Академический Документы
Профессиональный Документы
Культура Документы
a r t i c l e i n f o
abstract
Article history:
Received 13 June 2011
Received in revised form
19 December 2011
Accepted 22 December 2011
Available online 14 January 2012
Friction under lubricated conditions is known to depend on pressure, and experimental determination
of this dependence has typically quantied pressure as a uniform value over the contact area. However,
non-uniform pressure distributions at the contact interfaces of draw-bend tests have been reported
from various experiments and simulations. A previous study by the authors had evaluated the
conventional methodology, which assumes uniform pressure distribution to estimate friction coefcients from draw-bend friction tests, and has concluded that the conventional methodology is only
valid for measuring an average friction coefcient over the pressure range, which exists in a specic
draw-bend system. In this paper, a new methodology to determine friction coefcients from draw-bend
friction tests considering the pressure non-uniformity is suggested. In the methodology, contact
pressure maps obtained from simulations, instead of the uniform pressure assumption, are included in
the analysis of test data to measure the pressure dependency of friction coefcient. The proposed
method is applied to friction measurement of aluminum sheets with anisotropic mill-nish surface and
friction coefcients were obtained as functions of contact pressure, sliding velocity and sliding
direction. Lastly, the obtained friction data were implemented into a nite element code, and circular
cup drawing experiments and simulations were performed to validate the methodology.
& 2011 Elsevier Ltd. All rights reserved.
Keywords:
Friction measurement
Friction modeling
Friction anisotropy
Contact pressure
Finite element modeling
Cup drawing
1. Introduction
The draw-bend friction test closely mimics the motion and
deformation of a sheet metal drawn over a die radius. The friction
coefcient is measured between a metal strip and a cylindrical
pin while the strip slides, and bends and unbends around the pin.
A conventional method to determine a friction coefcient from
test data is based on the assumptions that (i) pin/strip contact
angle is equal to the geometric wrap angle, and (ii) pressure
distribution at the contact is uniform. However, these assumptions are not true due to the stiffness of metal strips. Alinger et al.
[1] measured the strains on the outer surfaces of the strips during
the draw-bend test and noticed that the real contact angle is less
than a geometric wrap angle from the longitudinal strain path.
Coubrough et al. [2] observed the existence of pressure peaks at
the pin/strip contact using a contact sensor on the pin. They also
observed that the real contact angle is less than the geometric
wrap angle. The study by Kim et al. [3] using 3D nite element
simulations of the tests clearly revealed and quantied the
phenomena concerning wrap angle and pressure variation.
Pereira et al. [4,5] investigated the time-dependent nature of
contact pressure response over the die radius in a channel
Corresponding author. Tel.: 905 525 9140x27841; fax: 905 525 9140.
E-mail address: jainmk@mcmaster.ca (M.K. Jain).
0890-6955/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmachtools.2011.12.011
70
Y.S. Kim et al. / International Journal of Machine Tools & Manufacture 56 (2012) 6978
Strip
P
Grip &
Load Cell
FX (Inlet Strip Tension)
Pin
Grip &
Load Cell
(Outlet Strip Tension) FY
Fig. 1. Draw-bend friction test, (a) schematic layout and (b) experimental set-up.
Step 1)
12.7 mm pin
25.4 mm pin
Step 2)
25 mm
42 mm pin
Step 3)
45 mm
Inlet
Outlet
71 mm
F X F Y
2rw
Y.S. Kim et al. / International Journal of Machine Tools & Manufacture 56 (2012) 6978
71
Fig. 4. Contact pressure (MPa) distribution on 901 wrap area of strip (12.7 mm pin case without friction) shown as (a) an overlay on the deformed FE mesh, and (b) as a 3D
pressure (in MPa)distancewrap angle plot.
Fig. 5. 3D pressure maps (in MPa) from simulations for the 3 pin diameters, (a) frictionless case and (b) Stribeck friction model.
Table 1
Input parameters for Stribeck friction model.
Values
mBL
mEHL
0.131
0.001
4.5 10 4
5.4 10 3
1.85 mm
1.2 Pa s
LBL
LEHL
Ra
Friction coefficient
Parameters
0.15
0.1
L=
0.05
0
Log (L)
Fig. 6. Input Stribeck friction model.
PRa
72
Y.S. Kim et al. / International Journal of Machine Tools & Manufacture 56 (2012) 6978
In the other two cases (25.4 mm and 42 mm pins), the length of the
nely meshed region was increased to 45 mm and 71 mm, respectively, while keeping mesh sizes equal to the 12.7 mm pin case.
Stressstrain data were selected from experimental results
available at hand for aluminum sheets and an isotropic elastic
plastic material model was used. All simulations were performed
with the nite element software called H3DMAP [15], which uses
an explicit solution method that is well suited to large deformation
problems. Simulations closely followed the general procedure of
0.14
0.1
Friction coefficient
Friction coefficient
0.15
0.05
0.09
Input Stribeck curve
Predicted friction curve
0.04
Conventional method
0
20
40
Contact pressure (MPa)
60
-0.01
0
20
30
40
Contact pressure (MPa)
50
60
0.2
0.1
0.075
0.05
Friction Coefficient
Friction coefficient
0.15
10
0.045
0.025
0.15
0.1
v = 6.35 mm/sec
v = 25.4 mm/sec
v = 101.6 mm/sec
0.05
0
12.7 mm
25.4 mm
Pin diameter
42 mm
0
10
Table 2
Tension forces from frictionless simulations.
F 0X
756.2
FB
25.4 mm pin
F 0Y
917.8
116.6
F 0X
756.2
FB
42 mm pin
F 0Y
822.1
65.9
F 0X
756.2
FB
F 0Y
788.7
32.5
Table 3
Tension forces from simulations with Stribeck friction.
12.7 mm pin
FX
756.2
FT
c1
25.4 mm pin
FY
1004.6
86.8
1.052
FX
756.2
FT
c2
30
0.2
42 mm pin
FY
875.0
52.9
1.034
FX
756.2
FT
c3
Friction Coefficient
12.7 mm pin
20
Pin Diameter (mm)
0.15
v = 6.35 mm/sec
v = 25.4 mm/sec
v = 101.6 mm/sec
0.1
0.05
0
FY
818.9
30.2
1.020
10
20
Pin Diameter (mm)
30
Y.S. Kim et al. / International Journal of Machine Tools & Manufacture 56 (2012) 6978
into the force required to drive the motion. But for step 2, dynamic
relaxation method [16] has been applied so that any dynamic
responses aroused by motions in steps 1 and 2 can be damped out
73
Fig. 11. Pressure maps obtained from simulations (a) 12.7 mm and (b) 25.4 mm
diameter pins.
v = 6.35 mm/s
v = 25.4 mm/s
v = 101.6 mm/s
0.3
v = 6.35 mm/s
v = 25.4 mm/s
v = 101.6 mm/s
0.3
0.2
0.1
0.2
0.1
0
0
20
40
P (MPa)
60
80
20
40
P (MPa)
60
80
Fig. 12. Friction coefcients determined by proposed method (a) RD strips (b) TD strips.
6
5
2
v (mm/s)
1
P (MPa)
v (mm/s)
P (MPa)
74
Y.S. Kim et al. / International Journal of Machine Tools & Manufacture 56 (2012) 6978
where mBL, mEHL, LBL, and LEHL are the parameters decided by
experiments. The lubricant number L is again a function given as
nZ
crit
FTy = y FN
PRa
y, TD
x, RD
crit
FTx = x FN
FTy
Stick region
FTx
Fig. 14. Stick/slip criteria for anisotropic friction model.
The proposed method was applied to the friction measurement of mill-nish AA5754-O sheets. The tests were conducted
0.35
0.35
0.3
p = 70 MPa
0.25
0.3
0.2
p = 40 MPa
0.15
0.25
Retrieved Friction
p = 10 MPa
p = 70 MPa
0.15
0.1
0.1
0.05
0.05
Retrieved Friction
0.2
p = 40 MPa
p = 10 MPa
0
0
20
40
60
80 100
Sliding velocity (mm/s)
120
20
40
60
80 100
Sliding velocoty (mm/s)
Fig. 16. Input friction model data vs. retrieved friction coefcient (a) RD sliding (b) TD sliding.
120
Y.S. Kim et al. / International Journal of Machine Tools & Manufacture 56 (2012) 6978
using the draw-bend test setup and two existing pins of 12.7 mm
and 25.4 mm diameters. Mill-nish aluminum sheets have
strong directional anisotropy in surface roughness. Therefore, the
friction measurements were performed on strips cut along the
rolling direction (RD) and transverse direction (TD) of the sheets.
The tests were also performed at three sliding speeds (6.35 mm/s,
25.4 mm/s and 101.6 mm/s) to measure the sliding speed dependency of the friction coefcient. Sheet forming lubricant, Parker
MP404, was applied to the test strips by dipping the specimens
into lubricant solution diluted with Hexane. By this way, 1.7
2.3 g/m2 of lubrications were applied to the strips. Fig. 10 shows
the friction coefcients estimated by applying the conventional
method to the test data. Obtained friction coefcients show the
same trend as other researchers observed in their tests that the
friction coefcient decreases as the pin size increases.
To obtain pressure maps at current test conditions for the
proposed method, simulations similar to the ones in Section 2
were performed for two frictionless pin (12.7 mm and 25.4 mm)
cases. Due to the high gradient of contact pressure near the strip
edge as observed in the previous simulations (see Figs. 5 and 6),
mesh densities were doubled in width direction of the strip
model. Fig. 11 shows the obtained pressure maps for 12.7 mm
and 25.4 mm diameter pins.
Fig. 12 shows the friction coefcients determined by applying the
proposed method to the test data. Because the tests were performed
using two different size pins, friction coefcients were obtained as
linear functions of pressure. Friction coefcient is higher in rolling
75
direction than transverse direction, which agrees with the test results
by other researchers for mill-nish aluminum sheets [1819]. Also, it
can be observed that at higher sliding speeds the friction coefcients
become more pressure independent and the difference between
RD and TD is reduced. This is because the contact becomes more like
hydrodynamic lubrication condition at higher sliding speed where
the load is carried more by lubricant lms [20].
0.02
0.01
0.05
Input Friction Model
0.04
Retrieved Friction
0.03
0
0
0.01
0.02
0.03
0.04
0.05
Fig. 17. Input friction model data vs. retrieved friction coefcient for sliding cases
in various directions (with p 110 MPa and v 50 mm/s).
57.8
Upper Die
Sheet
R6.35
Lower
R6.35
Punch
48
Fig. 18. Experimental cup drawing, (a) test set-up, (b) punch and die geometric details.
76
Y.S. Kim et al. / International Journal of Machine Tools & Manufacture 56 (2012) 6978
In the cup drawing tests two sheet samples were drawn until
the failure of the cup was observed. Fig. 22 shows the failed
samples, where the RD of the sheets is marked with an arrow. It
shows that both sheets failed in RD. This type of directional
failure is typically attributed to the material anisotropy of the
sheets in the literature. However, the strong frictional anisotropy
35
77 mm/s
30
35
25
20
15
10
5
0
5
10
15
Punch Displacement (mm)
77 mm/s
25
20
15
10
5
0
20
30
5
10
15
Punch Displacement (mm)
20
Fig. 20. Finite Punch force/displacement curves (a) experiments (b) simulations.
112
110
RD
TD
108
112
106
104
102
100
0
20
40
60
Punch Speed (mm/s)
80
110
RD
TD
108
106
104
102
100
0
20
40
60
Punch Speed (mm/s)
Fig. 21. Cup size as a function of punch speed from (a) experiments (b) simulations.
80
Y.S. Kim et al. / International Journal of Machine Tools & Manufacture 56 (2012) 6978
8. Conclusions
1. A new methodology to determine pressure dependent friction
coefcients from draw-bend friction tests is developed. The
contact pressure maps in the whole pin/strip interface are
obtained from draw-bend simulations and included in the
analysis of test data.
77
2. The proposed friction measurement method and its implementation into a FE code considers local contact conditions
such as the sliding speed, contact pressure, sliding direction
and anisotropic stick/slip criterion in the form of an ellipse at
the macroscopic level.
3. The anisotropic friction model obtained by the new methodology
and its implementation into FE code is validated through circular
cup drawing experiments and simulations. The comparisons
between experiments and simulations show good agreements
in the aspects of punch force, cup size and failure location.
4. While the material anisotropy is not considered in the constitutive model, the phenomenon of failure in the rolling
direction of the sheet is accurately predicted. The results
indicate that the anisotropy of the material ow behavior in
the cup ange leading to variation in ange width and material
direction dependent fracture can arise from frictional anisotropy and not only from the material crystalline anisotropy.
5. The proposed methodology of draw-bend friction tests and
analysis of data is quite general and can be extended to obtain
friction coefcient representation in other forms such as
higher order polynomial functions of pressure.
Acknowledgments
The authors are grateful to Mr. L. D. Kenny of Novelis Global
Technology Center for his technical advice and for performing the draw-bend tests. The authors also acknowledge the
help of M. Bruhis at McMaster University with cup drawing
experiments.
A1
A2
FX
[b] Friction forces FT are obtained from
F T F Y F X F B
FY
Fig. A1. A schematic of the forces being applied to the sheet in a draw bend process.
A3
78
Y.S. Kim et al. / International Journal of Machine Tools & Manufacture 56 (2012) 6978
[4]
[5]
A5
[6]
n
X
F T test j F T step3 j 2
[7]
A6
[8]
mP a0 a1 P a2 P2
[10]
A7
[11]
A8
[15]
[12]
[13]
[14]
Substituting a0, a1 and a2 obtained from Eq. (A8) into Eq. (A7),
a optimum friction coefcient model expressed as a second order
polynomial of pressure can be determined.
[16]
[17]
References
[18]
[1] M.J. Alinger, S. Sriram, D.K. Matlock, An investigation of strain evolution in
thebending-under-tension friction test, Minerals, Metals and Materials
Society/AIME, Sheet MetalForming Technology, USA, 1999., pp. 129-140.
[2] G.J. Coubrough, M.J. Alinger, C.J. Van Tyne, Angle of contact between sheet anddie
during stretch-bend deformation as determined on the bending-under-tension
friction testsystem, Journal of Materials Processing Technology (2002), pp. 69-75.
[3] Y.S. Kim, M.K. Jain, D.R. Metzger, A nite element study of capstan friction
Test, Proceedings of the 8th International Conference on Numerical Methods
[19]
[20]