International Journal of Machine Tools & Manufacture 50 (2010) 10571064

Contents lists available at ScienceDirect

International Journal of Machine Tools & Manufacture

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

A comparative study on dry milling and little quantity lubricant milling

based on vibration signals
Weiwu Zhong a,n, Dongbiao Zhao a, Xi Wang b
a
College of Mechanical and Electrical Engineering, Nanjing University of Aeronautics and Astronautics, P.O. Box 245, 210016 Nanjing, Jiangsu, Peoples Republic of China
b
School of Aeronautical and Manufacturing Engineering, Nanchang Hangkong University, 330063 Nanchang, Jiangxi, Peoples Republic of China

a r t i c l e in fo abstract

Article history: Green manufacturing is the theme of manufacturing industry in the 21st century. In order to realize
Received 23 April 2010 green manufacturing, it is critical to decrease the usage of cutting uid in machining as much as
Received in revised form possible. Presently, there are still a lot of difculties in the adoption of dry cutting and MQL cutting for
26 August 2010
various reasons. This paper presents a new method called little quantity lubrication (LQL) in machining
Accepted 27 August 2010
and a comparative study on dry milling and LQL milling based on vibration signals. The vibration signals
Available online 7 September 2010
were acquired from workpiece surface in peripheral milling and were analyzed in time domain, fre-
Keywords: quency domain and timefrequency domain. The results show that vibration signals can be signicantly
Vibration signal affected by cutting uid in milling process. For the sake of reducing vibration and cutting uid usage,
Milling
process parameters should be considered while deciding whether or not and how to apply cutting uid.
Lubricant
This research gives a valuable insight in applying LQL in machining.
Fast Fourier transform
Wavelet packet transform & 2010 Elsevier Ltd. All rights reserved.

1. Introduction friendly characteristics. But they raise special demands to

In order to improve workpiece quality and prolong tool life,
cutting uid is introduced to cutting zone in metal cutting
process. In general, cutting uid used in metal machining plays an
important role in promoting productive efciency and improving
process quality. An ample amount of cutting uid used in
machining can induce serious pollution of the environment in
the whole manufacturing process. Oil mist, smoke, chemical
particles and bacteria, which are caused by cutting uid can easily
pose a threat to the health of production workers [1,2]. Processes
of steam cleaning and high-temperature treatment, which are
used to remove the impurities on the workpiece surface, can
generate the pollution of secondary effect [3]. Otherwise, from
economic consideration, the cost of using cutting uid is very
expensive. From the viewpoint of environment protection and
cost saving, it is critical to decrease the usage of cutting uid in
machining as much as possible in order to keep pace with the new
trend of green manufacturing. The common approaches of green
manufacturing relating to cutting uid are dry cutting and MQL
cutting.
Dry cutting and MQL cutting have been accepted as successful
applications in machining because of their environmentally
n MQL in turning of AISI-1040 steel. This study compares the
Corresponding author. Tel./fax: + 86 0 25 84896370
E-mail addresses: zhongweiwu2008@yahoo.com.cn (W. Zhong),
zdbme@nuaa.edu.cn (D. Zhao).
machine tools and cutting tools. Especially, there are still a lot
of difculties in the adoption of high-speed dry cutting because of
the high cutting force and temperature in machining. Without
cutting uid, the function of cooling and lubrication will not exist,
and this will cause severe friction and adhesion, reduce tool life
and affect surface quality of workpiece.
As can be seen from Table 1, there is a large interval of
lubricant ow rate between MQL and conventional wet lubrica-
tion (300300 000 ml/h), which is rarely used. Such a method of
little quantity lubrication is called LQL in this paper. In this work,
LQL means that the ow rate of cutting uid is 50300 ml/min.
Till now, far less attention has been paid to the application of LQL
in machining. The consideration of introducing LQL in machining
aims at achieving a good cooling and lubricating effect by using a
small amount of cutting uid. By this means, cutting process is in
the best condition (not shorten tool life, not worsen the surface
quality of workpiece), and the usage of cutting uid can be
reduced greatly to alleviate the negative effects of cutting uid on
human health and environment.
Many scholars have conducted research on the effect of
lubricant on machining process. Wang and Chang [4] conducted
an experimental study of surface roughness in slot-end milling.
The results showed that dry-cut roughness was reduced by
applying cutting uid. Dhar et al. [5] investigated the effect of
mechanical performance of MQL to completely dry lubrication
for the turning of AISI-1040 steel based on experimental

0890-6955/$ - see front matter & 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijmachtools.2010.08.011
1058 W. Zhong et al. / International Journal of Machine Tools & Manufacture 50 (2010) 10571064

Table 1
Flow rate of cutting uid for different cutting styles.

Cutting style Dry cutting MQL cutting Others Wet cutting

Flow rate of cutting uid 0 2300 ml/h Z 300 l/h

Cutting uid Cutting oil Cutting oil/emulsion Cutting oil/emulsion

Table 2
Experimental design.
Spindle
Nozzle Group no. Test no. n (rpm) ar (mm) f (mm/tooth)
Digital Valve
1 3000 0.5 0.05
End mill 2 6000 0.5 0.05
Driving card 1 3 10000 0.5 0.05
Workpiece Vibration sensor 4 14000 0.5 0.05
5 18000 0.5 0.05
Computer
Machine bed 6 3000 1.5 0.05
7 6000 1.5 0.05
Acquisition card 2 8 10000 1.5 0.05
9 14000 1.5 0.05
10 18000 1.5 0.05
Amplifier
11 3000 3.0 0.05
12 6000 3.0 0.05
Fig. 1. Experimental setup. 3 13 10000 3.0 0.05
14 14000 3.0 0.05
15 18000 3.0 0.05
measurement of cutting temperature, chip reduction coefcient,
cutting forces, tool wears, surface nish and dimensional devia- 16 18000 3.0 0.010
17 18000 3.0 0.015
tion. Jayal and Balaji [6] conducted an experimental study into the
4 18 18000 3.0 0.020
effects of different lubricant application methods on tool wear. 19 18000 3.0 0.025
Tsao [7] reported an experimental study of cutting uid effect in 20 18000 3.0 0.030
milling aluminum alloy in comparing the milling force responses
and ank wear under various cutting conditions. Aldo et al. [8]
investigated tool wear mechanisms and tool life for an end milling signals. Vibration signals were sampled by data acquisition card
operation under four different cooling and lubricant conditions. Advantech PCI-1714U, the sampling frequency was 2 MHz. In
From these researches, it can be known that no attention has been order to assure the comparability among the tests, vibration
paid to the effect of lubricant on cutting vibration. signals were sampled at the same position that was approxi-
Cutting vibration is mainly caused by the cutting forces and mately in the middle of every stroke.
motions involved in cutting process. Yesilyurt [9] conducted a Experimental design is shown in Table 2. n is spindle speed at
research of end mill breakage detection using mean frequency rpm, ar is radial depth of cut, f is feed per tooth. There are 20 tests
analysis of scalogram. His results showed that vibration signals in in the experiment, which are divided into four groups. Each of the
machining contained rich information of cutting state. Generally, 20 tests was conducted according to dry cutting (Cd), LQL cutting
cutting vibration is considered to be a very harmful phenomenon at 150 ml/min (C150) and LQL cutting at 300 ml/min (C300).
because it can lead to tool wear and worse surface quality of Therefore, there are 60 tests in the experiment altogether.
workpiece. In this paper, a comparison study of vibration signals
on dry cutting and LQL cutting is conducted in time domain,
frequency domain and timefrequency domain. The main purpose
3. Results and discussion
of this work is to reveal the effects of LQL on cutting process and
provide an effective instruction to apply LQL to machining.
In metal machining process, vibration is caused by cyclic varia-
tions in the dynamic components of cutting force [10]. Different
ow rates of cutting uid may cause different friction character-
2. Experimental setup
istics of the secondary deformation zone between cutting tool and
chip and the third deformation zone between cutting tool and
Experimental setup is illustrated in Fig. 1. The experimental workpiece. The change in the friction characteristics may result in
work was performed on Mikron UCP600 5-axis machining center. the change of dynamic cutting force. Correspondingly, the features
Up milling was performed in the experiment. The workpiece of random vibration of cutting tool and workpiece can be changed,
material used was aeronautic aluminum alloy 7050-T7451 in the which in turn affects the cutting vibration characteristics.
form of a 150 mm  150 mm  45 mm block. A new diamond-
coated at end mill (20 mm diameter, 3-ute) was used in the
tests. Axial depth of cut (aa) in all the tests was xed to 5 mm in 3.1. Time domain analysis
order to limit the amount of testing times. Cutting uid was a
fully synthetic water soluble coolant in a volumetric concentra- A typical time waveform of vibration signal is shown in Fig. 2.
tion of 1:20, which was provided accurately by a digital valve to According to sampling rate and spindle speed, it can be calculated
cutting zone. In order to obtain a higher injection speed of cutting out that 8571 data are sampled in every cutting cycle. A cutting
uid, the diameter of injection nozzle was only 1 mm. Piezo- cycle is dened as the time required for the cutting edge to
electric accelerometer sensor was used to sample vibration complete a 3601 cut. Fig. 2 shows that the vibration signals are
 W. Zhong et al. / International Journal of Machine Tools & Manufacture 50 (2010) 10571064 1059

0.2

Magnitude, V
0

-0.2
0 0.5 1 1.5 2 2.5
Sampling data point x 104

0.2
Magnitude, V

-0.2
0 0.5 1 1.5 2 2.5
Sampling data point x 104

0.5
Magnitude, V

-0.5
0 0.5 1 1.5 2 2.5
Sampling data point x 104

Fig. 2. Time waveform of vibration signals. n 14000 rpm, f 0.05 mm/tooth, ar 0.5 mm. (a) Vibration signals of Cd, (b) vibration signals of C150 and (c) vibration signals
of C300.

almost periodic time-varying. The almost periodic time-varying

characteristic is due to the intermittent machining characteristic 0.1 0.14
of milling process. Each period consists of three sub-periods, Cd
0.12 Cd
which are created by three utes of cutting tool. It can be seen 0.08 C150 C150

RMS value, V
RMS value,V

C300 0.1 C300

from Fig. 2 that cutting vibration in milling was mainly due to the 0.06 0.08
intermittent cutting force. There is no signicant difference in
amplitude among the signals of Cd, C150 and C300. From the 0.04 0.06
shape of time waveform in Fig. 2, it can be seen that cutting uid 0.04
0.02 0.02
has a signicant inuence on cutting vibration. Three sub-periods
in every cutting cycle are obscure for dry cutting, but they are 0 0
easy to distinguish for C300. Therefore, it can be concluded that n1 n2 n3 n4 n5 n1 n2 n3 n4 n5
cutting uid can speed up the attenuation of vibration in milling. RMS value of group 1 RMS value of group 2
With almost all of the signals the most common parameter to
look at is the RMS value. RMS value is usually regarded as a 0.14
0.16 Cd
meaningful parameter in quantifying the intensity of a signal. In
0.14 Cd 0.12 C150
order to get the accurate values, all the RMS values are calculated C150
0.12
RMS value, V

RMS value, V

0.1 C300
from the data in two cutting cycles. The RMS values of vibration C300
0.1 0.08
signal are shown in Fig. 3.
0.08
From Fig. 3(ac) it can be seen that the RMS values of vibration 0.06
0.06
signal are almost constant at three different cutting uid ow 0.04
0.04
rates when cutting is conducted at lower speeds (3000 and 0.02 0.02
6000 rpm). When cutting is conducted at a higher speed (10 000, 0 0
14 000 and 18 000 rpm), RMS value of C150 is less than that of Cd, n1 n2 n3 n4 n5 f1 f2 f3 f4 f5
and with an increase in ar, the RMS value of C300 tends to be RMS value of group 3 RMS value of group 4
greater than that of C150. This phenomenon can be explained as
Fig. 3. RMS value of vibration signals. Main speed: n1 3000 rpm, n2 6000 rpm,
follows: when cutting speed is low, the heat generated in cutting
n3 10 000 rpm, n4 14 000 rpm, n5 18 000 rpm. Feed per tooth: f1 0.010 mm,
process is small, so the cutting temperature is relatively low. In f2 0.015 mm, f3 0.020 mm, f4 0.025 mm, f5 0.030 mm. (a) RMS value of group
such a cutting process, the main function of cutting uid is 1, (b) RMS value of group 2, (c) RMS value of group 3 and (d) RMS value of group 4.
lubrication. In fact, there is only little lubricant being introduced
to cutting zone. Therefore, the ow rate of cutting uid has little
effect on vibration signal. When cutting speed increases, the
cutting heat generated in cutting process increases, the tempera- the same as that of 300 ml/min, but the cooling capacity of
ture of workpiece in cutting zone increases and metal material is 300 ml/min is greater than 150 ml/min. Thus, the softening effect
softened. The lubrication effect of ow rate 150 ml/min is almost on workpiece in cutting zone for 150 ml/min is bigger than that
1060 W. Zhong et al. / International Journal of Machine Tools & Manufacture 50 (2010) 10571064

for 300 ml/min. As a result, the RMS value of C300 is greater than by cutting heat, is weakened by cutting uid. Therefore, cutting
that of C150. It can also be found that increase in cutting speed force increases. As a result, the RMS value of vibration signal rises
results in dramatic increase in RMS value. up with increase in cutting uid.
Fig. 3(d) shows that the RMS value of vibration signal increases
with an increase in feed per tooth and the RMS value increases
approximately linearly with increase in ow rate of cutting uid. 3.2. Frequency domain analysis
The reason may be that increase in feed per tooth will lead to a
rise in cutting force. Therefore, RMS value of vibration signals Analysis in frequency domain is often useful in providing
increases. Because of the higher cutting speed (n18 000 rpm) interesting information for vibration signals. Usually, frequency
and the bigger radial depth of cut (ar 3 mm), more cutting heat analysis is a research method for stationery signal, which can
will be generated. In this condition, cooling effect of cutting uid reveal the composition of signal frequency components. It has
is greater than lubrication effect; softening effect, which is caused been known from Fig. 2 that vibration signals in milling process
are almost periodic time-varying. Although vibration signal is not
strictly periodic, it is reasonable to apply frequency analysis
4 method on the research of vibration signal in milling.
Axis Z:magnitude (V)

The widely used fast Fourier transform (FFT) provides a way to

3 nd out the frequency content of a measured signal. Assuming the
ow rate of cutting uid inuences the frequency contents of the
2 measured signal in machining, FFT then gives an inside view of
this process. In this work, spectra of cutting vibration signals are
1 represented by FFT. In order to eliminate the effect of static
component, the vibration signals were subtracted by the mean
0 before performing FFT. Fig. 4 shows the spectra of group 2. From
15
10 Fig. 4, it can be seen that the frequency component of vibration
10 signal is concentrated in the low-frequency band of 020 kHz.
Axi 5 5 x 104
sY , Hz There are few frequency components in 4080 kHz. This phenom-
ency
0 0 X:fre u
q enon also exists in other tests of the experiment. Another
Axis conclusion can be made from Fig. 4 that the magnitude of vibra-
Fig. 4. FFT of group 2. Axis Y: 13: CdC150C300, n 3000 rpm; 46: CdC150
tion increases when the cutting speed is increased.
C300, n 6000 rpm; 79: CdC150C300, n 10 000 rpm; 1012: CdC150C300, It can be seen from Figs. 57 that the vibration spectra have
n 14 000 rpm; 1315: CdC150C300, n 18 000 rpm. three main peak frequency bands: 02, 38 and 1016 kHz. The

15 15
Axis Z:magnitude (V)

Axis Z:magnitude (V)

10 10

5 5

0 0
10 15
2 2
Ax 5
is 1 x 104 Ax 1 x 104
Y X is
0 0 Axis Y 10 0 Axis
X

Fig. 5. FFT of group 1. Axis X: frequency, Hz. Axis Y: 13: CdC150C300, n 3000 rpm; 46: CdC150C300, n 6000 rpm; 79: CdC150C300, n 10 000 rpm; 1012:
CdC150C300, n 14 000 rpm; 1315: CdC150C300, n 18 000 rpm.

10 20
Axis Z:magnitude (V)

Axis Z:magnitude (V)

5 10

0 0
10 15
2 2
Ax 5 Ax
is 1 x 104 is 1 x 104
Y X Y X
0 0 Axis 10 0 Axis

Fig. 6. FFT of group 2. Axis X: frequency, Hz. Axis Y: 13: CdC150C300, n 3000 rpm; 46: CdC150C300, n 6000 rpm; 79: CdC150C300, n 10 000 rpm; 1012:
CdC150C300, n 14 000 rpm; 1315: CdC150C300, n 18 000 rpm.
 W. Zhong et al. / International Journal of Machine Tools & Manufacture 50 (2010) 10571064 1061

20 15

Axis Z:magnitude (V)

Axis Z:magnitude (V)

10
10
5

0 0
10 15
2 2
Ax 5
is 1 x 104 Ax 1 x 104
Y X is Y X
0 0 Axis 10 0 Axis

Fig. 7. FFT of group 3. Axis X: frequency, Hz. Axis Y: 13: CdC150C300, n 3000 rpm; 46: CdC150C300, n 6000 rpm; 79: CdC150C300, n 10 000 rpm; 1012:
CdC150C300, n 14 000 rpm; 1315: CdC150C300, n 18 000 rpm.

20 15
Axis Z:magnitude (V)

Axis Z:magnitude (V)

10
10
5

0 0
10 15
2 2
Ax 5 Ax
is 1 x 10 4
is 1 x 104
Y X Y X
0 0 Axis 10 0 Axis

Fig. 8. FFT of group 4. Axis X: frequency, Hz. Axis Y: 13: CdC150C300, f 0.010 mm; 46: CdC150C300, f 0.015 mm; 79: CdC150C300, f 0.020 mm; 1012: Cd
C150C300, f 0.025 mm; 1315: CdC150C300, f 0.030 mm.

amplitude in the three main frequency bands changes with ow the amplitude of frequency band in 02 kHz increases, the
rate of cutting uid. Fig. 5 shows the spectra of group 1. At the amplitude of frequency band in 38 kHz decreases, which is
lower cutting speeds of 3000, 6000, 10 000 and 14000 rpm, with similar to Figs. 5 and 6. At the higher cutting speeds of 10 000,
increase in ow rate of cutting uid, the amplitude of spectrum 14 000 and 18 000, the amplitude in 02 kHz increases for
shows different characteristics: it increases in the frequency C150, while the amplitude in 38 kHz decreases for C300.
region 02 kHz, but it decreases in the frequency region 38 kHz. The amplitude of frequency bands 1016 kHz has a large increase
There is no obvious change in 1016 kHz. At the higher cutting for C300 at higher cutting speeds of 10 000, 14 000 and
speed of 18 000 rpm, the amplitude of signals in the frequency 18 000 rpm. Fig. 3(c) shows that the RMS value of C300 is
band of 1016 kHz increases quickly with increase in ow rate of greater than that of C150 at cutting speeds of 10 000, 14 000
cutting uid. From Fig. 3(a), it can be seen that the RMS value of and 18 000 rpm. It can also be concluded that the higher RMS
C300 is greater than that of C150 at the speed of 18 000 rpm. value of C300 is caused by increase in amplitude in the frequency
Therefore, it can be concluded that the higher RMS value of C300 region 1016 kHz. Another phenomenon can also be found
is caused by increase in amplitude in frequency region 1016 kHz. from Fig. 7 that the main frequency component of vibration
Fig. 6 shows the spectra of group 2. It can be seen from Fig. 6 signals distributes in a wider region with increase in cutting
that, at the lower cutting speeds of 3000, 6000 and 10 000 rpm, speed.
increase in ow rate of cutting uid results in increase in Fig. 8 shows the spectra of group 4. As can be seen from
amplitude of frequency bands 02 kHz, but results in decrease Fig. 4(d) that there is no obvious change in spectrum at different
in amplitude of frequency bands 38 kHz. There is no obvious cutting uid ow rates in the frequency bands of 02 and
change in 1016 kHz. At the higher cutting speeds of 14 000 and 38 kHz, but the frequency component in frequency bands
18 000 rpm, the amplitude in 02 kHz increases for C150, while 1016 kHz increases with increase in cutting uid ow rate.
the amplitude in 38 kHz decreases for C300. The amplitude of Considering the conclusion from Fig. 3(d) that the RMS value will
frequency bands 1016 kHz has a large increase for C300 at the increase linearly with the supply of cutting uid, it can also be
higher cutting speeds of 14 000 and 18 000 rpm. Fig. 3(b) shows concluded that an increase in the RMS value is contributed to an
that the RMS value of C300 is greater than that of C150 at cutting increase in amplitude in frequency bands 1016 kHz.
speeds of 14 000 and 18 000 rpm. It can be concluded that the
higher RMS value of C300 is caused by an increase in amplitude in
frequency region 1016 kHz. Another phenomenon can be found 3.3. Timefrequency domain analysis
that the main frequency component of vibration signals distri-
butes in a wider region with increase in cutting speed. Wavelet packet transforms (WPTs) is often used to analyze the
Fig. 7 shows the spectra of group 3. With increase in ow rate frequency content of a signal and provide a possibility of time
of cutting uid at the lower cutting speeds of 3000 and 6000 rpm, frequency localization. In this study, wavelet db3 and entropy
1062 W. Zhong et al. / International Journal of Machine Tools & Manufacture 50 (2010) 10571064

standard Shannon are adopted to analyze vibration signals in Feature vector E2 can describe energy distribution in the
order to reveal the effect of cutting uid on different frequency frequency band of 019 531 Hz. In order to express the energy
bands. Signals are decomposed by 12 layers wavelet packet distribution of vibration signals clearly, E2 is compressed from 80
decomposition and wavelet packet coefcient series of 4096 (212) dimensions to 10 dimensions by
frequency bands are obtained. It can be calculated out that the
E3 t1 ,t2 ,. . .,t10 : 5
width of each frequency band is 244 Hz. Considering that the
most energy of vibration signals is concentrated in the frequency where
band of 020 kHz, the rst 80 wavelet packets are researched,
X
8 X
16 X
80
which contain the signal component of 019 531 Hz. t1 Ti , t2 Ti ,. . .,t10 Ti :
4096 frequency bands are determined by: i1 i9 i 73

f j,i j1fs =2i ,jfs =2i : 1 Fig. 9(a) shows the energy ratios of t1t10 in group 1. It can be
seen from Fig. 9(a) that there are three energy peaks in the
researched frequency region, which are band 1(01953 Hz), bands
where f (j,i) is the jth frequency band in ith layer, j 1,2, y, 2i, fs is
24(19547812 Hz) and bands 68(976715 625 Hz). There are
sampling rate. The energy of each frequency band can be obtained
the same phenomena in other three groups.
by
In order to reveal the distribution characteristic of signal
Z t X
n   energy, a new feature vector can be obtained by
E12,j s212,j t dt xj,k 2 2
0 k1 E4 t1 ,t2 t3 t4 ,t6 t7 t8 : 6
where j 0,1, y, 79, t is the whole machining time, n the length of
analyzed data, xj,k the amplitude of the kth discrete time point in Fig. 9(bd) shows energy ratios of the three frequency regions
the jth frequency band. (band 1, bands 24 and bands 68) in group 1. As can be seen
Feature vector E1 is constructed by from Fig. 9(b), energy ratio of C150 is bigger than that of Cd,
E1 E12,0 ,E12,1 ,. . .,E12,79 : 3 whereas energy ratio of C300 is smaller than that of C150. It can
be seen from Fig. 9(c) that, in bands 24, the energy ratio of C150
and C300 is smaller than that of Cd. Fig. 9(d) shows that, in bands
Considering the energy value may be too big when the amount 68, the energy ratio of C300 is much bigger than that of Cd and
of analyzed data is too large, E1 is normalized by C150, the energy ratio of C150 is smaller than that of Cd. It can
also be seen from Fig. 9(bd) that the energy ratio decreases in
E2 T1 ,T2 ,. . .,T80  E12,0 =E,E12,1 =E,. . .,E12,79 =E: 4
band 1 and increases in bands 68 with increase in cutting speed.
From Figs. 10 and 11, it can be seen that there are almost the same
where phenomena in group 2 and group 3.
X
79 From the results above it is clear that the ow rate of cutting
E E12,j : uid has a signicant impact on the energy distribution in
j0 different frequency bands. On the one hand, little quantity

90
80
70 100
Energy ratio, %

60
80
Energy ratio,%

50
40 60
30
20 40
S11
10 20
S6
0
0
1

S1
3
5
7
9

Frequency ba 1 3 5 7 9 11 13 15
nd

70 50
60
Energy ratio,%

40
Energy ratio,%

50
40 30
30 20
20
10
10
0 0
1 3 5 7 9 11 13 15 1 3 5 7 9 11 13 15
Fig. 9. Energy ratio of group 1. S1S3, 13:CdC150C300,n 3000 rpm; S4S6, 46: CdC150C300, n6000 rpm; S7S9,79: CdC150C300, n10 000 rpm; S10S12,
1012: CdC150C300, n14 000 rpm; S13S15, 1315:CdC150C300, n18 000 rpm. (b) Energy ratio of band 1, (c) energy ratio of bands 24 and (d) energy ratio of bands 68.
 W. Zhong et al. / International Journal of Machine Tools & Manufacture 50 (2010) 10571064 1063

80 80
70 70

Energy ratio, %
Energy ratio, %
60 60
50 50
40 40
30 30
20 20
10 10
0 0
1 3 5 7 9 11 13 15 1 3 5 7 9 11 13 15

80
70
Energy ratio, %

60
50
40
30
20
10
0
1 3 5 7 9 11 13 15
Fig. 10. Energy ratio of group 2. 13:CdC150C300, n 3000 rpm; 46: CdC150C300, n 6000 rpm; 79: CdC150C300, n 10 000 rpm; 1012:CdC150C300,
n 14 000 rpm; 1315: CdC150C300, n 18 000 rpm. (a) Energy ratio of band 1, (b) energy ratio of bands 24 and (c) energy ratio of bands 68.

70 80
60 70
Energy ratio, %

Energy ratio, %

50 60
50
40
40
30
30
20 20
10 10
0 0
1 3 5 7 9 11 13 15 1 3 5 7 9 11 13 15

70
60
Energy ratio, %

50
40
30
20
10
0
1 3 5 7 9 11 13 15
Fig. 11. Energy ratio of group 3. 13: CdC150C300, n 3000 rpm; 46: CdC150C300, n6000 rpm; 79: CdC150C300, n10 000 rpm; 1012: CdC150C300,
n 14 000 rpm; 1315: CdC150C300, n 18 000 rpm. (a) Energy ratio of band 1, (b) energy ratio of bands 24 and (c) energy ratio of bands 68.

lubricant in machining results in the increase of energy ratio in
band 1, but the energy ratio in band 1 will decrease when the ow
rate of lubricant reaches a certain threshold, which is about
150 ml/min, but on the other hand, in bands 68, little quantity
lubricant can result in the decrease of energy ratio, but the
energy ratio can reach to a high value when the ow rate of
lubricant exceeds a certain threshold. Generally speaking,
vibration energy concentrates in a lower frequency region means
a high-quality cutting process. Therefore, it is benecial to apply
an appropriate ow rate of lubricant to assure a high-quality
machining process.
From Fig. 12, it can be seen that the energy ratio drops with an
increase in cutting uid in band 1 and bands 24, the energy
ratios of C150 and C300 are much higher than that of Cd in bands
68. It can be concluded that, for high cutting speed (18 000 rpm)
and big ar, the introduction of lubricant can result in a small
energy ratio in band 1 and a large energy ratio in bands 68.
4. Conclusions
In this paper, a comparison research of drying milling and LQL
milling has been experimentally conducted. It has been found
that vibration in peripheral milling process is signicantly
affected by cutting uid. Some interesting conclusions can be
made through different analysis methods, including time domain
method, frequency domain method and timefrequency domain
method.
1064 W. Zhong et al. / International Journal of Machine Tools & Manufacture 50 (2010) 10571064
50
40
Energy ratio, %
30
20
10
0 0
1 3 5 7
70
60
Energy ratio, %
50
40
30
20
10
0
1 3 5 7
Fig. 12. Energy ratio of group 4. 13: CdC150C300, f 0.010 mm; 46: CdC150C300, f 0.015 mm; 79: CdC150C300, f 0.020 mm; 1012: CdC150C300,
f 0.025 mm; 1315: CdC150C300, f 0.030 mm. (a) Energy ratio of band 1, (b) energy ratio of bands 24 and (c) energy ratio of bands 68.
The specic conclusions from this work are:
(1). Cutting vibration in peripheral milling is mainly generated by
intermittent cutting force, which is inherent in milling
process.
(2). Cutting uid can speed up the attenuation of cutting
vibration.
(3). At higher cutting speed, higher feed per tooth and higher
radial depth of cut, higher ow rate of cutting uid tends to
cause higher RMS value. Therefore, it is reasonable to apply
dry cutting or less quantity of uid cutting in rough
machining. In such machining conditions, for dry cutting, it
can not only avoid the consumption of cutting uid, but also
have a mild vibration; for less quantity of uid cutting, it can
decrease RMS value of vibration, which is benecial to
machines and cutting tools. In nishing process, it is
appropriate to apply LQL, which can not only assure a high
quality of workpiece, but also reduce the usage of cutting
uid as much as possible.
When the ow rate of lubricant is less than a certain
threshold, lubricant applied in machining results in
increase in energy ratio in low-frequency region, but the
energy ratio of low-frequency region will decrease when
the ow rate of lubricant reaches a certain threshold,
which is about 150 ml/min. On the contrary, little
lubricant can result in decrease in energy ratio in high-
frequency region, but the energy ratio can reach to a high
value when the ow rate of lubricant exceeds a certain
threshold. Therefore, it is benecial to apply an appro-
priate ow rate of lubricant to assure a high-quality
machining process.
100
80
Energy ratio, %
60
40
20
9 11 13 15 1 3 5 7 9 11 13 15
9 11 13 15
Acknowledgements
The authors would like to thank the National Natural Science
Foundation of China (Grant no.: 50875120) and Aeronautics
Science Foundation of China (Grant no.: 2008ZE52049) for
funding this work.
References
[1] H.S. Abdalla, W. Baines, G. McIntyre., C. Slade., Development of novel
sustainable neat-oil metal working uids for stainless steel and titanium
alloy machining. Part 1. Formulation development, International Journal of
Advanced Manufacturing Technology 34 (2007) 2133.
[2] S.Y. Hong, Mark Broomer, Economical and ecological cryogenic machining of
AISI 304 austenitic stainless steel, Clean Products and Processes 2 (2000)
157166.
[3] X.B. Wang, Green machining technology in metal cutting, Chinese Journal of
Mechanical Engineering (in Chinese) 36 (2000) 611.
[4] M.Y. Wang, H.Y. Chang, Experimental study of surface roughness in slot end
milling AL2014-T6, International Journal of Machine Tools and Manufacture
44 (2004) 5157.
[5] N.R. Dhar, M.T. Ahmeda, S. Islama, An experimental investigation on effect of
minimum quantity lubrication in machining AISI 1040 steel, International
Journal of Machine Tools and Manufacture 47 (2007) 748753.
[6] A.D. Jayal, A.K. Balaji, Effects of cutting uid application on tool wear in
machining: interactions with tool-coatings and tool surface features, Wear
267 (2009) 17231730.
[7] C.C. Tsao, An experiment study of hard coating and cutting uid effect in
milling aluminum alloy, International Journal of Advanced Manufacturing
Technology 32 (2007) 885891.
[8] B.J. Aldo, E.D. Anselmo, T.F. Fernando, Tool wear and tool life in end milling of
155 PH stainless steel under different cooling and lubrication conditions,
International Journal of Advanced Manufacturing Technology 43 (2009) 756764.
[9] I. Yesilyurt, End mill breakage detection using mean frequency analysis of
scalogram, International Journal of Machine Tools and Manufacture 46 (2006)
450458.
[10] D.E. Dimla Sr., The correlation of vibration signal features to cutting tool wear
in a metal turning operation, International Journal of Advanced Manufactur-
ing Technology 19 (2002) 705713.