You are on page 1of 15

Journal of Cleaner Production 209 (2019) 1508e1522

Contents lists available at ScienceDirect

Journal of Cleaner Production


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

Parameter optimization during minimum quantity lubrication milling


of TC4 alloy with graphene-dispersed vegetable-oil-based cutting
fluid
Ming Li a, Tianbiao Yu a, *, Lin Yang a, Hongyu Li a, Rongchuang Zhang b, Wanshan Wang a
a
School of Mechanical Engineering and Automation, Northeastern University, Shenyang 110819, China
b
School of Control Engineering, Northeastern University at Qinhuangdao, Qinhuangdao 066004, China

a r t i c l e i n f o a b s t r a c t

Article history: Minimum Quantity Lubrication has been widely used in the titanium alloy milling process as an
Received 3 August 2018 advanced and clean means of cooling and lubrication. The Minimum Quantity Lubrication parameters
Received in revised form have a significant influence on the milling characteristics and so, determining an optimal Minimum
15 November 2018
Quantity Lubrication parameter combination is vital to obtaining the best milling characteristics. In this
Accepted 16 November 2018
Available online 17 November 2018
study, Minimum Quantity Lubrication with graphene-dispersed vegetable-oil-based cutting fluids was
adopted in the milling of TC4 alloy, where the cutting fluids were prepared by dispersing graphene
nanoparticles into the vegetable-oil-based cutting fluid to improve the milling characteristics of the TC4
Keywords:
Minimum quantity lubrication milling
alloy. The integrated Taguchi-Principal component analysis-Gray relational analysis optimization method
Graphene was used to evaluate the effects of the Minimum Quantity Lubrication parameters on the milling char-
Vegetable-oil-based cutting fluid acteristics and obtain the optimal Minimum Quantity Lubrication parameter combination. The milling
Parameter optimization characteristics of TC4 alloy, namely, the milling force, milling temperature, surface micro-hardness, and
TC4 alloy surface roughness were evaluated and analyzed, and the optimal Minimum Quantity Lubrication
parameter combination was obtained. A verification experiment was conducted and the results indicated
that all the four milling characteristics were significantly improved after the optimization process. The
improvement rates of the milling force, milling temperature, surface micro-hardness, and surface
roughness are 18.13%, 13.59%, 8.36%, and 24.82%, respectively. In summary, appropriately chosen Mini-
mum Quantity Lubrication parameters can enhance the lubrication and cooling properties of the oil film
and improve the milling characteristics. The results of this study attest to the feasibility of the integrated
Taguchi-Principal component analysis-Gray relational analysis optimization method and provide an
experimental basis for the application of graphene additive in Minimum Quantity Lubrication milling.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction improve the machining characteristics during the processing of


difficult-to-machine materials (Debnath et al., 2014). This tech-
As an alloy which has excellent all-round performance, TC4 is nology can not only reduce the employed volume of cutting fluid
widely used in the aerospace, shipbuilding, petrochemical, and but also enhance the ability of the cutting fluid to penetrate the
automobile industries, in addition to medical applications. How- cutting zone (Vazquez et al., 2015). Meanwhile, vegetable-oil-based
ever, TC4 is a difficult-to-machine material due to its low thermal cutting fluids are widely adopted owing to their good biodegrad-
conductivity, large friction coefficient, high chemical activity, and ability, low environmental impact, and non-toxicity (Lawal et al.,
low elasticity modulus (Tao et al., 2017). Minimum Quantity 2012). Thus, the use of MQL with such cutting fluids opens up
Lubrication (MQL) is a method that has been widely used to several promising applications.
Many studies have been conducted on the machining charac-
teristics of vegetable-oil-based cutting fluids. Wang et al. (2016)
* Corresponding author. investigated the lubrication properties at the wheel/workpiece
E-mail addresses: limingeducn@163.com (M. Li), tianbiaoyudyx@gmail.com interface during the flood grinding and MQL grinding of the nickel-
(T. Yu), jwdcmyl1011@gmail.com (L. Yang), m18640389373@163.com (H. Li), based alloy GH4169, with seven typical vegetable oils (castor oil,
zhangrongchuang@mail.neuq.edu.cn (R. Zhang), wws@mail.neu.edu.cn (W. Wang).

https://doi.org/10.1016/j.jclepro.2018.11.147
0959-6526/© 2018 Elsevier Ltd. All rights reserved.
M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522 1509

Nomenclature NTB Nominal-the-better


LTB Larger-the-better
TC4 An aþb titanium alloy n Total number of experiments
MQL Minimum Quantity Lubrication yi Result of the i-th experiment
PCA Principal component analysis m The target value of yi
GRA Gray relational analysis yi ðkÞ Original SNR
GH4169 A nickel-based superalloy xi ðkÞ Normalized SNR
AISI American Iron and Steel Institute aðkÞ The target value of yi ðkÞ
Cu Copper GRC Gray relational coefficient
Al2O3 Aluminum oxide εi ðkÞ Experimental result GRC
Ag Silver D0i ðkÞ The absolute value of the deviation
TiO2 Titanium dioxide x0 ðkÞ The ideal value of xi ðkÞ
MoS2 Molybdenum disulfide Dmin The minimum value of D0i ðkÞ
CNT Carbon nanotube Dmax The maximum value of D0i ðkÞ
SiO2 Silicon dioxide z Distinguishing coefficient
PTFE Polytetrafluoroethylene ε GRC matrix
ANOVA Analysis of Variance Rjl Correlation coefficient array
SNR Signal-to-noise ratio Cov Covariance
N Spindle rotation speed lk Eigenvalue
fz Feed rate per tooth V ik Eigenvector
ae Radial immersion depth Im Unit matrix
ap Axial immersion depth Ymk The k-th principal component
wt.% Weight concentration ai The weight of GRC
W Cutting fluid concentration GRG Gray relational grade
Q Cutting fluid flow rate g Experimental result GRG
P Gas pressure Tmax The maximum milling temperature
PC Personal computer MH Surface micro-hardness
Fc Resultant force Ra Surface roughness
Fcmax Maximum resultant force DF Degree of freedom
Fx The x-axis milling force Seq. SS Sum of squares
Fy The y-axis milling force Adj. MS Mean of squares
Fz The z-axis milling force F Variance ratio
STB Smaller-the-better P Probability of significance

soybean oil, rapeseed oil, corn oil, sunflower oil, peanut oil, and oil facilitated longer tool life than the coconut oil and soybean oil.
palm oil) and paraffin oil were selected as the base oil during MQL. This is due to the higher density and heat transfer coefficient of
Vegetable oils had better lubrication properties than both mineral canola oil which allowed for maximum cooling and lubrication
oil and flood cutting fluid due to the high binding energy, low compared to the coconut and soybean oils. Ribeiro Filho et al.
friction coefficient. Stachurski et al. (2018) investigated the effect of (2017) evaluated the tapping characteristics of cast aluminum
hob cutter sharpening on cutting blade surface condition during alloy in MQL tapping with different vegetable fluids. They
MQL-CCA. The results showed that the amount of cutting fluid can confirmed that the fluid A (ECOCUT) could reduce the torque,
be greatly reduced without affecting the machining performance. minimize the friction, and avoid tool breakages. Gajrani et al. (2017)
Therefore, this technology can greatly reduce the environmental studied the biodegradation and cutting characteristics of a bio-
impact. Khan et al. (2009) studied the effects of MQL with vegetable cutting fluid and mineral oil in the MQL turning of hardened AISI
oil-based cutting fluid on the chip-tool interface temperature, chip H-13 steel. They stated that the bio-cutting fluid performed better
formation mode, tool wear, and surface roughness. Compared with with its higher thermal conductivity, higher specific heat, and
flood machining, the average chip-tool interface temperature was better ability to penetrate the cutting zone as compared to mineral
reduced by 10%, and the chip formation modes, tool wear, and oil, and the MQL with bio-cutting fluid could reduce the sticking
surface finish were improved in varying degrees. Therefore, MQL is and sliding zones. The results contribute to the further application
an environmentally friendly and efficient cooling method. Agrawal of bio-cutting fluid in the machining field. Li et al. (2016) evaluated
and Patil (2018) evaluated the machining effects of aloe vera oil in the cutting performance of different vegetable oils (castor oil,
the MQL turning of M2 Steel. Compared with conventional cutting soybean oil, rapeseed oil, corn oil, sunflower oil, peanut oil, and
fluid, the surface roughness and tool wear were reduced by 6.7% palm oil) in the MQL grinding of high-temperature nickel-base
and 0.14%, respectively, when aloe vera oil was used. Rahim and alloy. They concluded that palm oil provided the second-lowest
Sasahara (2011) studied the cutting performance of Ti-6Al-4V grinding force, lowest grinding temperature and lowest energy
alloy in MQL drilling using palm oil. Compared with synthetic es- ratio coefficient, and was the optimal base oil among the seven
ters, the palm oil generated lower cutting forces and cutting tem- vegetable oils. This is mainly because vegetable oils with higher
peratures. This was mainly because the thin boundary lubrication viscosities have greater lubrication effects and can significantly
film formed in the cutting zone reduced the friction and heat at the reduce the grinding force. This research provides a theoretical basis
tool-workpiece interfaces. Gunjal and Patil (2018) investigated the for the application of palm oil in the cutting field. Rahim and
cutting performance of hardened AISI 4340 steel in MQL turning HiroyukiSasahara (2011) evaluated the effects of palm oil and a
with vegetable-oil-based cutting fluids. They found that the canola synthetic ester on surface integrity in the MQL drilling of Inconel
1510 M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522

718. Compared with synthetic ester, palm oil exhibited excellent (2011) evaluated the machining characteristics of aluminum 6061
performance with regard to micro-hardness, surface defects, sub- in nano-diamond nanofluid-assisted MQL micro-drilling.
surface deformation, and surface roughness at all cutting speeds Compared with compressed air lubrication and pure MQL, the
and feed rates. This is mainly because palm oil, with its higher nano-diamond nanofluid MQL significantly increased the number
viscosity, could provide effective lubrication and cooling in MQL of drilled holes, reduced drilling torques and thrust forces, and
machining. Guo et al. (2017) mixed castor oil with other vegetable eliminated the remaining chips and burrs to enhance the quality of
oils and evaluated their lubrication performance in the MQL drilled holes. In addition, some other additives, such as hexagonal
grinding of the nickel-based alloy GH4169. They concluded that the boron nitride nanoparticle (Talib et al., 2017), nanoboric acid
comprehensive lubricating performances of the mixed oils were (Vamsi Krishna et al., 2010), SiO2 nanoparticle (Sayuti et al., 2014),
better than that of castor oil, and soybean/castor oil achieved the lignin (Zhang and Jun 2014), PTFE nanoparticle (Sartori et al., 2018),
best performance. Mia et al. (2018b) studied an approach to cleaner and ionic liquid (Goindi et al., 2018), were also dispersed in the
production for machining hardened steel from the aspect of envi- cutting fluids. In addition, Zhang et al. (2017) combined the mate-
ronmental friendliness. The Pugh matrix environmental approach rial removal and plastic stacking to improve the prediction model of
was adopted to establish the sustainability assessment model, and the grinding force, while considering the lubrication conditions,
the research results showed that MQL is an environmentally such as flood, MQL, and nanofluid MQL. This greatly improves the
friendly, low consumption and efficient cooling and lubrication accuracy of the prediction model of the grinding force under
method. lubrication conditions. Yang et al. (2017) studied the critical
In addition, in order to further improve the cooling and lubri- maximum undeformed equivalent chip thickness for ductile-brittle
cation performances, various nanoparticles were dispersed into transition in grinding of zirconia ceramics with different lubrication
traditional cutting fluids based on the principle of heat transfer conditions. This greatly improves the accuracy of the grinding force
enhancement by the addition of solids (Choi and Eastman, 1995). prediction model and the surface topography prediction model for
Meanwhile, the investigation and optimization of related parame- zirconia grinding under different lubrication conditions.
ters were conducted to improve the machining characteristics. Meanwhile, several attempts have been made by researchers to
Many studies have been conducted to understand the optimize machining parameters in order to obtain excellent all-
machining characteristics of nanoparticle-dispersed cutting fluids. round machining performance. Cetin et al. (2011) studied the
Most such studies involve the dispersion of nanoparticle additives machining characteristics of vegetable-based cutting fluids and
into a cutting fluid, followed by testing. These nanoparticle addi- evaluated the effects of cutting parameters on machining charac-
tives are of four main types, mainly, metals, metal oxides, metal teristics in the turning of AISI 304L. They used Taguchi's mixed-
sulfides, and carbide (Sharma et al., 2016; Wang et al., 2017a). level parameter design to design the experiments, applied regres-
Pashmforoush and Delir Bagherinia (2018) dispersed copper (Cu) sion analyses to predict the surface roughness, and cutting and feed
nanoparticles into traditional cutting fluid to prepare a copper forces, and finally, used ANOVA to determine the effects of the
nanofluid and evaluated the machining performance in the MQL cutting parameters on the machining characteristics. They
grinding of Inconel 738 alloy. They concluded that the copper concluded that the feed rate and cutting depth were more influ-
nanofluid could improve the wheel loading and surface roughness ential than the cutting fluids and spindle speed, and that sun-
as compared to the dry condition and conventional fluid. Chetan flower- and canola-based cutting fluids performed better than the
et al. (2016) studied the machining performances of Nimonic in others. Kuram et al. (2013) used the D-optimal design to optimize
MQL turning with alumina (Al2O3) and silver (Ag) nanofluids. They the cutting fluids and cutting parameters in the end milling of AISI
concluded that the alumina nanofluid could reduce the cutting 304. Mathematical models for the process responses were devel-
forces, tool wear and chip curing due to the smaller contact angle, oped and conclusions regarding the effects of the milling parame-
greater spreadability, and small droplet size, while the silver ters and cutting fluid types on the milling performances criteria are
nanofluid could provide good surface finish and reduce abrasive obtained. Sarıkaya and Güllü (2014) used Taguchi design and
wear due to the nano-ball bearing effect. Wang et al. (2017b) response surface methodology to improve the machining perfor-
evaluated the lubricating properties of the wheel/workpiece mances in CNC turning under MQL. The optimal parameter com-
interface in MQL grinding with an Al2O3 nanofluid. They pointed bination and corresponding mathematical models are achieved,
out that the appropriate concentration of the Al2O3 nanofluid could and the machining performances were greatly improved. Nam et al.
provide good lubricating properties, and the nanofluid could form a (2015) used the response surface methodology and genetic algo-
lubricating oil film and reduce the friction significantly. Najiha et al. rithm to optimize the environmentally benign nanofluid-based
(2016) investigated the tool wear mechanisms in the MQL milling MQL micro-drilling process. The optimal values of the process
of AA6061-T6 alloy using TiO2 nanofluid. They concluded that the factors were obtained and the regression models of drilling torques
adhesion of the work material contributed most to the tool damage and thrust forces were demonstrated. Ooi et al. (2015) used a fuzzy
while the major benefit of the TiO2 nanofluid MQL was the edge logic-based approach to achieve the optimal machining parameters
integrity, and the composition with 2.5% TiO2 nanofluid was iden- in the milling of Al-6061-T6 alloy. Compared with the experimental
tified as most suitable to control the tool damage. Zhang et al. results, the proposed fuzzy model exhibited a high degree of reli-
(2015) evaluated the lubricating properties of MoS2 nanoparticles ability. Sarıkaya and Güllü (2015) performed the multi-response
in MQL grinding with different types of vegetable oils as the base optimization of MQL parameters by using a Taguchi-based gray
oil. They concluded that the MoS2-dispersed palm oil-based relational analysis in the MQL turning of Haynes 25 alloy. The best
nanofluid could produce the best lubricating property due to the combination of a vegetable-based cutting fluid, fluid flow rate of
high content of saturated fatty acids and considerable film-forming 180 mL/h, and cutting speed of 30 m/min were achieved. Sarıkaya
property of carboxyl groups in palm oil. Sharma et al. (2015) studied et al. (2016) studied the machinability of cobalt-based Haynes 25
the machining performances of AISI D2 steel in MQL turning with superalloy in CNC turning with three cooling methods. The Tagu-
carbon nanotube (CNT) nanofluid. They stated that MQL turning chi's signal to noise ratio (S/N) analysis was adopted to obtain the
with CNT nanofluid outperformed the MQL with conventional optimal parameter combination, and the machinability of the ma-
cutting fluid with regard to cutting zone temperature and surface terial was significantly improved. Mia et al. (2018a) optimized the
roughness. This is mainly because CNTs could increase the thermal machining parameters by using the signal-to-noise ratio-based
conductivity and heat carrying capacity of the fluid. Nam et al. optimization in MQL turning. Optimum combinations for the
M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522 1511

minimum surface roughness and maximum material removal rate, milling parameters were kept constant. The relatively reasonable
the minimum principal flank wear, and the lowest auxiliary flank levels of the orthogonal experiment factor were determined by a
wear were obtained. Viswanathan et al. (2018) optimized the per- series of single factor experiments before the orthogonal experi-
formance characteristics of magnesium AZ91D alloy by using the ment. The factors and levels of the orthogonal experiment are
Taguchi-based gray relational analysis in MQL turning. The opti- shown in Table 1. Based on Taguchi design method, the L16 ð43 Þ
mum combination of 90 m/min cutting speed, 0.1 mm/rev feed rate, orthogonal experiment table was selected to design the experi-
and 0.5 mm depth of cut was obtained as a result. Rapeti et al. ment. A total of 16 experiments were performed, as shown in
(2018) evaluated machining performances of AISI 1040 steel in Table 8. The main milling parameters are the spindle rotation speed
MQL turning with vegetable oil based nano cutting fluids. The (N ¼ 1062 rpm), feed rate per tooth (fz ¼ 0.02 mm/z), radial im-
optimal parameter combination was obtained by gray relational mersion depth (ae ¼ 0.4 mm) and axial immersion depth
analysis, and the machining performances were significantly (ap ¼ 0.6 mm), as shown in Table 2.
improved.
However, although many researchers have attempted the 2.2. Preparation of graphene-dispersed cutting fluid
dispersion of a variety of nanoparticles into cutting fluids, and
studied their machining characteristics and parameter optimiza- In this study, few-layer graphene was chosen as the nanoparticle
tion, only a few have attempted to incorporate graphene nano- additive to prepare graphene-dispersed cutting fluids. Graphene is
particles in vegetable-oil-based cutting fluids. Graphene was the the first truly two-dimensional, man-made material. The few-layer
first-ever two-dimensional material to be discovered and is the graphene used in this study had a thickness of less than 5 nm and
carbon material with the highest known thermal conductivity. The an average diameter of 10 mm. The structure diagram of few-layer
thermal conductivity of graphene is as high as 5300 W/(m K) and it graphene is shown in Fig. 1. Meanwhile, the vegetable-oil-based
has very good lubrication properties (Balandin et al., 2008; Berman cutting fluid LB2000 (supplied by ITW Rocol North America Co.,
et al., 2014). Overall, graphene is a promising additive for cutting Ltd.) was selected as the base oil for the graphene-dispersed cutting
fluids to improve cooling and lubrication performances. fluids thanks to its low environmental impact, good biodegrad-
To address this gap, in this study, graphene nanoparticles were ability, and non-toxicity. The properties of graphene and LB2000
used as additive and dispersed into a vegetable-oil-based cutting are shown in Tables 3 and 4, respectively. In addition, the high-
fluid to improve the cooling and lubrication properties. The L16 ð43 Þ precision electronic scale was selected for weighing the graphene
orthogonal experiment table was selected to design the experiment nanoparticle and vegetable-oil-based cutting fluid with a range of
based on the Taguchi design method. The MQL parameters (the 200 g and an accuracy of 0.001 g, as shown in Fig. 2.
cutting fluid concentration, cutting fluid flow rate, and gas pres- The graphene-dispersed cutting fluids were prepared by
sure) were selected as the control factors, and the milling charac- dispersing a certain amount of graphene nanoparticles in the
teristics (the milling force, milling temperature, surface micro- LB2000 by using a two-step method, i.e., the preparation of the
hardness, and surface roughness) were selected as the response nanoparticles and nanoparticle-dispersed cutting fluids are not
characteristics. Meanwhile, the effect of the MQL parameters on the performed at the same time. During the dispersion process, the
four milling characteristics were evaluated and analyzed using the graphene-dispersed cutting fluids were first stirred using a mag-
signal-to-noise ratio (SNR) analysis based on the Taguchi method. netic stirrer (for 30 min) and then dispersed using an ultrasonic
Then, the unified optimal MQL parameter combination for the four dispersion instrument (40 kHz, 80 W) (for 1 h), as depicted in Fig. 2.
milling characteristics was obtained by the integrated Taguchi- This dispersion process was repeated several times until the gra-
Principal component analysis (PCA)-Gray relational analysis phene nanoparticles were uniformly dispersed in the LB2000. The
(GRA) optimization method. In addition, the influencing mecha- graphene-dispersed cutting fluids were stable and no agglomera-
nisms of the MQL parameters on the milling characteristics were tion was observed during the entire experiment process.
discussed. The findings of this study are expected to indicate the
feasibility of the application of graphene additive in cutting fluids
and provide certain technical references for industrial application. 2.3. Machine tool and MQL system

2. Experiment and method In this study, a four-axis vertical milling machining center
(TH5650, Shenyang machine tool (Group) Co., Ltd.) shown in
In this study, a series of milling experiments of the TC4 alloy Fig. 3(c) was adopted to carry out a series of experiments. The main
under cooling/lubrication conditions with different MQL parame- technical parameters are: the numerical control system of FANUC
ters were conducted. Meanwhile, the effects of the different MQL 0i-MB, maximum spindle speed of 6000 rpm, maximum spindle
parameters on the milling force, milling temperature, surface power of 35 kW, and working range (X  Y  Z) of
micro-hardness, and surface roughness were evaluated. In addition, 850 mm  500 mm  630 mm.
the optimal MQL parameter combination was obtained using the In addition, an Accu-Lube MQL system (supplied by ITW Rocol
multi-objective optimization method. North America Co., Ltd.) was used to transport the graphene-
dispersed cutting fluid to the milling zone. This MQL system is
2.1. Experimental design comprised of an air pump, filter, pulse generator, pulse pump, oil
tank, and oil-gas mixing tube, as depicted in Fig. 3(a). Compressed
In order to evaluate the effects of the different MQL parameters
on the milling force, milling temperature, surface micro-hardness, Table 1
and surface roughness, the orthogonal experiments were Control factors and levels of the orthogonal experiment.
designed based on the Taguchi design method.
Control factors Symbol Level
In this study, the MQL parameters (the cutting fluid concen-
Levels 1 2 3 4
tration, cutting fluid flow rate, and gas pressure) were considered
as the control factors, and the milling characteristics (the milling Concentration (wt.%) W 4 0.05 0.1 0.15 0.2
force, milling temperature, surface micro-hardness, and surface Flow rate (mL/h) Q 4 30 60 90 120
Gas pressure (MPa) P 4 0.4 0.5 0.6 0.7
roughness) were considered as the response characteristics; the
1512 M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522

Table 2
Main milling parameters of the orthogonal experiment.

Milling parameters Value

Milling style Up milling


Spindle rotation speed N (rpm) 1062
Feed rate per tooth fz (mm/z) 0.02
Radial immersion depth ae (mm) 0.4
Axial immersion depth ap (mm) 0.6

Table 3
Properties of graphene.

Property Average layer thickness (nm) Mean diameter (mm) Number of layers Specific surface area (m2/g) Appearance

Value <5 10 1e5 360e450 Black powder

Table 4
Properties of LB2000

Property Density (at 20  C) (g/cm3) Flash point ( C) Pour point ( C) Appearance

Value 0.92 320 20 Dark blue fluid

Table 5
Chemical composition of TC4.

Element Ti Fe C N H O Al V

Composition (wt.%) Balance 0.30 0.10 0.05 0.015 0.20 5.5e6.8 3.5e4.5

Table 6
Mechanical properties of TC4.

Property Tensile strength (MPa) Yield strength (MPa) Elongation (%) Poison ratio Modulus of elasticity (GPa) Hardness (HV)

Value 990 830 14 0.342 114 312

Table 7 used to remove the heat affected layer. The chemical composition
Details of milling tool. and mechanical properties of the TC4 alloy are shown in Tables 5
Property Value and 6, respectively.
Diameter (mm) 6
In addition, TC4 is a difficult-to-machine material due to its low
Total length (mm) 50 thermal conductivity, large friction coefficient, high chemical ac-
Cutting edge length (mm) 16 tivity and low elasticity modulus. The carbide tool is ideal for cut-
Number of flutes 4 ting TC4 alloy due to its high hot hardness, good toughness, good
Coating TiAlN
wear resistance, high thermal conductivity, low chemical activity
Matrix material Ultra-fine tungsten steel
Helix angle (degree) 45 and good impact resistance (A. Ghani et al., 2013). Meanwhile, as a
Rake (degree) 8 widely used coating material, the TiAlN coating has many advan-
Relief angle (degree) 14 tages, such as high hardness, high oxidation temperature, good
thermal hardness, strong adhesion, low friction coefficient, low
thermal conductivity and so on. It is especially suitable for cutting
air at a certain pressure is mixed with a certain amount of cutting difficult-to-machine materials such as titanium alloy. Therefore, the
fluid discharged by the pulse pump and then injected into the TiAlN coated milling tool is adopted in this work. A solid carbide
cutting zone. The pressure of the compressed air and the flow rate end-mill tool (GM-4E-D6.0, Zhuzhou Cemented Carbide Cutting
of the graphene-dispersed cutting fluid are regulated by the air Tools Co., Ltd.) was adopted for use in the experiments. The char-
pump and pulse pump, respectively. In addition, the nozzle position acteristic parameters of the tool are: a diameter of 6 mm, length of
parameters are shown in Fig. 3(b), with the nozzle distance (L) of 50 mm, spiral angle of 45-degree, and teeth number of four. The
25 mm and the nozzle angle (q) of 45-degree. matrix material of the milling tool was ultra-fine tungsten steel
while the coating material was TiAlN. Detailed information
2.4. Work material and cutting tool regarding the milling tool is shown in Table 7. In addition, the
suspension length of milling tool in each experiment was set to
The workpiece material used in this study was TC4 alloy. The 30 mm.
dimension of the workpiece was 80 mm (length)  30 mm
(width)  15 mm (height). Before the experiments, the oxide layer 2.5. Measurement of the milling characteristics
and the heat affected layer on the surface of the workpiece needed
to be removed. Grinding was used to remove the oxide layer and a In order to evaluate the effects of the MQL parameters on milling
solution containing 20% nitric acid and 5% hydrofluoric acid was characteristics, the milling force, milling temperature, surface
M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522 1513

Table 8
Orthogonal experiment arrangements and results.

No. MQL parameter Milling characteristic SNR of milling characteristic

W Q P Fcmax Tmax MH Ra Fcmax Tmax MH Ra

1 0.05 30 0.4 38.928 242.771 398.531 0.492 31.805 47.704 52.009 6.161
2 0.05 60 0.5 31.877 225.856 347.433 0.405 30.070 47.077 50.817 7.851
3 0.05 90 0.6 32.870 194.764 359.126 0.411 30.336 45.790 51.105 7.723
4 0.05 120 0.7 35.353 235.988 391.153 0.496 30.969 47.458 51.847 6.090
5 0.10 30 0.5 28.004 181.781 331.917 0.342 28.944 45.191 50.421 9.319
6 0.10 60 0.4 28.699 193.235 346.852 0.365 29.157 45.722 50.803 8.754
7 0.10 90 0.7 27.607 173.341 337.214 0.339 28.820 44.778 50.558 9.396
8 0.10 120 0.6 28.203 197.759 369.231 0.357 29.006 45.923 51.346 8.947
9 0.15 30 0.6 28.699 199.783 352.528 0.352 29.157 46.011 50.944 9.069
10 0.15 60 0.7 29.097 195.562 349.539 0.348 29.277 45.826 50.870 9.168
11 0.15 90 0.4 33.565 208.257 389.762 0.408 30.518 46.372 51.816 7.787
12 0.15 120 0.5 32.274 229.212 375.922 0.397 30.177 47.205 51.502 8.024
13 0.20 30 0.7 33.863 231.461 396.551 0.379 30.595 47.290 51.966 8.427
14 0.20 60 0.6 28.898 194.267 358.286 0.346 29.217 45.768 51.085 9.218
15 0.20 90 0.5 36.247 217.163 372.529 0.413 31.185 46.736 51.423 7.681
16 0.20 120 0.4 39.623 238.246 401.672 0.508 31.959 47.541 52.077 5.883

Fig. 1. Structure diagram of few-layer graphene.

Fig. 3. Schematic diagram of the milling experiment ((a) MQL system, (b) experiment
schematic, and (c) measurement system).

introduced, which is the maximum value of the resultant force


(total cutting force) Fc . The resultant force Fc is computed using Eq.
(1).
Fig. 2. Preparation process of the graphene-dispersed cutting fluids ((a) few-layer qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
graphene and high-precision electronic scale, (b) pure LB2000, (c) magnetic stirrer, Fc ¼ F 2x þ F 2y þ F 2z (1)
(d) ultrasonic dispersion instrument, and (e) graphene-dispersed cutting fluid).

where Fc is the resultant force, Fx , Fy and Fz are the measuring


milling forces along the x-axis, y-axis, and z-axis, respectively.
micro-hardness and surface roughness were measured under
cooling/lubrication conditions with different MQL parameters. In
addition, in order to reduce the error, each experiment was 2.5.2. Milling temperature
repeated 30 times. Milling temperature is a parameter that significantly affects the
surface integrity and service performance of the workpiece; lower
milling temperatures indicate better milling performance. The
2.5.1. Milling force measurement of milling temperature under cooling/lubrication
Milling force is a vital milling characteristic that directly affects conditions with different MQL parameters was performed using the
surface quality and machining accuracy. As shown in Fig. 3, a 3D temperature-measuring system shown in Fig. 3(c), which includes
force-measuring system (9257B, Kistler Co., Ltd) was used to a commercial K-type thermocouple (TT-K-36-SLE, OMEGA) with a
measure the milling force under cooling/lubrication conditions diameter of 0.30 mm, a multichannel signal processing system, and
with different MQL parameters along the x-axis, y-axis, and z-axis. a PC. The surface temperature measurement strategy is depicted in
This system included a three-component dynamometer, a multi- Fig. 4. The K-type thermocouple was inserted into a 0.30-mm-
channel charge amplifier, an A/D data acquisition card, and a PC. diameter hole in the workpiece to measure the surface tempera-
The sample frequency of the milling force was set at 7000 Hz. In ture. The sample frequency of milling temperature was set at
order to comprehensively evaluate the cooling/lubrication condi- 20 KHz. As shown in Fig. 4, the K-type thermocouple was clamped
tion on the milling force, the maximum resultant force Fcmax is by two identical workpieces. Meanwhile, the positioning pins were
1514 M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522

Fig. 4. Schematic diagram of surface temperature measurement strategy.

used to position the two workpieces to reduce the positioning error.


In addition, the 0.3-mm-diameter thermocouple mounting hole is
assembled from two 0.3-mm-diameter circular slots. As shown in
Fig. 4, the 0.3-mm-diameter circular slot is made by milling with
the micro ball end mill (R0.15 mm).

2.5.3. Surface micro-hardness


Surface micro-hardness is a parameter that significantly affects
the service reliability of the workpiece. The work hardening of the
workpiece surface during the milling process is generally uneven,
and this uneven work hardening greatly reduces the service reli-
ability of the workpiece. Therefore, smaller values of surface micro-
hardness is generally desirable. In this study, the surface micro-
hardness under cooling/lubrication conditions with different MQL
parameters were measured using the micro-hardness tester
(TUKON1102, Wilson) shown in Fig. 5(a) with an indenter load of 20
gf and a holding time of 10 s.

2.5.4. Surface roughness


Surface roughness of the workpiece is an important indicator for
evaluating milling machining characteristics, and a smaller value of
surface roughness is indicative of better service performance of the
workpiece. Therefore, the smaller surface roughness is desirable. In Fig. 6. Flowchart for integrated Taguchi-PCA-GRA optimization method.
this study, the surface roughness of the workpiece under cooling/
lubrication conditions with different MQL parameters were
measured by a laser confocal microscope (LEXT OLS4100, Japan GRA optimization method was given. Firstly, the orthogonal ex-
Olympus Corporation), as shown in Fig. 5(b). periments based on Taguchi method were performed and the
milling characteristics were measured. Meanwhile, SNR of the
2.6. Optimization method milling characteristics was calculated and used to evaluate the ef-
fects of the MQL parameters on the milling characteristics. Then,
In this study, the MQL parameters (the cutting fluid concen- the PCA-GRA method was conducted and applied to obtain the
tration, cutting fluid flow rate, and gas pressure) were considered optimal MQL parameter combination. Finally, the verification ex-
as the control factors and the milling characteristics (the milling periments were conducted to verity the feasibility of the optimal
force, milling temperature, surface micro-hardness, and surface MQL parameter combination. Overall, this optimization process
roughness) were considered as the response characteristics. In or- could be divided into three phases.
der to comprehensively evaluate the effects of the MQL parameters
on the milling characteristics and obtain the optimal MQL param- 2.6.1. Phase I: Taguchi-method analysis
eter combination, a multi-objective optimization was conducted. The orthogonal experiments were designed based on the
As shown in Fig. 6, the flow chart of integrated Taguchi-PCA- Taguchi method. Meanwhile, SNR was used to evaluate the effects
of the MQL parameters on milling characteristics. In addition, the
significance of the MQL parameters for the milling characteristics
was evaluated by the ANOVA of the SNR of the milling
characteristics.
Step 1: Design the orthogonal experiments and measure the
milling characteristics.
Step 2: Calculate the SNR of the experiment results.
The SNR of the milling characteristics is an indicator of the
robustness of the milling characteristics and the higher the SNR
Fig. 5. (a) Micro-hardness tester and (b) laser confocal microscope. value, the better the setting of the MQL parameters will be.
M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522 1515

Generally, there are three types of SNR, namely, smaller-the-better 2 3


(STB), nominal-the-better (NTB), and larger-the-better (LTB). In this ε1 ð1Þ ε1 ð2Þ / ε1 ðnÞ
6 ε2 ð1Þ ε2 ð2Þ / ε2 ðnÞ 7
study, the desired objective is to obtain smaller milling force, ε¼6
4 «
7 (5)
smaller milling temperature, smaller surface micro-hardness, and « « « 5
smaller surface roughness. Therefore, the SNR of the milling char- εm ð1Þ εm ð2Þ / εm ðnÞ
acteristics all belongs to the STB class of the SNR as shown in Eq. (2).
where ε is the GRC matrix; m ¼ 16, n ¼ 4.
8 !
>
>
> 1X n
(b) Calculate the correlation coefficient array
>
> 10lg y2 STB
>
> n i¼1 i    
>
>
>
> ! Covðεi ðkÞ; εi ðlÞÞ j ¼ 1; 2; /; n;
>
< Rjl ¼ (6)
1X n
sεi ðkÞ  sεi ðlÞ l ¼ 1; 2; /; n
SNR ¼ 10lg ðyi  mÞ2 NTB (2)
>
> n i¼1
>
>
>
> ! where n ¼ 4; Covðεi ðjÞ; εi ðlÞÞ is the covariance of εi ðjÞ and εi ðlÞ; sεi ðjÞ
>
>
>
> 1X n
1 and sεi ðlÞ are the standard deviations of εi ðjÞ and εi ðlÞ, respectively.
>
: 10lg
>
n i¼1 y2i
LTB
(c) Calculate the eigenvalues and eigenvectors
where n is the total number of experiments, yi is the result of the i-
ðR  lk I m ÞV ik ¼ 0 ðk ¼ 1; 2; /; nÞ (7)
th experiment, and m is the target value of yi .
Pn
where lk is the eigenvalue and k¼1 lk ¼ n; V ik is the eigenvector
and V ik ¼ ½ak1 ak2 / akn T ; I m is the unit matrix.

(d) Calculate the weight of GRC


2.6.2. Phase II: multi-objective optimization based on the PCA-GRA
method
The k-th principal component is calculated as shown in Eq. (8).
Based on the above SNR data of the milling characteristics, the
principal component analysis (PCA) and gray relational analysis X
n
(GRA) are integrated to transform multiple responses into a single Ymk ¼ εm ðiÞ Vik (8)
response. The PCA is used to calculate the weight of each gray i¼1
relational coefficient (GRC). The principal components are arranged in descending order
Step 3: Preprocess data. with respect to the variance; hence, the first principal component
Data preprocessing in the GRA involves normalizing the above Ym1 accounts for the largest variance in the data and is generally
SNR data. The data preprocessing also belongs to three classes, used to replace all the response characteristics. Therefore, the rate
namely the STB, NTB, and LTB. Since the SNR is to be as large as of contribution each response characteristic of the first principal
possible, the SNR of the milling characteristics belongs to the LTB component is generally used as the weight of the GRC. The weight
class. As shown in Eq. (3), the above SNR data is normalized. of GRC is calculated, as shown in Eq. (9).
8
> maxyi ðkÞ  yi ðkÞ ai ¼ V 2i1 (9)
>
> STB
>
> maxy i ðkÞ  minyi ðkÞ
>
>
>
< where ai is the weight of GRC; V i1 is the eigenvector of the first
maxjyi ðkÞ  aðkÞj  jyi ðkÞ  aðkÞj
xi ðkÞ ¼ NTB (3) principal component.
>
> maxjy i ðkÞ  aðkÞj  minjyi ðkÞ  aðkÞj
>
> Step 6: Calculate the gray relational grade (GRG).
>
>
>
> yi ðkÞ  minyi ðkÞ The GRG is the weighted sum of the GRC, as shown in Eq. (10).
: LTB
maxyi ðkÞ  minyi ðkÞ
X
n

where i ¼ 1  m, k ¼ 1  n, xi ðkÞ2½0; 1; yi ðkÞ is the original SNR,


g¼ ai εi (10)
i¼1
xi ðkÞ is the normalized SNR, and aðkÞ is the target value of yi ðkÞ.
Step 4: Calculate the GRC. Pn
where g is the GRG; ai is the weight of GRC and i¼1 ai ¼ 1.
After the data preprocessing, the normalized SNR is used to
calculate GRC. Per Eq. (4), the GRC is calculated as:
2.6.3. Phase III: verification experiment
Step 7: The verification experiment is conducted to verify the
Dmin þ z$Dmax
εi ðkÞ ¼ (4) feasibility of the optimal MQL parameter combination.
D0i ðkÞ þ z$Dmax

where εi ðkÞ is GRC; D0i ðkÞ is the absolute value of the deviation 3. Results and discussions
between x0 ðkÞ and xi ðkÞ and D0i ðkÞ ¼ jx0 ðkÞ  xi ðkÞj; x0 ðkÞ is the
ideal value of xi ðkÞ and x0 ðkÞ ¼ 1; Dmin is the minimum value of In this section, the results obtained from experiments under
D0i ðkÞ and Dmin ¼ min$min$D0i ðkÞ; Dmax is the maximum value of cooling/lubrication conditions with different MQL parameters are
ci ck
D0i ðkÞ and Dmax ¼ max$max $D0i ðkÞ; z is the distinguishing coeffi- summarized. Firstly, based on the Taguchi-method analysis, the
ci ck
cient between zero and one and is normally selected as 0.5. effects of the MQL parameters (the cutting fluid concentration,
Step 5: Calculate the weight of GRC based on PCA. cutting fluid flow rate, and gas pressure) in terms of the milling
Based on the above GRC data, the PCA is used to calculate the force, milling temperature, surface micro-hardness, and surface
weight of GRC. roughness were evaluated. Then, based on the PCA-GRA method
analysis, multi-objective optimization was performed and the
(a) Establish the GRC matrix optimal MQL parameter combination was obtained. Finally, the
1516 M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522

Fig. 8. Measurement results of the milling characteristics ((a) milling force, (b) milling
temperature, (c) surface micro-hardness, and (d) surface roughness).

cutting fluid flow rate, and gas pressure to the milling force is
55.39%, 15.90%, and 26.20%, respectively. Hence, for the milling
force, the most significant MQL parameter is the cutting fluid
concentration W, followed by the gas pressure P, whereas the least
influential parameter is the cutting fluid flow rate Q. As shown in
Table 10 and Fig. 9(a), when the cutting fluid concentration was
increased from 0.05 wt% to 0.2 wt%, the SNR responses of the
milling force were first increased and then decreased, and reached
the maximum when the cutting fluid concentration was 0.1 wt%. As
shown in Fig. 10, this is because the appropriate amount of gra-
phene additives could enhance the anti-friction and load-bearing
capacities of the oil film formed in the milling zone. However, if
the cutting fluid concentration is too low, the enhancement effect
will be weakened due to insufficient lubrication. Meanwhile, when
the cutting fluid concentration is too high, the relatively large
Fig. 7. Flowchart for this work.
quantity of graphene additives could easily block the milling zone
and hinder the formation of the oil film, resulting in the weakening
feasibility of the optimal MQL parameter combination was verified of the enhancement effect due to insufficient lubrication. When the
by using the verification experiments. In addition, the causative cutting fluid flow rate was increased from 30 mL/h to 120 mL/h, the
mechanisms of the effects of the MQL parameters on the milling SNR responses of the maximum resultant force first increased and
characteristics were discussed. The flowchart for this work is then decreased, reaching the maximum when the flow rate was
shown in Fig. 7. 60 mL/h. This may be due to insufficient lubrication in the milling
zone when the cutting fluid flow rate is too small. Meanwhile, a
too-large cutting fluid flow rate could reduce the jet velocity of the
3.1. Evaluation of milling characteristics based on the Taguchi
cutting fluid, decrease the ability of the cutting fluid jet to penetrate
method analysis
the gas barrier layer around the milling tool, and cause insufficient
lubrication. When the gas pressure was increased from 0.4 MPa to
The orthogonal experiments based on Taguchi method were
0.7 MPa, the SNR responses of the maximum resultant force Fcmax
performed and the experiment results were obtained. The
also first increased and then decreased, reaching the maximum
orthogonal experiment arrangements are tabulated in Table 8 and
when the gas pressure was 0.6 MPa. This is because the ability of
the experiment results are summarized in Fig. 8 and Table 8.
the cutting fluid jet to penetrate the gas barrier layer around the
Meanwhile, the SNR analysis based on the Taguchi method was also
milling tool is poor when the gas pressure is less than 0.6 MPa, thus
performed. The SNR of the milling characteristics are calculated by
resulting in insufficient lubrication. Meanwhile, when the gas
Eq. (2) and are also given in Table 8. In addition, the ANOVA of the
pressure is more than 0.6 MPa, the flyoff of cutting fluid jet grad-
SNR of the milling characteristics was obtained as shown in Table 9.
ually increases as the gas pressure increases, which also could
For the milling force, the ANOVA of the SNR of the maximum
result in insufficient lubrication. As shown in Fig. 9(a), it was
resultant force Fcmax is summarized in Table 9. According to the F
observed that the optimal MQL parameter combination for the
distribution table, F0.01(3,6) ¼ 9.7795, F0.05(3,6) ¼ 4.76;
minimum Fcmax can be designated as W2Q2P3, namely, 0.1 wt%
F(W) > F0.01(3,6), F(Q) > F0.01(3,6), F(P) > F0.01(3,6), implying that
cutting fluid concentration (level 2), 60 mL/h cutting fluid flow rate
the cutting fluid concentration, cutting fluid flow rate, and gas
(level 2), and 0.6 MPa gas pressure (level 3).
pressure all have highly significant effects on the milling force.
For the milling temperature, the ANOVA of the SNR of the
Meanwhile, the contribution rate of the cutting fluid concentration,
M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522 1517

Table 9
The ANOVA table of the SNR of the milling characteristics.

Milling characteristic W Q P Error Total

Fcmax DF 3 3 3 6 15
Seq. SS 8.9578 2.5704 4.2360 0.4067 16.1708
Adj. MS 2.9859 0.8568 1.4120 0.0678
F 44.06 12.64 20.83
P 0.000 0.005 0.001
Contribution 55.39% 15.90% 26.20%
Tmax DF 3 3 3 6 15
Seq. SS 6.2078 2.9739 1.9741 1.0072 12.1630
Adj. MS 2.0693 0.9913 0.6580 0.1679
F 12.33 5.91 3.92
P 0.006 0.032 0.073
Contribution 51.04% 24.45% 16.23%
MH DF 3 3 3 6 15
Seq. SS 1.6123 1.3026 0.9629 0.4939 4.3717
Adj. MS 0.5374 0.4342 0.3210 0.0823
F 6.53 5.27 3.90
P 0.026 0.040 0.074
Contribution 36.88% 29.80% 22.03%
Ra DF 3 3 3 6 15
Seq. SS 10.2981 4.7569 5.4471 0.4469 20.9489
Adj. MS 3.4327 1.5856 1.8157 0.0745
F 46.09 21.29 24.38
P 0.000 0.001 0.001
Contribution 49.16% 22.71% 26.00%

Table 10
Mean SNR response table of the milling characteristics.

Milling characteristic MQL parameter

W Q P

Fcmax Level 1 30.795 30.125 30.860


Level 2 28.982 29.430 30.094
Level 3 29.782 30.215 29.429
Level 4 30.739 30.528 29.915
Delta 1.813 1.097 1.431
Rank 1 3 2
Tmax Level 1 47.007 46.549 46.835
Level 2 45.403 46.098 46.552
Level 3 46.353 45.919 45.873
Level 4 46.833 47.031 46.338
Delta 1.604 1.112 0.962
Rank 1 2 3
MH Level 1 51.445 51.335 51.676
Level 2 50.782 50.894 51.041
Level 3 51.283 51.226 51.120
Level 4 51.638 51.693 51.310
Delta 0.856 0.799 0.636
Rank 1 2 3
Ra Level 1 6.956 8.244 7.146
Level 2 9.104 8.748 8.219
Level 3 8.512 8.147 8.739
Level 4 7.802 7.236 8.271
Delta 2.148 1.512 1.593
Rank 1 3 2

milling temperature is shown in Table 9. From the F distribution gas pressure P. As shown in Table 10 and Fig. 9(b), when the cutting
table, F(W) > F0.01(3,6), and hence, the cutting fluid concentration fluid concentration was increased from 0.05 wt% to 0.2 wt%, the
has a highly significant effect on the milling temperature; SNR responses of the milling temperature first increased and then
F0.05(3,6) < F(Q) < F0.01(3,6), indicating that the cutting fluid flow decreased, and reached the maximum when the cutting fluid
rate has a significant effect on the milling temperature; and concentration was 0.1 wt%. This is mainly because the appropriate
F(P) < F0.05(3,6), indicating that the effect of the gas pressure on the amount of graphene additives could help enhance the cooling ca-
milling temperature is not significant. Meanwhile, the contribution pacity of the oil film formed in the milling zone based on the
rate of the cutting fluid concentration, cutting fluid flow rate, and principle that the presence of solid particles enhance the heat
gas pressure to the milling temperature are 51.04%, 24.45%, and transfer. However, when the cutting fluid concentration is too low,
16.23%, respectively. Therefore, for the milling temperature, the the enhancement effect will be weakened due to insufficient
most significant MQL parameter was the cutting fluid concentra- cooling. Meanwhile, when the cutting fluid concentration is too
tion W, followed by the cutting fluid flow rate Q, and the least is the high, the relatively large quantity of graphene additives could easily
1518 M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522

increase in gas pressure, which could also contribute to insufficient


cooling. From to Fig. 9(b), it can be seen that the optimal MQL
parameter combination for the minimum milling temperature is
W2Q3P3, which stands for 0.1 wt% cutting fluid concentration
(level 2), 90 mL/h cutting fluid flow rate (level 3), and 0.6 MPa gas
pressure (level 3).
With regard to the surface micro-hardness, the ANOVA of the
SNR of the surface micro-hardness is shown in Table 9. According to
the F distribution table, F0.05(3,6) < F(W) < F0.01(3,6), which implies
that the cutting fluid concentration has a significant effect on the
surface micro-hardness; F0.05(3,6) < F(Q) < F0.01(3,6), indicating
that the flow rate of the cutting fluid also has a significant effect on
the surface micro-hardness; F(P) < F0.05(3,6), which leads us to
conclude that the effect of the gas pressure on the surface micro-
hardness is not significant. Meanwhile, the contribution rate of
the cutting fluid concentration, cutting fluid flow rate, and gas
pressure to the surface micro-hardness is 36.88%, 29.80%, and
22.03%, respectively. Hence, for the surface micro-hardness, the
most significant MQL parameter is the cutting fluid concentration
W, followed by the cutting fluid flow rate Q, while the least influ-
Fig. 9. Mean SNR response graph of the milling characteristics ((a) milling force, (b)
milling temperature, (c) surface micro-hardness, and (d) surface roughness).
ential is the gas pressure P. As can be seen from Table 10 and
Fig. 9(c), when the cutting fluid concentration was increased from
0.05 wt% to 0.2 wt%, the SNR responses of the surface micro-
hardness first increased and then decreased, reaching the
maximum when the cutting fluid concentration was 0.1 wt %. When
the cutting fluid flow rate increased from 30 mL/h to 120 mL/h, the
SNR responses of the surface micro-hardness first increased and
then decreased, reaching the maximum when the cutting fluid flow
rate was 60 mL/h. When the gas pressure was increased from
0.4 MPa to 0.7 MPa, the SNR responses of the surface micro-
hardness also first increased and then decreased, with the
maximum value corresponding to a gas pressure of 0.5 MPa. Work
hardening in the milling process is mainly caused by the combined
effects of the friction, extrusion, and milling heat generated in the
tool-workpiece interface. Among them, the friction and extrusion
could improve the work hardening, while milling heat could
weaken the work hardening. From the above analysis of the milling
force and milling temperature, it is evident that inappropriate MQL
parameters could cause insufficient lubrication and cooling. Thus,
insufficient lubrication could increase work hardening whereas
insufficient cooling could reduce it. When the MQL parameters
values are very small, the strengthening effects are far greater than
Fig. 10. Schematic diagram of milling under cooling/lubrication conditions with
the weakening effects while large MQL parameters result in the
different cutting fluid concentrations ((a) higher cutting fluid concentration, (b)
appropriate cutting fluid concentration, and (c) lower cutting fluid concentration). strengthening effects being far greater than the weakening effects.
From Fig. 9(c), it can be seen that the optimal MQL parameter
combination for the minimum surface micro-hardness was
block the milling zone and hinder the formation of the oil film, W2Q2P2, which stands for 0.1 wt% cutting fluid concentration
thereby weakening the enhancement effect due to insufficient (level 2), 60 mL/h cutting fluid flow rate (level 2), and 0.5 MPa gas
cooling. When the cutting fluid flow rate was increased from pressure (level 2).
30 mL/h to 120 mL/h, the SNR responses of the milling temperature The ANOVA of the SNR of the surface roughness is shown in
were first increased and then decreased, and reached the Table 9. According to the F distribution table, F(W) > F0.01(3,6),
maximum when the cutting fluid flow rate was 90 mL/h. This may F(Q) > F0.01(3,6), F(P) > F0.01(3,6), implying that the cutting fluid
be due to the insufficient cooling in the milling zone when the concentration, the cutting fluid flow rate, and the gas pressure all
cutting fluid flow rate is too small. Meanwhile, a too-large cutting have a highly significant effect on the surface roughness. Mean-
fluid flow rate could reduce the cutting fluid jet velocity, decrease while, the contribution rate of the cutting fluid concentration,
the ability of the cutting fluid jet to penetrate the gas barrier layer cutting fluid flow rate, and gas pressure to the surface roughness is
around the milling tool, and cause insufficient cooling. When the 49.16%, 22.71%, and 26.00%, respectively. Therefore, for the surface
gas pressure was increased from 0.4 MPa to 0.7 MPa, the SNR re- roughness, the most significant MQL parameter is the cutting fluid
sponses of the milling temperature also first increased and then concentration W, followed by the gas pressure P, while the cutting
decreased, and reached the maximum when the gas pressure was fluid flow rate Q is the least significant parameter. As can be seen
0.6 MPa. This is because the ability of the cutting fluid jet to from Table 10 and Fig. 9(d), when the cutting fluid concentration
penetrate the gas barrier layer around the milling tool is poor when was increased from 0.05 wt% to 0.2 wt%, the SNR responses of the
the gas pressure is less than 0.6 MPa, resulting in insufficient surface roughness first increased and then decreased, reaching the
cooling. Meanwhile, when the gas pressure is more than 0.6 MPa, maximum when the cutting fluid concentration was 0.1 wt%. When
the flyoff of the cutting fluid jet gradually increases with the the cutting fluid flow rate was increased from 30 mL/h to 120 mL/h,
M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522 1519

the SNR responses of the surface roughness first increased and then Table 12
decreased, with the maximum corresponding to a cutting fluid flow Eigenvalues and percentages of the principal component.

rate of 60 mL/h. When the gas pressure was increased from 0.4 MPa Principal component Eigenvalue Percentage Cumulative
to 0.7 MPa, the SNR responses of the surface roughness first First 3.525 88.13% 88.13%
increased and then decreased, reaching the maximum when the Second 0.251 6.28% 94.41%
gas pressure was 0.6 MPa. Insufficient lubrication and cooling leads Third 0.161 4.02% 98.43%
to deterioration in the surface quality and increase in the surface Fourth 0.063 1.57% 100%

roughness. As shown in Fig. 9(d), the optimal MQL parameter


combination for the minimum surface roughness is W2Q2P3,
which indicates 0.1 wt% cutting fluid concentration (level 2), 60 mL/ Table 13
Eigenvectors and contribution of the first principal component.
h cutting fluid flow rate (level 2), and 0.6 MPa gas pressure (level 3).
Overall, it could be concluded that the cutting fluid concentra- Milling characteristic Eigenvector contribution
tion is the more significant MQL parameter affecting the milling Fcmax 0.514 26.43%
characteristics rather than the cutting fluid flow rate and gas Tmax 0.498 24.78%
pressure in the MQL milling of TC4 alloy with graphene-dispersed MH 0.488 23.77%
vegetable-oil-based cutting fluid. However, the optimal MQL Ra 0.500 25.02%

parameter combinations for the milling force, milling temperature,


surface micro-hardness, and surface roughness are different.
Therefore, a multi-objective optimization process is needed to respectively. Moreover, the variance contribution for the first
obtain a unified optimal MQL parameter combination for the four principal component characterizing all the original variables (the
milling characteristics (the milling force, milling temperature, above GRC array) is as high as 88.13%, as can be seen in Table 12.
surface micro-hardness, and surface roughness). Therefore, the square of the corresponding eigenvector of the first
principal component was selected as the weight of the corre-
sponding milling characteristic, and the weights of the corre-
3.2. Multi-objective optimization based on PCA-GRA method sponding milling characteristic used to calculate the GRG are
26.43%, 24.78%, 23.77%, and 25.02%, respectively.
In order to obtain a unified optimal MQL parameter combination Finally, the GRG was calculated using Eq. (10), as shown in
for the four milling characteristics (the milling force, milling tem- Table 11. The ANOVA of the GRG is shown in Table 15. From the F
perature, surface micro-hardness and surface roughness), multi- distribution table, F0.01(3,6) ¼ 9.7795, F0.05(3,6) ¼ 4.76;
objective optimization based on the PCA-GRA method is per- F(W) > F0.01(3,6), indicating the cutting fluid concentration has a
formed. Based on the SNR data calculated in section 3.1, the highly significant effect on the GRG; F(Q) < F0.05(3,6),
normalized SNR of milling characteristics was calculated by using F(P) < F0.05(3,6), leading to the conclusion that the effects of the
Eq. (3) and the GRC of milling characteristics was calculated by cutting fluid flow rate and gas pressure on the GRG are not signif-
using Eq. (4), as illustrated in Table 11. icant. Meanwhile, the contribution rate of the cutting fluid con-
In order to provide an objective assessment of the relative centration, cutting fluid flow rate, and gas pressure to the GRG is
importance for each milling characteristic in the GRG, the PCA-
method was introduced to determine the corresponding weight
for each milling characteristic. The above GRC data were used to Table 14
calculate the correlation coefficient array and determine the cor- Mean GRG response table of the milling characteristics.
responding eigenvalues and eigenvectors, as shown in Tables 12 Level W Q P
and 13. The square of the eigenvector could represent the contri-
1 0.448 0.607 0.465
bution rate of the corresponding milling characteristic to the
2 0.821 0.674 0.594
principal component. Table 13 shows that the contribution rates of 3 0.594 0.604 0.663
the milling force, milling temperature, surface micro-hardness, and 4 0.490 0.469 0.631
surface roughness are 26.43%, 24.78%, 23.77%, and 25.02%, Delta 0.372 0.205 0.199
Rank 1 2 3

Table 11
Normalized SNR, GRC, and GRG of the milling characteristics.

No. Normalized SNR GRC GRG

Fcmax Tmax MH Ra Fcmax Tmax MH Ra Value Rank

1 0.049 0.000 0.041 0.079 0.345 0.333 0.343 0.352 0.343 15


2 0.602 0.214 0.760 0.560 0.557 0.389 0.676 0.532 0.537 9
3 0.517 0.654 0.587 0.524 0.509 0.591 0.548 0.512 0.539 8
4 0.316 0.084 0.139 0.059 0.422 0.353 0.367 0.347 0.373 14
5 0.960 0.859 1.000 0.978 0.927 0.780 1.000 0.958 0.916 2
6 0.893 0.677 0.769 0.817 0.823 0.608 0.684 0.732 0.714 5
7 1.000 1.000 0.917 1.000 1.000 1.000 0.858 1.000 0.966 1
8 0.941 0.609 0.441 0.872 0.894 0.561 0.472 0.796 0.687 7
9 0.893 0.579 0.684 0.907 0.823 0.543 0.613 0.843 0.709 6
10 0.855 0.642 0.729 0.935 0.775 0.583 0.648 0.885 0.725 3
11 0.459 0.455 0.158 0.542 0.480 0.479 0.373 0.522 0.465 11
12 0.568 0.171 0.347 0.610 0.536 0.376 0.434 0.562 0.479 10
13 0.435 0.142 0.067 0.724 0.469 0.368 0.349 0.645 0.459 12
14 0.874 0.662 0.599 0.949 0.798 0.596 0.555 0.908 0.718 4
15 0.246 0.331 0.395 0.512 0.399 0.428 0.452 0.506 0.446 13
16 0.000 0.056 0.000 0.000 0.333 0.346 0.333 0.333 0.337 16
1520 M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522

Table 15 (W2Q2P3) is not in the orthogonal experiment arrangement table.


The ANOVA table of the GRG. Therefore, it is necessary to verify the optimal MQL parameter
Factor DF Seq. SS Adj. MS F P Contribution combination through additional experiments.
W 3 0.3338 0.1113 13.01 0.005 59.07%
The initial MQL parameter combination is W1Q3P3 (Experiment
Q 3 0.0885 0.0295 3.45 0.092 15.66% No. 3), namely, 0.05 wt% cutting fluid concentration (level 1),
P 3 0.0915 0.0305 3.57 0.087 16.19% 90 mL/h cutting fluid flow rate (level 3), and 0.6 MPa gas pressure
Error 6 0.0513 0.0086 (level 3). The verification experiments were performed and
Total 15 0.5652
repeated 30 times, and the experiment results are shown in
Table 16 and Fig. 12. The results of the verification experiment
indicated that the milling force (maximum resultant force) was
reduced from 32.870 N to 26.911 N, the milling temperature was
decreased from 194.764  C to 168.296  C, the surface micro-
hardness was reduced from 359.126 H V to 329.117 H V, and the
surface roughness was decreased from 0.411 mm to 0.309 mm. The
improvement rates of the milling force, milling temperature, sur-
face micro-hardness, and surface roughness are 18.13%, 13.59%,
8.36%, and 24.82%. In conclusion, it is clearly shown that the mul-
tiple response characteristics in MQL milling of TC4 alloy with
graphene-dispersed vegetable-oil-based cutting fluid were signif-
icantly improved through the integrated Taguchi-PCA-GRA opti-
mization method.
Overall, in this work, the parameter optimization was per-
formed to improve the milling characteristics during minimum
quantity lubrication milling of TC4 alloy with graphene-dispersed
vegetable-oil-based cutting fluid. Meanwhile, many researchers
have applied nanoparticle-dispersed cutting fluid to the turning
and grinding. In the turning process, the Carbon nanotubes-
dispersed cutting fluid was adopted and showed superior perfor-
mances than pure cutting fluids in terms of cutting temperature
and surface roughness (Sharma et al., 2015). Meanwhile, Taguchi
Fig. 11. Mean GRG response graph of the experiment results.
design and response surface methodology were used to improve
the machining performances and achieved significant results
59.07%, 15.66%, and 16.19%, respectively. Hence, for the GRG, the (Sarıkaya and Güllü, 2014). In the grinding process, the machining
most significant MQL parameter was the cutting fluid concentra- characteristics in terms of friction coefficient, specific grinding
tion W, followed by the gas pressure P, while the least significant is
the cutting fluid flow rate Q. As shown in Table 14 and Fig. 11, when
the cutting fluid concentration was increased from 0.05 wt% to
0.2 wt%, the GRG first increased and then decreased, reaching the
maximum when the cutting fluid concentration was 0.1 wt%. When
the cutting fluid flow rate was increased from 30 mL/h to 120 mL/h,
the GRG first increased and then decreased, with the maximum
corresponding to a cutting fluid flow rate of 60 mL/h. When the gas
pressure was increased from 0.4 MPa to 0.7 MPa, the GRG also
increased initially and then decreased, reaching the maximum
when the gas pressure was 0.6 MPa. From Fig. 10, it can be seen that
the optimal MQL parameter combination for the GRG was W2Q2P3,
which represents 0.1 wt% cutting fluid concentration (level 2),
60 mL/h cutting fluid flow rate (level 2), and 0.6 MPa gas pressure
(level 3).

3.3. Verification experiment

As described in section 3.2, the optimal MQL parameter com-


bination is identified as W2Q2P3. However, as shown in Table 8, it
could be seen that the optimal MQL parameter combination Fig. 12. Results of verification experiment ((a) milling force, (b) milling temperature,
(c) surface micro-hardness, and (d) surface roughness).

Table 16
Experiment verification table.

milling characteristic Initial MQL parameter combination (W1Q3P3) Optimal MQL parameter combination (W2Q2P3) Improvement rate

Fcmax (N) 32.870 26.911 18.13%


Tmax ( C) 194.764 168.296 13.59%
MH (HV) 359.126 329.117 8.36%
Ra (mm) 0.411 0.309 24.82%
M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522 1521

energy, and surface roughness were improved significantly by us- MQL parameter combination was W2Q2P3. The cutting fluid
ing the MQL with MoS2-dispersed cutting fluid (Zhang et al., 2015). concentration has a highly significant effect, while the effects
Therefore, the nanoparticle additives are effective for improving of the flow rate and gas pressure are not significant. Finally,
the machining characteristics of traditional cutting fluids. In addi- the unified optimal MQL parameter combination was veri-
tion, the graphene-dispersed vegetable-oil-based cutting fluid is fied by an experiment, which ensured the robustness of the
also very promising in other machining areas, such as turning, optimal results. The verification experiment results indicated
grinding and so on. that all the four milling characteristics were significantly
improved through the integrated Taguchi-PCA-GRA optimi-
4. Conclusions and future work zation process. The improvement rates of the milling force,
milling temperature, surface micro-hardness, and surface
In this study, the milling characteristics of TC4 alloy under roughness are 18.13%, 13.59%, 8.36%, and 24.82%, respectively.
cooling/lubrication conditions with different MQL parameters are In addition, the investigation results attest to the feasibility of
investigated experimentally. Meanwhile, the effects of the MQL the integrated Taguchi-PCA-GRA optimization method and
parameters on the four milling characteristics (the milling force, also provides an experimental basis for the application of the
milling temperature, surface micro-hardness, and surface rough- graphene additive in MQL milling of TC4 alloy with
ness) were evaluated and analyzed using the SNR analysis based on graphene-dispersed vegetable-oil-based cutting fluid.
the Taguchi method. In addition, the unified optimal MQL param- (6) The graphene additive is effective for improving the
eter combination for the four milling characteristics is obtained machining performances in terms of the milling force, mill-
using the integrated Taguchi-PCA-GRA optimization method. The ing temperature, surface micro-hardness, and surface
following conclusions could be drawn from the study. roughness by improving the cooling and lubrication char-
acteristics of the vegetable-oil-based cutting fluid. Overall,
(1) The cutting fluid concentration, cutting fluid flow rate, and the application of the graphene-dispersed vegetable-oil-
gas pressure all have a pronounced effect on the milling based cutting fluid in the machining process is very mean-
force. Meanwhile, the most significant MQL parameter is the ingful and promising.
cutting fluid concentration W, followed by the gas pressure P, (7) MQL can greatly reduce the consumption of cutting fluid,
while the least is the cutting fluid flow rate Q. In addition, the which can reduce environmental pollution and high cost
optimal MQL parameter combination for the minimum caused by the subsequent processing of cutting fluid. In
milling force was W2Q2P3. This is because the appropriate addition, MQL can also enhance the ability of cutting fluid to
MQL parameter could enhance the lubrication property of enter the cutting zone, which can greatly improve the cool-
the oil film formed in the milling zone in the MQL milling of ing and lubrication effect. In general, MQL is a highly efficient
TC4 alloy with graphene-dispersed vegetable-oil-based cut- and low-cost machining technology.
ting fluid.
(2) With regard to the milling temperature, the cutting fluid In future work, the fundamental mechanisms of the effect of the
concentration has a highly significant effect, and the cutting graphene-dispersed vegetable-oil-based cutting fluid on the mill-
fluid flow rate has a moderately significant effect, while the ing characteristics (the milling force, milling temperature, surface
effect of the gas pressure is not significant. In addition, the micro-hardness, and surface roughness) need to be further studied.
optimal MQL parameter combination for minimum milling More theoretical modeling is required for better understanding of
temperature was W2Q3P3. This is because the appropriate these fundamental mechanisms. Meanwhile, the fundamental
MQL parameter could enhance the cooling property of the oil mechanisms of the effect of the graphene-dispersed vegetable-oil-
film formed in milling zone. based cutting fluid on tool wear are still to be further studied. In
(3) Both the cutting fluid concentration and the cutting fluid addition, the specific cutting energy needs to be further studied to
flow rate have a significant effect on the surface micro- predict the energy consumption during MQL milling of TC4 alloy
hardness, while the effect of the gas pressure is not signifi- with graphene-dispersed vegetable-oil-based cutting fluid.
cant. Meanwhile, the most significant MQL parameter is the
cutting fluid concentration W, followed by the cutting fluid Acknowledgement
flow rate Q, while the gas pressure P is the least significant. In
addition, the optimal MQL parameter for the minimum sur- This work was supported by grants from the Major Project of
face micro-hardness was W2Q2P2. This is likely because the Ministry of Industry and Information Technology of China
appropriate MQL parameter combination could enhance the (201675514), the Joint Funds of the National Natural Science
lubrication and cooling properties of the oil film to reduce Foundation of China (U1508206), the National Key Research and
the extrusion, friction, and cutting heat. Development Program of China (2017YFA0701200), and the Science
(4) The cutting fluid concentration, the cutting fluid flow rate and Technology Planning Project of Shenyang City (18006001). The
and the gas pressure all have a highly significant effect on the authors would like to thank Professor Ji Zhao, Jiashun Shi, and Lida
resulting surface roughness. Meanwhile, the most significant Zhu for excellent technical support and my wife (Hongxiao Li) for
MQL parameter is the cutting fluid concentration W, fol- critically reviewing the manuscript.
lowed by the gas pressure P, while the least significant is the
cutting fluid flow rate Q. In addition, the optimal MQL References
parameter for the minimum surface roughness was W2Q2P3.
This is mainly because appropriate MQL parameter could A. Ghani, J., Che Haron, C.H., Hamdan, S.H., Md Said, A.Y., Tomadi, S.H., et al., 2013.
Failure mode analysis of carbide cutting tools used for machining titanium
enhance the lubrication and cooling properties of the oil film
alloy. Ceram. Int. 39 (4), 4449e4456.
to improve the surface quality and reduce the surface Agrawal, S.M., Patil, N.G., 2018. Experimental study of non edible vegetable oil as a
roughness. cutting fluid in machining of M2 Steel using MQL. Procedia Manufacturing 20,
(5) The unified optimal MQL parameter combination for the four 207e212.
Balandin, A.A., Ghosh, S., Bao, W., Calizo, I., Teweldebrhan, D., Miao, F., Lau, C.N.,
milling characteristics is obtained by integrating the Taguchi 2008. Superior thermal conductivity of single-layer graphene. Nano Lett. 8 (3),
method, PCA method, and GRA method. The unified optimal 902.
1522 M. Li et al. / Journal of Cleaner Production 209 (2019) 1508e1522

Berman, D., Erdemir, A., Sumant, A.V., 2014. Graphene: a new emerging lubricant. 2862e2875.
Mater. Today 17 (1), 31e42. 
Ribeiro Filho, S.L.M., Vieira, J.T., de Oliveira, J.A., Arruda, E.M., Branda~o, L.C., 2017.
Cetin, M.H., Ozcelik, B., Kuram, E., Demirbas, E., 2011. Evaluation of vegetable based Comparison among different vegetable fluids used in minimum quantity
cutting fluids with extreme pressure and cutting parameters in turning of AISI lubrication systems in the tapping process of cast aluminum alloy. J. Clean.
304L by Taguchi method. J. Clean. Prod. 19 (17), 2049e2056. Prod. 140, 1255e1262.
Chetan, Behera, B.C., Ghosh, S., Rao, P.V., 2016. Application of nanofluids during Sarıkaya, M., Güllü, A., 2014. Taguchi design and response surface methodology
minimum quantity lubrication: a case study in turning process. Tribol. Int. 101, based analysis of machining parameters in CNC turning under MQL. J. Clean.
234e246. Prod. 65, 604e616.
Choi, S.U.S., Eastman, J.A., 1995. Enhancing thermal conductivity of fluids with Sarıkaya, M., Güllü, A., 2015. Multi-response optimization of minimum quantity
nanoparticles. Asme Fed 231 (1), 99e105. lubrication parameters using Taguchi-based grey relational analysis in turning
Debnath, S., Reddy, M.M., Yi, Q.S., 2014. Environmental friendly cutting fluids and of difficult-to-cut alloy Haynes 25. J. Clean. Prod. 91, 347e357.
cooling techniques in machining: a review. J. Clean. Prod. 83, 33e47. Sarıkaya, M., Yılmaz, V., Güllü, A., 2016. Analysis of cutting parameters and cooling/
Gajrani, K.K., Ram, D., Sankar, M.R., 2017. Biodegradation and hard machining per- lubrication methods for sustainable machining in turning of Haynes 25 su-
formance comparison of eco-friendly cutting fluid and mineral oil using flood peralloy. J. Clean. Prod. 133, 172e181.
cooling and minimum quantity cutting fluid techniques. J. Clean. Prod. 165. Sartori, S., Ghiotti, A., Bruschi, S., 2018. Solid lubricant-assisted minimum quantity
Goindi, G.S., Sarkar, P., Jayal, A.D., Chavan, S.N., Mandal, D., 2018. Investigation of lubrication and cooling strategies to improve Ti6Al4V machinability in finishing
ionic liquids as additives to canola oil in minimum quantity lubrication milling turning. Tribol. Int. 118, 287e294.
of plain medium carbon steel. Int. J. Adv. Manuf. Technol. 94 (1), 881e896. Sayuti, M., Erh, O.M., Sarhan, A.A.D., Hamdi, M., 2014. Investigation on the
Gunjal, S.U., Patil, N.G., 2018. Experimental investigations into turning of hardened morphology of the machined surface in end milling of aerospace AL6061-T6 for
AISI 4340 steel using vegetable based cutting fluids under minimum quantity novel uses of SiO2 nanolubrication system. J. Clean. Prod. 66, 655e663.
lubrication. Procedia Manufacturing 20, 18e23. Sharma, A.K., Tiwari, A.K., Dixit, A.R., 2016. Effects of Minimum Quantity Lubrication
Guo, S., Li, C., Zhang, Y., Wang, Y., Li, B., Yang, M., Zhang, X., Liu, G., 2017. Experi- (MQL) in machining processes using conventional and nanofluid based cutting
mental evaluation of the lubrication performance of mixtures of castor oil with fluids: a comprehensive review. J. Clean. Prod. 127, 1e18.
other vegetable oils in MQL grinding of nickel-based alloy. J. Clean. Prod. 140, Sharma, P., Sidhu, B.S., Sharma, J., 2015. Investigation of effects of nanofluids on
1060e1076. turning of AISI D2 steel using minimum quantity lubrication. J. Clean. Prod. 108,
Khan, M.M.A., Mithu, M.A.H., Dhar, N.R., 2009. Effects of minimum quantity lubri- 72e79.
cation on turning AISI 9310 alloy steel using vegetable oil-based cutting fluid. Stachurski, W., Sawicki, J., Wo jcik, R., Nadolny, K., 2018. Influence of application of
J. Mater. Process. Technol. 209 (15), 5573e5583. hybrid MQL-CCA method of applying coolant during hob cutter sharpening on
Kuram, E., Ozcelik, B., Bayramoglu, M., Demirbas, E., Simsek, B.T., 2013. Optimization cutting blade surface condition. J. Clean. Prod. 171, 892e910.
of cutting fluids and cutting parameters during end milling by using D-optimal Talib, N., Nasir, R.M., Rahim, E.A., 2017. Tribological behaviour of modified jatropha
design of experiments. J. Clean. Prod. 42, 159e166. oil by mixing hexagonal boron nitride nanoparticles as a bio-based lubricant for
Lawal, S.A., Choudhury, I.A., Nukman, Y., 2012. Application of vegetable oil-based machining processes. J. Clean. Prod. 147, 360e378.
metalworking fluids in machining ferrous metalsda review. Int. J. Mach. Tool Tao, Z., Yaoyao, S., Laakso, S., Jinming, Z., 2017. Investigation of the effect of grinding
Manufact. 52 (1), 1e12. parameters on surface quality in grinding of TC4 titanium alloy. Procedia
Li, B., Li, C., Zhang, Y., Wang, Y., Jia, D., Yang, M., 2016. Grinding temperature and Manufacturing 11, 2131e2138.
energy ratio coefficient in MQL grinding of high-temperature nickel-base alloy Vamsi Krishna, P., Srikant, R.R., Nageswara Rao, D., 2010. Experimental investigation
by using different vegetable oils as base oil. Chin. J. Aeronaut. 29 (4), on the performance of nanoboric acid suspensions in SAE-40 and coconut oil
1084e1095. during turning of AISI 1040 steel. Int. J. Mach. Tool Manufact. 50 (10), 911e916.
Mia, M., Dey, P.R., Hossain, M.S., Arafat, M.T., Asaduzzaman, M., Shoriat Ullah, M., Vazquez, E., Gomar, J., Ciurana, J., Rodríguez, C.A., 2015. Analyzing effects of cooling
Tareq Zobaer, S.M., 2018a. Taguchi S/N based optimization of machining pa- and lubrication conditions in micromilling of Ti6Al4V. J. Clean. Prod. 87,
rameters for surface roughness, tool wear and material removal rate in hard 906e913.
turning under MQL cutting condition. Measurement 122, 380e391. Viswanathan, R., Ramesh, S., Subburam, V., 2018. Measurement and optimization of
Mia, M., Gupta, M.K., Singh, G., Kro lczyk, G., Pimenov, D.Y., 2018b. An approach to performance characteristics in turning of Mg alloy under dry and MQL condi-
cleaner production for machining hardened steel using different cooling- tions. Measurement 120, 107e113.
lubrication conditions. J. Clean. Prod. 187, 1069e1081. Wang, Y., Li, C., Zhang, Y., Li, B., Yang, M., Zhang, X., Guo, S., Liu, G., Zhai, M., 2017a.
Najiha, M.S., Rahman, M.M., Kadirgama, K., 2016. Performance of water-based TiO2 Comparative evaluation of the lubricating properties of vegetable-oil-based
nanofluid during the minimum quantity lubrication machining of aluminium nanofluids between frictional test and grinding experiment. J. Manuf. Process.
alloy, AA6061-T6. J. Clean. Prod. 135, 1623e1636. 26, 94e104.
Nam, J.S., Kim, D.H., Chung, H., Lee, S.W., 2015. Optimization of environmentally Wang, Y., Li, C., Zhang, Y., Yang, M., Li, B., Jia, D., Hou, Y., Mao, C., 2016. Experimental
benign micro-drilling process with nanofluid minimum quantity lubrication evaluation of the lubrication properties of the wheel/workpiece interface in
using response surface methodology and genetic algorithm. J. Clean. Prod. 102, minimum quantity lubrication (MQL) grinding using different types of vege-
428e436. table oils. J. Clean. Prod. 127, 487e499.
Nam, J.S., Lee, P.-H., Lee, S.W., 2011. Experimental characterization of micro-drilling Wang, Y., Li, C., Zhang, Y., Yang, M., Zhang, X., Zhang, N., Dai, J., 2017b. Experimental
process using nanofluid minimum quantity lubrication. Int. J. Mach. Tool evaluation on tribological performance of the wheel/workpiece interface in
Manufact. 51 (7), 649e652. minimum quantity lubrication grinding with different concentrations of Al2O3
Ooi, M.E., Sayuti, M., Sarhan, A.A.D., 2015. Fuzzy logic-based approach to investigate nanofluids. J. Clean. Prod. 142, 3571e3583.
the novel uses of nano suspended lubrication in precise machining of aerospace Yang, M., Li, C., Zhang, Y., Jia, D., Zhang, X., Hou, Y., Li, R., Wang, J., 2017. Maximum
AL tempered grade 6061. J. Clean. Prod. 89, 286e295. undeformed equivalent chip thickness for ductile-brittle transition of zirconia
Pashmforoush, F., Delir Bagherinia, R., 2018. Influence of water-based copper ceramics under different lubrication conditions. Int. J. Mach. Tool Manufact. 122,
nanofluid on wheel loading and surface roughness during grinding of Inconel 55e65.
738 superalloy. J. Clean. Prod. 178, 363e372. Zhang, Y., Jun, M.B.G., 2014. Feasibility of lignin as additive in metalworking fluids
Rahim, E., HiroyukiSasahara, 2011. An analysis of surface integrity when drilling for micro-milling. J. Manuf. Process. 16 (4), 503e510.
inconel 718 using palm oil and synthetic ester under MQL condition. Mach. Sci. Zhang, Y., Li, C., Ji, H., Yang, X., Yang, M., Jia, D., Zhang, X., Li, R., Wang, J., 2017.
Technol. 15 (1), 76e90. Analysis of grinding mechanics and improved predictive force model based on
Rahim, E.A., Sasahara, H., 2011. A study of the effect of palm oil as MQL lubricant on material-removal and plastic-stacking mechanisms. Int. J. Mach. Tool Manufact.
high speed drilling of titanium alloys. Tribol. Int. 44 (3), 309e317. 122, 81e97.
Rapeti, P., Pasam, V.K., Rao Gurram, K.M., Revuru, R.S., 2018. Performance evaluation Zhang, Y., Li, C., Jia, D., Zhang, D., Zhang, X., 2015. Experimental evaluation of MoS2
of vegetable oil based nano cutting fluids in machining using grey relational nanoparticles in jet MQL grinding with different types of vegetable oil as base
analysis-A step towards sustainable manufacturing. J. Clean. Prod. 172, oil. J. Clean. Prod. 87, 930e940.

You might also like