For a plane stress condition when shear stress is zero the yield
function becomes
a
/ /x jSy Sz ja /y jSz Sx ja /z0 jSx Sy ja 2r
where
Sx
c3
c c
Sy
rx 11 3 ry
3
3
c2
c
Sz
rx 1 ry
3
3
Sz
c2
r0
3
When the matrices C and C00 are taken as identity matrix the
above criteria reduces to isotropic case. For the simplicity of calculation, anisotropy parameters and the coefcients of L0 and L00 are
related as
3
72 3
6 0 7 6
0
0 7 a1
6 L12 7 6 1=3
76 7
6 0 7 6
6L 7 6 0
1=3 0 7
74 a2 5
6 21 7 6
7
6 0 7 6
2=3 0 5 a7
4 L22 5 4 0
0
0
1
L066
2
=r0 a
F /x jc2 c3 ja /y j2c2 c3 ja /z0 jc2 2c3 ja 23r
3
2=3 1=3 0
6
7
where T 4 1=3 2=3 0 5
c 3 c2
Sx
r0
3
c3
r0
3
in which X 01;2 and X 001;2 are the principal values of the linearly transformed stress deviator matrices {s},
c 3 c2
c
rx 3 ry
3
3
Sy
L011
L0011
2=3
32
a3
76 7
6 00 7 6
6 L12 7 6 1 2 4 4 0 76 a4 7
76 7
6 00 7 6
6 L 7 6 4 4 4 1 0 76 a5 7
76 7
6 21 7 6
76 7
6 00 7 6
2 2 0 54 a6 5
4 L22 5 4 2 8
0
0
0
0 9
a8
L0066
2
2 0
Fig. 2. Yield loci for Ti6Al4V alloy at 400 C (a) Hill 1948, (b) Barlat 1989, (c) Barlat 1996, (d) Barlat 2000 and (e) Cazacu Barlat.
339
340
a
a
rT
2=31 k 21=31 k
a
a
rC
2=31 k 21=31 k
Table 3
Barlat 1996 model material parameters for Ti6Al4V alloy.
Temperature (C)
C1
C2
C3
C4
ax
ay
az0
400
1.059
0.939
1.233
0.992
0.992
sx cr; sy dr
The material parameters are determined by solving two equations given below for each state of stress.
R1
1
Rxx Ryy Rxx Ryy 2 4R2xy
2
R2
1
Rxx Ryy Rxx Ryy 2 4R2xy
2
R3 Rzz
where
Rxx
2
1
1
1
2
1
C 11 C 12 C 13 rxx C 11 C 12 C 13 ryy
3
3
3
3
3
3
Ryy
2
1
1
1
2
1
C 12 C 22 C 23 rxx C 12 C 22 C 23 ryy
3
3
3
3
3
3
Rzz
2
1
1
1
2
1
C 13 C 23 C 33 rxx C 13 C 23 C 33 ryy
3
3
3
3
3
3
=ra
F j/1 c /2 dja j/3 c 2/4 dja /z0 j2/5 c /6 dja 2r
G qx
@/
@/
qy
0
@sx
@sy
1=a
Rxy C 66 rxy
Let rT0 and rC0 is yield stress in tension and compression respectively along the 0 orientation. The proposed orthotropic yield criterion is written as
rT0
R Cs
2
C 11
6C
6 12
6
6 C 13
C6
6
6
6
4
C 12
C 22
C 13
C 23
C 23
C 33
rC0
C 44
C 55
F
a
a
a
j/1 j k/1 j/2 j k/2 j/3 j k/3
F
a
a
a
j/1 j k/1 j/2 j k/2 j/3 j k/3
1a
1a
where
/1
2
1
1
C 11 C 12 C 13
3
3
3
/2
2
1
1
C 12 C 22 C 23
3
3
3
/3
2
1
1
C 13 C 23 C 33
3
3
3
3
7
7
7
7
7
7
7
7
5
rT90
C 66
where C is a fourth order transformation tensor with reference to
orthotropic (x, y, z) axes and F is the size of yield locus, k is based
on yield stress in tension/compression and material parameter a
in the yield function.
Table 4
Parameter description for Barlat 2000.
Experimental test
qx
qy
0 Tension
90 Tension
Balanced biaxial tension
2/3
1/3
1/3
1/3
2/3
1/3
1 r0
2 + r90
1 + 2rb
2 + r0
1 r90
2 + rb
rC90
1a
1a
F
a
a
a
jW1 j kw1 jw2 j kw2 jw3 j kw3
F
a
a
a
jW1 j kw1 jw2 j kw2 jw3 j kw3
where
1
2
1
w1 C 11 C 12 C 13
3
3
3
1
2
1
w2 C 12 C 22 C 23
3
3
3
341
a1
a2
a3
a4
a5
a6
a7
a8
400
0.982
0.853
0.787
0.976
0.986
0.787
1.132
1.426
w3
rTb
1
2
1
C 13 C 23 C 33
3
3
3
F
a
a
a
jX1 j kX1 jX2 j kX2 jX3 j kX3
1a
where
X1
1
1
2
C 11 C 12 C 13
3
3
3
X2
1
1
2
C 12 C 22 C 23
3
3
3
X3
1
1
2
C 13 C 23 C 33
3
3
3
rT0
a1
a1
1 k Ua1
1 W1 1 k U2 W2 U3 W3
a
a1
a1
a
a
1 k Ua1
1 W1 U1 1 k U2 W2 U3 W3 U2 U3
r T90
a1
a1
1 k Wa1
2 U2 1 k W1 U1 W3 U3
a
1 k W
a1
2
a1
a
a
W2 U2 1 k Wa1
1 U1 W3 U3 W1 W3
Table 6
Cazacu-Balart yield model material parameters for Ti6Al4V alloy.
Temperature (C)
C11
C22
C33
C12
C13
C23
C44
400
1.959
1.837
1.955
0.684
0.480
0.451
1.880
0.085
342
Fig. 5. Simulated deep drawn cups (a) Hill 1948, (b) Barlat 1989, (c) Barlat 1996, (d) Barlat 2000 and (e) Cazcau Barlat.
343
Parameters
Tests required
Hill 1948
Barlat 1989
Barlat 1996
Barlat 2000
Cazacu Barlat
4
3
8
8
8
Y
Y
Y
Y
Y
r0
r45
Y
Y
Y
r90
Y
Y
Y
rb
r0
r45
r90
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
Y
rb
rc
Y
Y
Y
Y = test required.
teria [16]. But, these models are not able to show yielding behavior
in uniaxial compression and biaxial tensile stress state. Inspite of
this drawback, these criteria are frequently used in nite element
simulations due to ease in identication of material parameters
[19].
On the other hand, loci obtained by Barlat 1996 and Barlat 2000
yield function shown in Fig. 2(c) and (d) nearly approximate the
experimental data points for Ti6Al4V alloy. But, accurate determination of material response in sheet metal forming requires
more accurate yield function which takes tensioncompression
asymmetry into account. However, Ti6Al4V alloy has HCP crystal structure and Bauschinger effect is more pronounced. Since
Cazacu Barlat yield function considers stress asymmetry, it is very
well suited for Ti6Al4V alloy. It is also evident from Fig. 2(e) that
the yield locus plotted by Cazacu Barlat yield function predicts
yielding very accurately with all the states of stress. Similar results
were observed by Cazacu and Barlat [9] for Mg0.5% Th alloy
which is also a HCP crystal structure.
In order to validate the yield criteria, the experiments have
been performed at 400 C. The maximum blank diameter of
54 mm was successfully drawn hence the limiting draw ratio
(LDR) of 1.8 was achieved. It is seen that LDR of Ti6Al4V alloy
is signicantly lower than other structural alloys such as austenitic stainless steel. The experimental deep drawn cup at 400 C
and simulated cups using different anisotropic yield criteria are
as shown in Figs. 4 and 5 respectively. Thickness distribution
Table 10
Total CPU time taken for simulation.
Yield model
Time (s)
Hill 1948
Barlat 1989
Barlat 1996
Barlat 2000
Cazacu Barlat
7166
8769
82,722
76,014
37,398
Table 7
Relative error for thickness distribution.
Normalized
distance
Expt.
thickness
Hill
1948
Relative
error
Barlat
1989
Relative
error
Barlat
1996
Relative
error
Barlat
2000
Relative
error
Cazacu
Barlat
Relative
error
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0.803
0.789
0.788
0.772
0.792
0.837
0.834
0.884
0.903
0.908
0.914
0.807
0.795
0.808
0.808
0.801
0.792
0.897
0.943
0.922
0.921
0.932
0.006
0.008
0.026
0.047
0.012
0.054
0.076
0.067
0.021
0.014
0.019
0.816
0.797
0.817
0.777
0.761
0.763
0.799
0.812
0.844
0.896
0.923
0.017
0.010
0.036
0.007
0.039
0.089
0.042
0.082
0.065
0.014
0.010
0.829
0.828
0.829
0.796
0.784
0.783
0.799
0.836
0.875
0.890
0.895
0.032
0.049
0.053
0.031
0.010
0.065
0.042
0.054
0.031
0.020
0.020
0.799
0.800
0.809
0.803
0.790
0.785
0.818
0.849
0.900
0.956
0.980
0.005
0.014
0.026
0.040
0.003
0.062
0.019
0.040
0.003
0.053
0.072
0.803
0.803
0.806
0.781
0.781
0.822
0.835
0.856
0.882
0.904
0.947
0.000
0.018
0.023
0.012
0.014
0.018
0.001
0.032
0.024
0.005
0.036
Table 8
Comparison of percentage relative error in thickness distribution.
Material model
Hill 1948
Barlat 1989
Barlat 1996
Barlat 2000
Cazacu -Barlat
3.1
0.0236
3.7
0.0283
3.6
0.0161
3.0
0.0232
1.6
0.0112
344