Professional Documents
Culture Documents
R
T
0
x t
2
dt
T
v
u
u
u
t
1
The analysis of variance (ANOVA) proved that both the
diamond mill wear and the two considered process
parameters, feed speed and depth of cut, significantly
influence the RMS parameter characterising the F
x
, F
y
and
F
z
signals. The interaction between feed speed and disk
wear seems to significantly influence RMS values, too.
Figure 8 shows the influence of these three factors on RMS
values extracted from the F
z
signal. The same effects have
been observed for F
x
and F
y
signals. Figure 9 shows the
effects of the interaction between feed speed and disk wear
on RMS values of the F
z
signal.
Plots of the RMS feature vs. diamond disk wear have
been produced for the three components of the cutting force
(see Fig. 10). The x-axis shows the five stages of diamond
mill wear distributed along disk tool life. The five stages of
wear have been put into relationship with the protrusion
height of the diamond grits by bond observed on disk
surface, as shown in Fig. 11.
Vt
Va
Fy, Ay
Fz, Az
Fx, Ax
Fy, Ay
Stone
diamond
disk
Stone
diamond disk
Cutting
groove
Fig. 4 Experimental scheme
Table 2 Experimental plan
Factors # Levels Levels
Cutting depth (mm) 3 1.01.5
Feed speed (mm/min) 3 200400
Spindle speed (rpm) 1 3,000
Replications 3
458 Int J Adv Manuf Technol (2007) 35:454467
Inspection of Fig. 10 shows that the RMS of the x and y
components of the cutting force are very low and that their
values stay low with the increase of tool wear (from 5 to
20 N). The values of RMS related to the z component
increase considerably with the increase of tool wear, from
about 35 to 110 N. This is due to the fact that disk cutting
proceeds along a line parallel to the y-axis; the contact
between disk and stone takes place in the y-z plane (see
Fig. 1). Moreover, the ratio between the depth of cut and
the disk diameter is very low, thus involving a low value of
force along the y-axis. The RMS feature along the x- and y-
axes may be neglected with regards to the z-axis.
Figures 12 and 13 show the changes in the relationship
between RMS feature related to F
z
signal and disk wear
with the increase of the feed speed (u) or of the depth of cut
(d). RMS increases significantly with the disk wear, with
the feed speed, and with the depth of cut.
Regression analysis underlined that the analytical rela-
tionships among RMS feature, mill wear T, feed speed u,
and depth of cut d is given by the following expression for
the F
z
signal:
log RMS
F
z
k
1
k
2
d k
3
u k
4
T k
5
u T 2
where k
1
, k
2
, k
3
, k
4
and k
5
are constants equal to 1.04,
0.619, 0.00478, 0.579 and 0.000938, respectively. The
adjusted coefficient of determination is about 96.5%, while
the hypotheses (normality and homogeneity of variance)
about the residuals are satisfied. The predictive capability
of the regression model is given by the value of the
predictive error sum of squares (PRESS) statistic, 96.2%,
respectively. The lack of fit test confirms that Eq. (2) may
be considered linear. Equation (2) may be written as:
Fig. 6 Time domain force signal monitored in Y direction: a at 1
wear stage; b at 3 wear stage; c at 5 wear stage
0
10
20
30
40
50
60
70
80
90
100
0 200 400 600 800
Time [ms]
Time [ms]
Time [ms]
F
x
[
N
]
F
x
[
N
]
F
x
[
N
]
(a)
0
10
20
30
40
50
60
70
80
90
100
0 200 400 600 800
(b)
0
10
20
30
40
50
60
70
80
90
100
0 200 400 600 800
(c)
Fig. 5 Time domain force signal monitored in the X direction: a at 1
wear stage; b at 3 wear stage; c at 5 wear stage
Int J Adv Manuf Technol (2007) 35:454467 459
RMS
F
z
e
k
1
e
k
2
d
e
k
3
u
e
k
4
T
e
k
5
uT
3
3.3 Acceleration signal analysis in the time domain
Figures 14ac, 15ac and 16ac show typical acceleration
signals in the time domain that are monitored in the X
(traverse to feed), Y(feed) and Z(vertical), directions,
respectively, at different time interval inside tool life.
As can be seen in these figures, the average values keep
constant to 0 with the progress of disk wear, while the
dispersion of the signals increases significantly.
RMS feature has been extracted for the three accelera-
tion signals, A
x
, A
y
, and A
z
, respectively.
The analysis of variance (ANOVA) proved that both the
diamond disk wear and the two considered process
parameters, feed speed and depth of cut, influence
significantly RMS parameter characterising A
x
, A
y
, and
A
z
signals. The interactions between feed speed and disk
wear seem to significantly influence RMS values, too.
Figure 17 shows the influence of these three factors on
RMS values extracted from the A
z
signal. The same effects
200
400
1 2 3 4 5
20
30
40
50
60
70
80
90
100
110
T [#]
u [mm/min]
a
v
e
r
a
g
e
o
f
R
M
S
o
f
F
z
[
N
]
Fig. 9 Interaction plot of RMS feature extracted by F
z
signal vs. feed
speed (u) and disk wear (T)
0
40
80
120
160
200
240
280
320
360
400
0 200 400 600 800
Time [ms]
Time [ms]
Time [ms]
F
z
[
N
]
F
z
[
N
]
F
z
[
N
]
(a)
0
40
80
120
160
200
240
280
320
360
400
0 200 400 600 800
(b)
0
40
80
120
160
200
240
280
320
360
400
0 200 400 600 800
(c)
Fig. 7 Time domain force signal monitored in Z direction: a at 1
wear stage; b at 3 wear stage; c at 5 wear stage
T [#] u [mm/min] d [mm]
5 4 3 2 1
4
0
0
2
0
0
1
.
5
1
.
0
100
85
70
55
40
a
v
e
r
a
g
e
R
M
S
o
f
F
z
[
N
]
Fig. 8 Main effect plots of
RMS feature extracted by F
z
signal vs. feed speed (u), depth
of cut (d) and disk wear (T)
460 Int J Adv Manuf Technol (2007) 35:454467
have been observed for A
x
and A
y
signals. Figure 18 shows
the effects of the interaction between feed speed and disk
wear on RMS values of the A
z
signal.
Plots of the RMS feature vs. diamond disk wear were
produced for the three acceleration signals (see Fig. 19).
Inspection of Fig. 19 shows that the average value of RMS
of the A
x
and A
z
signals increases much more than that of A
y
(0.50 vs. 0.20). Moreover, the standard deviation of the A
z
signal increases considerably with disk wear (about 0.11),
while that of both A
x
and A
y
remains approximately constant
with wear increasing. Therefore, A
z
is the acceleration signal
most critical and it has been deeply investigated.
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
125
130
135
140
0 1 2 3 4 5 6
wear stage T [#]
R
M
S
o
f
F
z
[
N
]
Fx - average-sigma Fx - average Fx - average+sigma Fy - average-sigma Fy - average
Fy - average+sigma Fz - average-sigma Fz - average Fz - average+sigma
Fig. 10 The RMS feature of F
x
, F
y
and F
z
vs. diamond disk wear (T)
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0 1 2 3 4 5 6
wear stages [#]
p
r
o
t
r
u
s
i
o
n
h
e
i
g
h
t
[
m
m
]
Fig. 11 Average protrusion
height of the diamond grits
vs. wear stages
Int J Adv Manuf Technol (2007) 35:454467 461
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
125
130
135
140
0 1 2 3 4 5 6
wear stage T [#]
R
M
S
o
f
F
z
[
N
]
d=1 mm - average-sigma d=1 mm - average d=1 mm - average+sigma
d=1.5 mm - average-sigma d=1.5 mm - average d=1.5 mm - average+sigma
Fig. 13 RMS feature of F
z
vs. diamond disk wear (T) for different values of the depth of cut (d= and 1.5 mm)
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
125
130
135
140
0 1 2 3 4 5 6
wear stage T [#]
R
M
S
o
f
F
z
[
N
]
u=200 mm/min - average-sigma u=200 mm/min - average u=200 mm/min - average+sigma
u=400 mm/min - average-sigma u=400 mm/min -average" u=400 mm/min - average+sigma
Fig. 12 RMS feature of F
z
vs. diamond disk wear (T) for different values of the feed speed (u=200 and 400 mm/min)
462 Int J Adv Manuf Technol (2007) 35:454467
Figures 20 and 21 show the changes in the relationship
between RMS feature related to A
z
signal and disk wear with
the increase of the feed speed (u) or of the depth of cut (d).
Regression analysis underlined that the analytical rela-
tionship among RMS feature of A
z
, mill wear T, feed speed
u and depth of cut d is given by the following expression:
log RMS
Az
h
1
h
2
d h
3
u h
4
Th
5
u T 4
where h
1
, h
2
, h
3
, h
4
and h
5
are constants equal to 2.66,
0.419, 0.0028, 0.331 and 0.000332, respectively. The
adjusted coefficient of determination is about 93.4%, while
the hypotheses (normality and homogeneity of variance)
about the residuals are satisfied. The predictive capability
of the regression model is given by the value of the
predictive error sum of squares (PRESS) statistic, 92.7%,
respectively. The lack of fit test confirms that Eq. (4) may
be considered linear. Equation (4) may be written as:
RMS
Az
e
h
1
e
h
2
d
e
h
3
u
e
h
4
T
e
k
5
uT
5
3.4 Correlation between force and acceleration signals
RMS feature of F
z
signal may be put into relationship with
RMS feature of A
z
signal by means of the following
regression equation:
RMS
F
z
j
1
j
2
RMS
A
z
6
where j
1
and j
2
are constants equal to 11.4 and 133,
respectively (see Fig. 22). The adjusted coefficient of
determination is about 90.6%, while the hypotheses about
the residuals are satisfied. Equation (6) shows that it is
possible to estimate the average value of the cutting force,
once the average value of the vibrations on the workpiece
during the manufacturing process is known and vice-versa.
A summary of the variations of the cutting-process
parameters under different cutting conditions is shown in
Table 3 for the two considered sensors. The arrows denote
the tendency of variation of the above cutting process
parameters with changing machining conditions. The
upward arrow indicates that the monitored parameter
increases with an increase in that cutting variable, while
the other cutting conditions are kept constant. Similarly, the
downward arrow implies that the monitored parameter
-8
-6
-4
-2
0
2
4
6
8
0 200 400 600 800 1000
Time [ms]
Time [ms]
Time [ms]
A
y
[
m
/
s
^
2
]
A
y
[
m
/
s
^
2
]
A
y
[
m
/
s
^
2
]
(a)
-8
-6
-4
-2
0
2
4
6
8
0 200 400 600 800 1000
(b)
-8
-6
-4
-2
0
2
4
6
8
0 200 400 600 800 1000
(c)
Fig. 15 Time domain acceleration signal monitored in Y direction:
a at 1 wear stage; b at 3 wear stage; c at 5 wear stage
-8
-6
-4
-2
0
2
4
6
8
0 200 400 600 800 1000
Time [ms]
Time [ms]
Time [ms]
A
x
[
m
/
s
^
2
]
A
x
[
m
/
s
^
2
]
A
x
[
m
/
s
^
2
]
(a)
-8
-6
-4
-2
0
2
4
6
8
0 200 400 600 800 1000
(b)
-8
-6
-4
-2
0
2
4
6
8
0 200 400 600 800 1000
(c)
Fig. 14 Time domain acceleration signal monitored in X direction:
a at 1 wear stage; b at 3 wear stage; c at 5 wear stage
Int J Adv Manuf Technol (2007) 35:454467 463
decreases with an increase in that cutting condition while
the other cutting conditions are kept constant.
The models (3) and (5) allow to determine disk wear by
knowing the values of the RMS features, feed speed, and
depth of cut. In this way, it is possible to control in-process
disk wear on the basis of RMS feature measured, once the
feed speed to use for machining is fixed.
Once it is demonstrated that the force or acceleration
sensors may be alternatively used to control the diamond
disk wear, the choice of sensor depends on economic
factors. Generally, the acceleration sensor is cheaper than
the force sensor and, therefore, may be preferred in sawing
applications.
4 Conclusions
The system developed and used in this investigation uses the
capability of acceleration methodology and force methodol-
ogy for the monitoring of stone cutting by a diamond disk.
With information from the force and acceleration signals
acquired during the machining processes, the characteristics
of the cutting activity may be determined and these will
provide a means of monitoring the tool wear. In the stone
cutting by diamond disk, RMS feature of acceleration and
force signals along the z-axis are more sensitive to tool wear
and process parameters. Thus, this would lead to the de-
velopment of progressive tool wear monitoring on-line. The
force and the acceleration signals may be put into relation-
ship in order to foresee the average values of force and
vibration on the workpiece during cutting. The choice of
200
400
1 2 3 4 5
0.2
0.3
0.4
0.5
0.6
0.7
0.8
T [#]
u [mm/min]
a
v
e
r
a
g
e
R
M
S
o
f
A
z
[
m
/
s
^
2
]
Fig. 18 Interaction plot of RMS feature extracted by A
z
vs. feed
speed (u) and disk wear (T)
-8
-6
-4
-2
0
2
4
6
8
0 200 400 600 800 1000
Time [ms]
Time [ms]
Time [ms]
A
z
[
m
/
s
^
2
]
A
z
[
m
/
s
^
2
]
A
z
[
m
/
s
^
2
]
(a)
-8
-6
-4
-2
0
2
4
6
8
0 200 400 600 800 1000
(b)
-8
-6
-4
-2
0
2
4
6
8
0 200 400 600 800 1000
(c)
Fig. 16 Time domain acceleration signal monitored in Z direction:
a at 1 wear stage; b at 3 wear stage; c at 5 wear stage
T [#] u [mm/min] d [mm]
5 4 3 2 1
4
0
0
2
0
0
1
.
5
1
.
0
0.68
0.56
0.44
0.32
0.20
a
v
e
r
a
g
e
R
M
S
o
f
A
z
[
m
/
s
^
2
]
Fig. 17 Main effect plots of
RMS feature extracted by A
z
vs.
feed speed (u), depth of cut (d)
and disk wear (T)
464 Int J Adv Manuf Technol (2007) 35:454467
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0 1 2 3 4 5 6
wear stages T [#]
R
M
S
o
f
A
z
[
m
/
s
^
2
]
u=200 mm/min - average-sigma u=200 mm/min - average u=200 mm/min - average+sigma
u=400 mm/min - average-sigma u=400 mm/min - average u=400 mm/min - average+sigma
Fig. 20 RMS feature of A
z
vs. diamond disk wear (T) for different values of the feed speed (u=200 and 400 mm/min)
0.00
0.20
0.40
0.60
0.80
1.00
1.20
0 1 2 3 4 5 6
wear stages T [#]
R
M
S
o
f
a
c
c
e
l
e
r
a
t
i
o
n
s
[
m
/
s
^
2
]
Ax - average-sigma Ax - average Ax - average+sigma Ay - average-sigma Ay - average
Ay - average+sigma Az - average-sigma Az - average Az - average+sigma
Fig. 19 RMS feature of A
x
, A
y
and A
z
vs. diamond disk wear (T)
Int J Adv Manuf Technol (2007) 35:454467 465
0.00
0.20
0.40
0.60
0.80
1.00
1.20
0 1 2 3 4 5 6
wear stages T [#]
R
M
S
o
f
A
z
[
m
/
s
^
2
]
d=1 mm - average-sigma d=1 mm - average d=1 mm - average+sigma
d=1.5 mm - average-sigma d=1.5 mm - average d=1.5 mm - average+sigma
Fig. 21 RMS feature of A
z
vs. diamond disk wear (T) for different values of the depth of cut (d=1 and 1.5 mm)
RMS of Az [m/s^2]
R
M
S
o
f
F
z
[
N
]
1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1
140
120
100
80
60
40
20
0
Fig. 22 Regression equation of
RMS feature of F
z
vs. RMS
feature of A
z
466 Int J Adv Manuf Technol (2007) 35:454467
sensor depends on economic factors. Generally, the acceler-
ation sensor is cheaper than the force sensor and, therefore,
may be preferred in sawing applications.
Acknowledgement This research work was financed by the Italian
M.I.U.R. (Ministry of Instruction, University and Research) under the
project entitled Intelligent sensor monitoring of stone machining
(MIUR-PRIN 2002).
References
1. Dimla Sr DE (2000) Sensor signals for tool-wear monitoring in
metal cutting operations: a review of methods. Int J Mach Tools
Manuf 40:10731098
2. Ravindra HV, Srinivasa YG, Krishnamurthy R (1997) Acoustic
emission for tool condition monitoring in metal cutting. Wear
212:7884
3. Ertekin YM, Kwon Y, Tseng TL (2003) Identification of common
sensory features for the control of CNC milling operations under
varying cutting conditions. Int J Mach Tools Manuf 43:897904
4. Dimla Sr DE, Lister PM (2000) On-line metal cutting tool
condition monitoring. I: force and vibration analyses. Int J Mach
Tools Manuf 40:739768
5. Ghani AK, Choudhury IA, Husni (2002) Study of tool life,
surface roughness and vibration in machining nodular cast iron
with ceramic tool. J Mater Process Technol 127:1722
6. El-Wardany TI, Gao D, Elbestawi MA (1996) Tool condition
monitoring in drilling using vibration signature analysis. Int J
Mach Tools Manuf 36(6):687711
7. Lim GH (1995) Tool-wear monitoring in machine turning. J Mater
Process Technol 51:2536
8. Chen JC, Chen W-L (1999) A tool breakage detection system
using an accelerometer sensor. J Intell Manuf 10:187197
9. Mehta NK, Pandey PC, Chakravarti G (1983) An investigation of
tool wear and the vibration spectrum in milling. Wear 91:219234
10. Guinea D, Ruiz A, Barrios LJ (1991) Multi-sensor integration-an
automatic feature selection and sate identification methodology
for tool wear estimation. Comput Ind 17:121130
11. Tonshoff HK, Warnecke G (1982) Research on stone sawing. In:
Daniel P (ed) Advances in ultrahard materials application
technology, vol. 1. De Beers Industrial Diamond Division, pp
3649
12. Asche J, Tonshoff HK, Friemuh T (1999) Cutting principles, wear
and applications of diamond tools in the stone and civil
engineering industry. In: Proc Int Workshop on Diamond Tool
Production, pp 151157
13. Luo SY, Liao YS (1995) Study of the behaviour of diamond
saw-blades in stone processing. J Mater Process Technol 51:
296308
14. Luo SY, Liao YS (1992) Wear characteristics of sintered diamond
composite during circular sawing. Wear 157:325337
15. Wright DN, Wapler H (1986) Investigations and prediction of
diamond wear when sawing. Ann CIRP 35:239244
16. Jahanmir S, Ramulu M, Koshy P (1999) Machining of ceramics
and composites. Marcel Dekker, New York
17. Polini S, Turchetta S (2004) Test protocol for micro-geometric
wear of sintherized diamond tools. Wear 257:246256
18. Turchetta S, Polini W, Carrino L (2002) Wear progression of
diamond mills: a micro-geometric study, Euro PM2002 European
Conference on Hard Materials and Diamond Tooling, Lausanne,
Switzerland
19. Scheffer C, Heyns PS (2001) Wear monitoring in turning
operations using vibration and strain measurements. Mech Syst
Signal Process 15(6):11851202
Table 3 The tendency of variation of the monitored process
parameters under different machining conditions
Investigated
parameters
Feed speed u
(mm/min)
Depth of cut d
(mm)
Disk
wear
RMS of Fz
RMS of Az
Int J Adv Manuf Technol (2007) 35:454467 467