International Journal of Science and Research (IJSR)

ISSN (Online): 2319-7064

Index Copernicus Value (2013): 6.14 | Impact Factor (2013): 4.438

Effects of Cutting Parameters on Surface Roughness

during End Milling of Aluminium under Minimum
Quantity Lubrication (MQL)
Okokpujie Imhade P. 1, Okonkwo Ugochukwu C.2
Department of Mechanical Engineering, Nnamdi Azikiwe University, Awka, Nigeria

Abstract: In this study an experimental investigation of effects of cutting parameters on surface roughness during end milling of
aluminium 6061 under minimum quantity lubrication (MQL) condition was carried out. The experiments were carried out to investigate
surface quality of the machined parameters and to developed mathematical models using least square techniques. Spindle speed (N),
feed rate (f), axial depth of cut (a) and radial depth of cut (r) has been chosen as input variables in order to predict surface roughness.
The experiment was designed by using central composite design (CCD) in which 30 samples were run in a CNC milling machine. Each
of the experimental result was measured using Mitutoyo surface tester. After the predicted surface roughness values have been obtained
the average percentage errors were calculated. The mathematical model developed by using least square method shows accuracy of
89.5% which is reasonably reliable for surface roughness prediction. With the obtained optimum input parameters for surface
roughness, production operations will be enhanced.

Keywords: CNC end milling, Minimum quantity lubrication (MQL), Surface roughness, Response surface methodology.

1. Introduction approach and, the fundamental approach involving

analytical means. The empirical approach is considered a
Milling is a process of generating machined surfaces by short-term and practical method, and it is the most suited
progressively removing a predetermined amount of materials approach for industrial applications. Ginta et al, (2009)
from the work-piece at a relatively slow rate of movement developed an effective methodology to determine the
by a milling cutter rotating at a comparatively high speed. performance of uncoated WC-Co inserts in predicting
The characteristic feature of the milling process is that each minimum surface roughness in end milling of titanium
milling cutter tooth removes its share of the stock in the alloys Ti-6Al-4V under dry conditions. Central composite
form of small individual chips. It is of three types which are: design (CCD) of response surface methodology was
Peripheral milling, Face milling and End milling. End employed to create an efficient analytical model for surface
milling is one of the most common metal removal operation roughness in terms of cutting parameters: cutting speed,
encountered in industrial process. It is widely used in the axial depth of cut, and feed per tooth. End milling tests were
manufacturing industries which include the automotive and conducted on Vertical Machining Center (VMC ZPS,
aerospace sectors, where quality is an important factor in the Model: MLR 542 with full immersion cutting and under dry
production of slots, pockets, precision molds, and dies. In condition. They concluded with CCD being a successful
end milling, the cutter generally rotates on an axis vertical to technique to predict the surface roughness produced in end-
the work-piece. It can be tilted to machine tapered surfaces. milling of titanium alloy Ti-6Al-4V using uncoated inserts
Cutting teeth are located on both the end face of the cutter under dry conditions. Linear CCD model proved inadequate
and the periphery of the cutter body. The quality of while quadratic CCD model is adequate with 95% accuracy.
machined surface is characterized by the accuracy of its The two developed models indicated that feed is the most
manufacture with respect to the dimensions specified by the predominant cutting condition followed by cutting speed and
designer. Each type of cutting tool leaves its own individual depth of cut. Interaction effect between cutting speed and
pattern which therefore can be identified. This pattern is feed will also give a high effect on surface roughness values.
known as surface finish or surface roughness.
Arokiadass et al, (2011) also studied the influence of four
The primary function of the MQL in metal machining machining parameters including spindle speed (N), feed rate
operations is to serve as a coolant, also as a lubricant thereby (f), depth of cut (d), and various percentage weight of silicon
reducing friction and tool wear. It is generally agreed that carbide (S) on surface roughness (Ra). The response surface
the application of MQL can improve the tool life and results methodology was employed to establish the mathematical
in good surface finish by reducing thermal distortion and relationship between the response and the various process
flushing away of machined chips. What is perhaps even parameters. The result they obtained shows that the
more important is to ensure proper filtration of the fluid as quadratic model is statistically significant for analysis of
suspended swarf can cause random deep scratches on the surface roughness. The value of R is 99.85 %, which
job. Predictive modeling of machining processes is the first indicates that the developed regression model is adequately
and the most important step for process control and significant at a 95% confidence level. Their model also
optimization. A predictive model is an accurate relationship indicated that the feed rate was the most dominant parameter
between the independent input variables and dependent on surface roughness followed by spindle speed and
output performance measures. There are two well-known %weight of SiC. Depth of cut has less influence on surface
approaches to obtain this relationship: the empirical roughness. They also concluded that the regression model is
Volume 4 Issue 5, May 2015
www.ijsr.net
Paper ID: SUB154684 2937
Licensed Under Creative Commons Attribution CC BY
 International Journal of Science and Research (IJSR)
ISSN (Online): 2319-7064
Index Copernicus Value (2013): 6.14 | Impact Factor (2013): 4.438
well fitted with the observed values and high correlation that with the desired response and exploring the effect of these
exists between fitted values and observed values. parameters on responses, including six steps (Chiang 2008).
These are in the order;
2. Materials and Methods
• Define the independent input variables and the desired
The work piece material used for the study is a rectangular responses with the design constants.
6061Aluminium blocks of 2000mm×50mm×5mm. Method • Adopt an experimental design plan.
used for the experimental investigations is explained thus: • Perform regression analysis with the quadratic model of
 Preparation of the vertical CNC milling machine system RSM.
ready for performing the machining operation, Cutting of • Calculate the statistical analysis of variance (ANOVA)
the work piece of the aluminium 6061 rectangle plate for the independent input variables in order to find which
into different sizes of 10, 15, 20, 25 and 30mm. A total parameter significantly affects the desired response.
of 30 pieces, for MQL condition • Determine the situation of the quadratic model of RSM
 Fixing of the high speed steel (HSS) end milling cutter of and decide whether the model of RSM needs screening
12mm diameter on the spindle taper of the machine variables or not.
 Mounting the work piece, clamped on a vice mounted on • Optimize and conduct confirmation experiment and
top of the table of the machine verify the predicted performance characteristics.
 Creating CNC part programs on CNC professional
In the current study, the relationship between the cutting
software for tool paths, with specific commands using
conditions and the technology parameters aspect is given as
different levels of spindle speed, feed rate, axial depth of
(Sabahudin et al, 2011).
cut and radial depth of cut, taking reference for Y axis,
Y = φ (N, f, a, r), (1)
and Z axis then performing end milling operation.
 After each machining the surface roughness of the work
Where Y is the desired machinability aspect and φ is the
piece was measured with the press-o-firm and mitutoyo
response function. The approximation of Y is proposed by
surface tester
using a non-linear (quadratic) mathematical model, which is
Detailed information on chemical composition of the 6061
suitable for studying the interaction effects of process
Aluminium is provided in table 1, and details of the
parameters on machinability characteristics. In the present
experimental outlay, only up-milling cutting mode was
work, the RMS-based second order mathematical model is
investigated.
given by
Y = β0+ β1x1 + β2x2 + β3x3 + β4x4 (2)
Table 1: Chemical Composition of Al-6061
Where βo is the free term of the regression equation, the
Element Mg Fe Si Cu Mn V Ti AL
Weight % 1.08 0.17 0.63 0.32 0.52 0.01 0.02 Remainder
coefficients, β1, β2, β3 and β4 values are the estimates of
corresponding parameters , x1, x2, x3, x4 are logarithmic
transformation of factors: spindle speed, cutting feed, axial
Table 2: Details of the Experimental Outlay
Exp. Material Mql Cutting Cutting Input Parameters Response
dept of cut and radial depth of cut, respectively.
Runs Condition Tool Parameters
1 to 30 Al-6061 10% boric High Cutting speed Surface The experimental plan is developed to assess the influence
alloy acid + base speed Feed rate Roughness of spindle speed (N), feed rate (f), axial depth of cut (a) and
oil SAE 40 steel
Axial depth of cut
radial depth of cut (r) on the surface roughness parameters
(Ra). Five levels were allocated for each cutting variable as
Radial depth of cut
given in table 3. The variable levels were chosen within the
intervals recommended by cutting tool manufacturer. Four
The experiment was performed on SIEG 3/10/0010 table top cutting variables at five levels led to a total of 30 tests for
CNC machine vertical milling centre. The vertical milling each condition.
centre has three (3) planes namely x, y and z planes.
Table 3: Factor levels to be used in the experimental design
Response surface methodology (RSM) was employed in the Levels
experimental design using second-order rotatable central Variable
-2 -1 0 1 2
composite design. By considering all the factorial corner Spindle speed [rpm] 1000 1500 2000 2500 3000
points, some of the central replicates and all the axial points Feed rate [mm/min] 100 150 200 300 500
second-order rotatable central composite design requires Radial depth of cut [mm] 0.5 1 1.5 2.0 2.5
between 25 to 33 experimental runs depending on the Axial depth of cut[mm] 10 15 20 25 30
number of the central replicates considered while a full
factorial design will require 54= 625 experimental runs. This Mathematical model of surface roughness was built for
explains the choice of second-order rotatable central MQL cutting condition. Percentage improvement in surface
composite design which tremendously reduces needed roughness expected to be occasioned by MQL was thereafter
number of experimental runs for the MQL cutting quantified. Furthermore, optimization of the arising model
conditions, which doubles the calculated number of was carried out to determine the coordinate of minimum
experimental runs. The design expert 9.0.1 was used in surface roughness.
analysis and presentation of results. The response surface
methodology (RSM) is the procedure for determining the The required number of experimental points for four-factor
relationship between the independent process parameters in the C.C.D with one replication of factorial and axial parts

Volume 4 Issue 5, May 2015

www.ijsr.net
Paper ID: SUB154684 2938
Licensed Under Creative Commons Attribution CC BY
 International Journal of Science and Research (IJSR)
ISSN (Online): 2319-7064
Index Copernicus Value (2013): 6.14 | Impact Factor (2013): 4.438
having, factorial design is = 2f = 24 = 16, the axial point or
star point is = 2×f = 2×4 = 8, where f= number of factors, The constant and exponents K, x, y, z, zr can be determined
the center point chosen for this experiment is 6, which is = by least squares method. The introduction of a replacement
16+8+6 = 30. Therefore the thirty experiments are carried gets the following expression:
out according to the blocked central composite design Y = InRa, β0 = InK, x1 = InN, x2 = Inf, x3 = In a, x4=Inr, x =
(CCD). β1, y = β2, z = β3, zr = β4 (5)
Therefore, eβ0 = K (6)
3. Mathematical Models
Linear model developed from the equation can be
The relationship between the surface roughness and cutting represented as follows:
independent variables can be represented by the following Y = β0+ β1x1 + β2x2 + β3x3 + β4x4 (7)
equation (Sabahudin et al, 2011).
Ra=K.Nx.fy.az.rzr (3) Where, x1, x2, x3, x4, are logarithmic transformation of
factors: spindle speed, feed rate, axial depth of cut and radial
Where, K is constant, and x, y, z and r are the exponents. depth of cut and β values are the estimates of corresponding
Equation (3) can be represented in mathematical form as parameters.
follows:
InRa = In K + x.In N + y.In f + z.In a + zr.In a (4)

Table 4: Experimental result for MQL environment

Std Factor 1 A: Spindle Factor 2 B: Feed rate Factor 3 C: Axial Factor 4 D: Radial Factor 5 E: Surface
Run
speed(rpm) (mm/min) depth of cut(mm) depth of cut(mm) roughness (Ra)(µm)
13 1 -1 -1 1 1 0.94
14 2 1 -1 1 1 0.85
8 3 1 1 1 -1 1.02
11 4 -1 1 -1 1 1.11
9 5 -1 -1 -1 1 0.92
24 6 0 0 0 2 1.1
1 7 -1 -1 -1 -1 0.9
25 8 0 0 0 0 1.01
5 9 -1 -1 1 -1 0.88
18 10 2 0 0 0 0.51
20 11 0 2 0 0 1.16
16 12 1 1 1 1 1.1
19 13 0 -2 0 0 0.5
4 14 1 1 -1 -1 0.98
22 15 0 0 2 0 1
23 16 0 0 0 -2 0.88
26 17 0 0 0 0 1.08
10 18 1 -1 -1 1 0.93
2 19 1 -1 -1 -1 0.74
27 20 0 0 0 0 1.08
17 21 -2 0 0 0 1.12
12 22 1 1 -1 1 1.07
15 23 -1 1 1 1 1.14
21 24 0 0 -2 0 0.92
30 25 0 0 0 0 0.96
3 26 -1 1 -1 -1 1.06
7 27 -1 1 1 -1 1.04
6 28 1 -1 1 -1 0.6
29 29 0 0 0 0 1.01
28 30 0 0 0 0 1.01

From equation (7), by minimizing the sum of the squares of β0 ∑x3+ β1∑ x1x3 + β2∑ x2x3 + β3∑x32 + β4∑x3x4 = ∑x3Yi
the residual, β0 ∑x4+ β1∑ x1x4 + β2∑ x2x4 + β3∑ x3x4 + β4∑x42 = ∑x4Yi
We have Since the surface roughness from the experiment has been
established, the analysis for the multiple regressions using
equations above are done to obtain regression coefficient
and the sum values calculated for xi, with the following
Solving the minimization, the resulting equations are as results:
follows ∑x1= 227.2231 ∑x1x2= 1212.728
nβ0+ β1∑x1 + β2∑x2 + β3∑x3 + β4∑x4 = ∑Yi ∑x2= 160.1149 ∑x1x3= 674.6051
β0 ∑x1+ β1∑x12 + β2∑x1x2 + β3∑x1x3 + β4∑x1x4 = ∑x1Yi ∑x3= 89.06798 ∑x1x4= 80.53167
β0 ∑x2+ β1∑x1x2 + β2∑x22 + β3∑x2x3 + β4∑x2x4 = ∑x2Yi ∑x4= 10.6339 ∑x1Yi= -15.7963
∑Yi= -2.00772 ∑x2x3= 475.3713
Volume 4 Issue 5, May 2015
www.ijsr.net
Paper ID: SUB154684 2939
Licensed Under Creative Commons Attribution CC BY
 International Journal of Science and Research (IJSR)
ISSN (Online): 2319-7064
Index Copernicus Value (2013): 6.14 | Impact Factor (2013): 4.438
∑x12=1722.695 ∑x2x4= 56.75883
227.2231 + 1722.695 + 1212.728 + 674.6051 +
∑x22= 857.8118 ∑x2Yi= -9.41013
80.53167β4 = -15.7963
∑x32=266.1206 ∑x3x4= 31.56074
∑x42=7.136489 ∑x3Yi= -5.98435 160.1149 + 1212.728 + 857.8118 + 475.3713 +
∑x4Yi=-0.23536 56.75883β4 = -9.41013
Substituting all the sums values into the simultaneous
equation of linear system as follows 10.6339 + 80.53167 + 56.75883 + 31.56074 +
7.136489β4 = -5.98435
30 + 227.2231 + 160.1149 +89.06798 +
10.6339β4 = -2.00772 89.06798 + 674.6051 + 475.3713 + 266.1206 +
31.56074β4 = -0.23536
Transform above equations into matrix form

After completing the solution for the matrix form, the And from equation 5, x = -0.3496, y =0.4013, z =-0.0144
regression coefficients estimated were; and zr = 0.1398
β0 = 0.4324 Finally, mathematical model of surface roughness for MQL
β1 = -0.3496 is:
β2 = 0.4013 Ra = 1.5409.N -0.3496.f 0.4013.a -0.0144.r 0.1398
β3 = -0.0144 Hence, the mathematical model of MQL condition is
β4 = 0.1398
From equation 6, K= e0.4324
Therefore, K = 1.5409
Table 5: Comparison between Measure Data and Predicted Data (MQL Condition)
Exp No. Spindle speed Feed Rate Axial depth of Radial Surface Roughness Predicted Percentage
(rpm) (mm/min) cut (mm) depth of cut (Ra) (µm) values(Ra) (µm) deviation (φi)
(mm)
1 1500 150 25 2 0.94 0.93 0.11
2 2500 150 25 2 0.85 0.78 7.60
3 2500 300 25 1 1.02 0.94 7.70
4 1500 300 15 2 1.11 1.25 -12.54
5 1500 150 15 2 0.92 0.95 -2.81
6 2000 200 20 2.5 1.1 0.99 10.33
7 1500 150 15 1 0.9 0.86 4.61
8 2000 200 20 1.5 1.01 0.92 9.07
9 1500 150 25 1 0.88 0.85 3.15
10 3000 200 20 1.5 0.51 0.79 -56.27
11 2000 500 20 1.5 1.16 1.32 -14.36
12 2500 300 25 2 1.1 1.04 5.71
13 2000 100 20 1.5 0.5 0.69 -39.07
14 2500 300 15 1 0.98 0.95 3.23
15 2000 200 30 1.5 1 0.91 8.69
16 2000 200 20 0.5 0.88 0.79 10.49
17 2000 200 20 1.5 1.08 0.92 14.96
18 2500 150 15 2 0.93 0.79 14.93
19 2500 150 15 1 0.74 0.72 2.96
20 2000 200 20 1.5 1.08 0.92 14.96
21 1000 200 20 1.5 1.12 1.17 -4.48
22 2500 300 15 2 1.07 1.04 2.34
23 1500 300 25 2 1.14 1.24 -8.77
24 2000 200 10 1.5 0.92 0.93 -0.82

Volume 4 Issue 5, May 2015

www.ijsr.net
Paper ID: SUB154684 2940
Licensed Under Creative Commons Attribution CC BY
 International Journal of Science and Research (IJSR)
ISSN (Online): 2319-7064
Index Copernicus Value (2013): 6.14 | Impact Factor (2013): 4.438
Exp No. Spindle speed Feed Rate Axial depth of Radial Surface Roughness Predicted Percentage
(rpm) (mm/min) cut (mm) depth of cut (Ra) (µm) values(Ra) (µm) deviation (φi)
(mm)
25 2000 200 20 1.5 0.96 0.92 4.34
26 1500 300 15 1 1.06 1.13 -6.96
27 1500 300 25 1 1.04 1.12 -8.22
28 2500 150 25 1 0.6 0.71 -18.81
29 2000 200 20 1.5 1.01 0.92 9.07
30 2000 200 20 1.5 1.01 0.92 9.07

Figure 2: Actual and Predicted Values of the Surface Roughness in MQL Condition

Similarly, the actual values gotten from the experiment and

the predicted values obtained from the developed 4. Effects of Cutting Parameters on Surface
mathematical model are depicted in figure 2. It can be seen Roughness under MQL Condition
that they have good agreement. Quantitatively, In order to
judge the accuracy of the experimental developed The effects of cutting parameters on surface roughness in
mathematical models, percentage deviation φi and average end milling of aluminium were investigated using contour
percentage deviation were used. The percentage plots of the results obtained in MQL conditions. The
deviation φi is stated thus: graphical evaluation was obtained by plotting surface
roughness values against the various cutting parameters
(axial depth of cut, radial depth of cut, spindle speed and
feed rate). Surface roughness values are simultaneously
plotted against two cutting parameters while keeping the
(10)
other two constant. Figures 3-4 show the experimental
Where φi: percentage deviation of single sample data, Ra(e):
results obtained from the effect of cutting parameters on
measured, Ra(e): predicted Ra(m) generated by a multiple
surface roughness.
regression equation.
Similarly, the average percentage deviation is stated thus:

Where : average percentage deviation of all sample data

n: the size of sample data.
For training data = = 10.54%

The result of average percentage deviation ( ) showed that

the training data set (n=30) was 10.54%. This means that the
statistical model could predict the surface roughness (Ra)
with about 89.5% accuracy of the training data set. For a full
test on the model created on the training data, table 6 shows
the predicted value for surface roughness and percentage Figure 3: Surface Roughness Contour Plot for Spindle
deviation from the measured or actual Ra values. Speed vs Feed Rate in MQL Condition

Volume 4 Issue 5, May 2015

www.ijsr.net
Paper ID: SUB154684 2941
Licensed Under Creative Commons Attribution CC BY
 International Journal of Science and Research (IJSR)
ISSN (Online): 2319-7064
Index Copernicus Value (2013): 6.14 | Impact Factor (2013): 4.438
axial depth of cut 20mm and radial depth of cut 1.5mm for
these conditions, the minimum surface roughness was
0.50µm.
• Interaction effect between spindle speed and feed rate also
possesses a major effect over the surface roughness,
followed by axial depth of cut and radial depth of cut.

References
[1] Arokiadass, R., Palaniradja, K., & Alagumoorthi, N.
(2011). Predictive modeling of surface roughness in end
milling of Al/SiCp metal matrix composite. Archive of
Applied Science Research, 3 (2) 228-236
[2] Benardos, P. G. & Vosniakos, G. C. (2002). Prediction
Figure 4: Surface Roughness Contour Plot for Axial Depth of surface roughness in CNC faces milling using neural
of Cut vs Radial Depth of Cut in MQL Condition networks and taguchi's design of experiments. Robotics
and Computer-Integrated Manufacturing 18 (5-6): 343-
Following conclusions can be deduced from figure 3 and 354.
figure 4. [3] Chapra, S.C. & Canale, R.P. Numerical methods for
engineers 5th edition. The Mcgraw-Hill Companies,
Spindle speed: An increase in spindle speed increases the [4] Datta, u., Lanjewar, R., & Balamurugan, G. (Groover
cutting force and eliminates the built-up edge (BUE) 2007). Development of Automatic CTC Roller Chasing
tendency. At low spindle speed (rpm), the unstable larger Machine. 13th National Conference on Mechanisms and
BUE is formed and also the chips fracture readily producing Machines (NaCoMM07), IISc, Bangalore, India,
the rough surface. As the spindle speed (rpm) increases, the December 12-13, 2007 NaCoMM122
BUE vanishes, chip fracture decreases, and hence, the [5] Ginta, T.L., Nurul, A.K.M., Amin, H.C.D., Mohd Radzi
roughness decreases. These findings were in line with & Lajis, M.A. (2009).Development of surface
observations made by Tosun and Mesut (2010); Korkut and roughness models in end milling titanium alloy ti-6al-4v
Donertas (2007) in related studies. Feed rate: An increase in using uncoated tungsten carbide inserts. European
feed rate significantly increases the surface roughness. Journal of Scientific Research, ISSN 1450-216X 28, 4,
Increasing feed rate increases vibration and heat generated, 542-551
which courses an increase in surface roughness. As the feed [6] Korkut, I., & Donertas, M. A. (2007). The influence of
rate is increased, chips become discontinuous and are feed rate and cutting speed on the cutting forces, surface
deposited between work piece and tool leading to increased roughness and tool–chip contact length during face
coefficient of friction and more interruption resulting in poor milling. Materials & Design, 28, 308–312.
surface finish. This finding is also supported by Arokiadass [7] Sabahudin, E., Joan, V.C., Edin. B., & Nadja, O.A.
et al (2011). Radial depth of cut: increasing the radial depth (2011). Experimental modeling of surface roughness
of cut will slightly increase the surface roughness. Axial parameters during cutting steel 30NiMo8, 15th
depth of cut: it has no significant effect on the surface International Research/Expert Conference, TMT.
roughness. This is supported by observation. [8] Tosun, N., & Huseyinoglu, M. (2010). Effect of MQL
on surface roughness in milling of AA7075-T6.
5. Conclusion Materials and Manufacturing Processes, 25, 793-798.
[9] Wang, M.Y., & Chang, H.Y. (2004). Experimental study
Experimental work is carried out on aluminium metal in of surface roughness in slot end milling, International
MQL environments. Through experimentation, the system Journal of Machine Tools & Manufacture 44, 51–57.
proved it is capable of predicting the surface roughness (Ra)
with about 89.5% accuracy in MQL environment. The
important conclusions drawn from the present research are
summarized as follows:
• The quadratic second order models developed to predict
the surface roughness value for the MQL cutting condition
could provide predictive values for surface roughness
pretty close to the actual values by applying the values of
the control parameter on the model.
• In the order of influence, spindle speed is the most
significant effect on the surface roughness, followed by
feed rate. However radial depth of cut has little effect on
the surface roughness and axial depth of cut has no
significant effect on the surface roughness.
• From the experimental values of table 4, the optimum or
minimum surface roughness during cutting process occurs
at spindle speed of 2000rpm, feed rate of 100mm/min,

Volume 4 Issue 5, May 2015

www.ijsr.net
Paper ID: SUB154684 2942
Licensed Under Creative Commons Attribution CC BY