Documente Academic
Documente Profesional
Documente Cultură
DOI 10.1007/s00170-015-8261-1
ORIGINAL ARTICLE
Abstract Nickel-aluminium bronze is used widely in seawater environment. The mechanistic method and analytical
method to predict helical end milling force are briefly described and compared. The mechanistic approach is shown
to depend on milling force coefficients determined from milling tests. By contrast, the analytical method is based on a
predictive machining theory, which regards the workpiece
material properties, tool geometry, cutting conditions and
types of milling as the input data. Each cutting edge of the
helical end cutter is discretized into a series of infinitesimal
elements along the cutter axis and the cutting action of which
is equivalent to the classical oblique cutting process. Thus, the
cutting force components applied on each element can be
calculated using a predictive oblique cutting model and the
total instantaneous cutting forces are obtained by summing
up the forces contributed by all cutting edges. The equation
of equivalent plane angle is derivation through coordinate
transformation. Experiments on machining nickel-aluminium
bronze under different cutting conditions were conducted to
validate the proposed model.
* Wenyu Yang
mewyang@mail.hust.edu.cn
1 Introduction
Nickel-aluminium bronze (NAB) is a copper-based alloy
which is widely used for propulsion and seawater handling
systems in naval platforms, such as marine propellers, bearings [1]. However, the study of machinability of NAB in the
published literature is little. Maybe, the researchers consider
the machinability of NAB is well, compared with difficult-tomachine alloys.
The helical end milling is used extensively in manufacturing of surfaces, such as propellers, turbines, and dies/molds.
Modelling of cutting forces is the basis for prediction of machine tool chatter, tool wear and breakage, cutting parameters
optimization and surface quality [2]. There mainly exist three
methods to model milling forces: mechanistic, numerical and
analytical methods. The mechanistic models are commonly
available to predict cutting forces especially for new materials
and tools. Typical approach for numerical modelling is finite
element method (FEM). Currently available FE models can be
used to predict cutting forces, stress and temperature distributions. Simulations of FEM are time consuming and the physical meaning is not clear. Due to the fact that analytical
methods are more physics based rather than experimental
modelling, it can reflect the actual physical phenomenon better than the other methods. Arrazola et al. [3] summarized the
capabilities and limitations of modelling approaches.
Two types of mechanistic cutting force models are found in
the machining literature. In the first model, the effects of
shearing mechanism and effects of rubbing and ploughing
mechanisms are lumped into one specific force coefficient
for each cutting force component. In the second model, the
shearing and ploughing effects are characterized separately by
the respective specific cutting and edge force coefficients.
Budak et al. [4] described and compared the mechanistic and
unified mechanics of cutting approaches to the prediction of
2 2z
tani0
N D
Vc
Rake
face
Milling cutter
Cutting edge
Chip
i0
Cutting
edge
t1
dz
ap
Workpiec
e
Workpiece
E-E Section
n
j 1
nr
j 1
j
(z)
zI dFr, j
j
r
Workpiece
zI
yI dFa, j dr
xI dFt , j
yI
dFt , j
Feed
xI
dFr, j dFa, j
Ps
3
32
d F t; j j ; z
0
0 54 d F r; j j ; z 5
1
d F a; j j ; z
7
The average cutting force calculated as follows:
3
2 32
Fx
Kt
cos2
2sin2
N
a
f
6 7
p t 4
cos2
K r 54 2sin2
4 Fy 5
8
0
0
K
a
Fz
2
3ex
0
0 5
4cos
st
8
Given the cutter geometry and immersion conditions, only
the specific cutting force coefficient remain unknown in the
right-hand side of Eq. (8). The coefficients can therefore be
determined by equating the measured cutting forces with the
corresponding expressions. The details can be found in Wan
et al. [6].
8 9 0
1
< xe =
ye @ 0
: ;
ze
0
0
cose
sine
18 9
0
< xs =
sine A ys
: ;
cose
zs
z fl A
Vc
zn
Pr
xn
Vc
yn
Pe
yI
zI
xs
P
Psh
sh
x
xI
Vs
Vn
chip
t2
dFt , j
w
dFr , j dFa , j
Ps
Ps
yn
yI
zI = zn
xI
Pn
zn n
xn
yc = y n
zc c z
fl
zc
xn
xc
yc
zc = z fl
h E
A
Vc
tool
ze
xe
t1
D B (o)
workpieces
ze sine ys cose zs
10
(xs, ys, zs) is obtained by the rotation of angle s of the coordinate system (x, y, z) around the z axis.
8 9 0
18 9
coss sins 0 < x =
< xs =
11
@ sins coss 0 A y
y
: ;
: s;
z
0
0
1
zs
(xc, yc, zc) is obtained by the rotation of angle (n n) of the
coordinate system (x, y, z) around the y axis.
8 9 0
18 9
cosn n 0 sinn n < x =
< xc =
A y
@
12
y
0
1
0
: c;
: ;
zc
sinn n 0 cosn n
z
(xfl, yfl, zfl) is obtained by the rotation of angle c of the coordinate system (xc, yc, zc) around the xc axis.
8 9 0
18 9
1
0
0
< xfl =
< xc =
yfl @ 0 cosc sinc A yc
13
: ;
: ;
zfl
0 sinc cosc
zc
From Eqs. (11), (12) and (13), zfl is calculated as:
zfl cosc sinn n coss sinc sins xs
cosc sinn n sins sinc coss ys
cosc cosn n zs
14
Since ze is parallel to zfl, from Eqs. (10) and (14), equivalent plane angle e can be obtained as:
cosc sinn n tans sinc coss
cosc cosn n
sinn n sins tanc coss
cosn n
16
The chip flow angle c on the rake face A is given with the
assumption that the friction force and the chip flow direction
are collinear and can be calculated from the following equation [10]:
cosn n sinn sinc tans cos2 n n cosc
cosn sinn n sinn tansinc cosc
tantans sinn n cosn n cos2 c 0
17
:
:
18
>
: m q hze q
ze 0; h
h
tane
15
8
>
>
>
<
:
m cose
ze 1hq1
ze 1 h; 0
V coss sinn q 11 hq
ze
:
:
>
>
m cose
cose hm
q1
>
:
ze 0; h
q hze
V coss sinn q 1h
V coss sinn q 1
q 1V s q 1V coss cosn
:
m
h
hcoss cosn n
20
Vc
V coss sinn
V coss cosn
; Vs
cosc cosn n
coss cosn n
22
: :
where ; m ; and represent the shear strain rate, the maximum shear strain rate and the shear strain. In this paper,
h = 0.025 mm, q = 3.
19
:
T T r m
1 Cln
s p A B p
1
T m T r
3
3
:0
23
:
where s ; ; and T represent the shear stress, the shear strain,
the shear strain rate and the absolute temperature, respectively.
The material characteristics are defined by the parameters A,
B, C, n, m, which stand for the yield strength, the strength coefficient, the strain rate constant, the strain hardening exponent and the thermal softening coefficient, respectively.
In cutting process, the boundary of the primary shear zone
can be considered adiabatic; the thermal conductivity is then
negligible. Assuming that a fraction (Taylor-Quinney
coefficient) of the plastic work is converted into heat, since
the temperature depends only on the coordinate ze under the
steady flow condition, consequently, the heat transfer equation
becomes:
dT
cose
:
dze cV sinn coss
dN s
28
Therefore, for the infinitesimal element of the jth cutting edge, the three differential cutting force components
dF t,j , dF r,j , dF a,j (tangential, radial, axial), applied to
point P in Fig. 2a, are evaluated from the following
matrix form:
9 2
8
coss cosn cossh sins sinsh
< d F t; j =
d F r; j 4
sinn cossh
;
:
d F a; j
sins cosn cossh coss sinsh
3
coss sinn
d Fs
cosn 5
dN s
sins sinn
24
29
n
2
4 2
25
As suggest by Dudzinski et al. [22], the mean friction coefficient may be a power function of the chip velocity:
f f 0V c
26
s t 1
dz
coss sinn
27
3 2
d F x; j j ; z
cos j z
4 d F y; j j ; z 5 4 sin j z
0
d F z; j j ; z
2
3
32
d F t; j j ; z
sin j z 0
cos j z 0 54 d F r; j j ; z 5
0
1
d F a; j j ; z
30
By integrating Eq. (30), respectively, the total cutting force
in each direction can be obtained. The algorithm for predicting
the cutting forces in end milling is shown in Fig. 3.
Fm
N Z
X
j1
z2
d F m; j dz;
m x; y; z
31
z1
where z1, z2 are the lower and upper axial engagement limits of
the in-cut portion.
5 Model validation
5.1 Experiment details
Table 1
Researchers
Relationship
= /4 +
= /4 + /2 /2
= /4 + + (s/s 1)/2
= + (CM arctan 2)/2 + arctan
= + arctan(s/s)
= + 54.7
cos1 s cos 4 2 sin 4 2
= /4 f + 5/8
Input parameters
Work piece material:A B C n m
Helical end geometry:D Ni0
Cuttingconditions: nr , ap , ae , f
Determine number of angular integration steps
L,number of axial integration steps K,entry
angle exit angle
s=s+1
Calculate the immersion angle of the sth discrete
point (height z)on the jth edge from Eq.(1)
Is the cutting
region or not?
j=j+1
K>s
N>j
Y
Sum the force of aallengagededges
i=i+1
L>i
Y
Output the cutting force variation with tool
rotation angle
Table 2
Test no
Spindle speed
(rpm)
Radial depth
of cut (mm)
1
2
3
4
5
6
7
1000
1000
1000
1600
1600
1600
2000
0.04
0.05
0.06
0.04
0.05
0.06
0.04
6
6
6
6
6
6
6
8
9
10
11
12
13
2000
2000
2000
2000
1800
1600
0.05
0.06
0.04
0.05
0.05
0.03
6
6
4
4
6
6
dynamometer or the tiny foundation vibrations of nearby machines in the experimental environment, slight perturbations
of measured cutting forces exist when the magnitudes of cutting force components attain a relatively low level. Even in
this condition, the trends of simulation and experimental data
are in good agreement. In Fig. 5a, since the experimental data
has run-out, the magnitude of cutting force did not match well,
but the shape of force profile during the cutter revolution is
captured well.
Table 3
Test no. 1
Kt
Kr
Ka
Fig. 5 Comparisons of
mechanistic model and
experimental results
250
Mechanistic
Mechanistic
Experiment
200
Cutting Forces(N)
Cutting Forces(N)
Fy
150
100
50
0
Fz
-50
-100
45
90
100
50
0
Fz
Fx
-50
Fx
0
Fy
150
135
180
225
270
315
-100
360
45
90
135
225
270
315
360
(b) Test 8
250
250
Mechanistic
Mechanistic
Experiment
Experiment
200
200
Fy
150
Cutting Forces(N)
Cutting Forces(N)
180
(a)Test 2
100
Fx
50
0
Fz
Fy
150
100
50
0
Fz
Fx
-50
-50
-100
Experiment
200
-100
0
45
90
135
180
225
270
315
360
45
90
135
180
225
270
315
360
Figure 7a, b is a comparison of mechanistic and analytical model result in cutting conditions for test 2 and test
8. It is shown that the proposed analytical method
closely agrees with mechanistic method results. The
mechanistic method requires a set of cutting tests for
each milling cutter conditions, whereas the analytical
method requires input data and using oblique cutting
model extend to end milling. The mechanistic model
is used widely on new materials or the materials properties not given. This method can be used in industry
since it calibrates cutting force coefficients efficiently.
The analytic model has a clear physical meaning, but
250
200
Cutting Forces (N)
200
Fy
150
100
50
0
Fy
150
100
50
0
Fz
Fz
-50
Fx
-50
Fx
-100
-100
0
45
90
135
180
225
270
315
360
45
90
315
360
Analytical
Experiment
100
50
50
-50
-50
-100
0
45
90
135
180
225
270
315
360
45
90
135
180
225
270
315
360
250
250
Analytical
Experiment
Analytical
200
Experiment
200
Fy
150
270
150
100
150
100
50
0
50
45
90
135
Fx
Fx
0
Fy
100
Fz
-50
Fz
-50
180
225
270
315
360
1
- 00
45
250
90
135 180 225 270
Tool rotation angle (deg)
(f) Test 11
315
360
250
Analytical
Analytical
Experiment
200
200
150
150
225
200
150
50
0
50
0
Fz
Fx
-50
-50
45
90
135
180
225
270
315
360
Experiment
Fy
100
100
-100
180
250
Analytical
Experiment
200
-100
135
250
-100
Experiment
Analytical
Experiment
Analytical
-100
45
90
135
180
225
270
315
360
250
Prediction(Analytical)
Experiment
Prediction(Mechanistic)
200
Fy
Cutting Forces (N)
Fy
150
100
50
0
Fz
Fx
-50
150
100
50
0
Fz
Fx
-50
-100
Prediction(Analytical)
Experiment
Prediction(Mechanistic)
200
-100
0
45
90
135
180
225
270
315
360
1.
2.
4.
However, the effect of tool run-out and deflection in milling process need to be considered in future work for more
accurate prediction of milling forces.
5.
6.
7.
8.
9.
10.
11.
12.
13.
135
180
225
270
315
360
References
6 Conclusion
90
(b) Test 8
3.
45
(a) Test 2
14.
19.