See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/26527438

A Study of the Effects of Machining Parameters on the Surface Roughness in

the End-Milling Process

Article · January 2007

Source: DOAJ

CITATIONS READS

120 5,783

3 authors, including:

Mohammed Hayajneh Montasser S. Tahat

Jordan University of Science and Technology Al-Balqa' Applied University
91 PUBLICATIONS 2,205 CITATIONS 20 PUBLICATIONS 287 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mohammed Hayajneh on 24 January 2016.

The user has requested enhancement of the downloaded file.

JJMIE
Volume 1, Number 1, Sep. 2007
ISSN 1995-6665
Pages 1 - 5
Jordan Journal of Mechanical and Industrial Engineering

A Study of the Effects of Machining Parameters on the Surface

Roughness in the End-Milling Process
Mohammed T. Hayajneh a,*, Montasser S. Tahat b, Joachim Bluhm c
a
Industrial Engineering Dep., Faculty of Engineering, Jordan University of Science and Technology, P.O. Box 3030, Irbid-22110, Jordan
b
Mechanical Engineering Department, Al-Huson Polytechnic, AlBalqa’ Applied University, P. O. Box 50, Huson-21510 , Jordan
c
Institute of Mechanics, University Duisburg-Essen, Campus Essen, 45117 Essen, Universitatsstr. 15, Germany

Abstract

A set of experiments designed to begin the characterization of surface quality for the end-milling process have been
performed. The objective of this study is to develop a better understanding of the effects of spindle speed, cutting feed rate
and depth of cut on the surface roughness and to build a multiple regression model. Such an understanding can provide
insight into the problems of controlling the finish of machined surfaces when the process parameters are adjusted to obtain a
certain surface finish. The model, which includes the effect of spindle speed, cutting feed rate and depth of cut, and any two-
variable interactions, predicted the surface roughness values with an accuracy of about 12%.

© 2007 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved

Keywords: Surface Finish; ANOVA; Pareto Chart; Surface Roughness; End milling;

One may ask a logical question: do we really need to

improve the methodology of metal cutting experimental
1. Introduction study? The answer to this question is given in the recent
CIRP working paper [7] the quote of which is as follows:
Metal cutting is one of the most significant “A recent survey by a leading tool manufacturer indicates
manufacturing processes in the area of material removal that in the USA the correct cutting tool is selected less than
[1]. Black [2] defined metal cutting as the removal of 50% of the time, the tool is used at the rated cutting speed
metal chips from a workpiece in order to obtain a finished only 58% of the time, and only 38% of the tools are used
product with desired attributes of size, shape, and surface up to their full tool life capability”. The same has been
roughness. found in an earlier survey of cutting regime selection on
The imperative objective of the science of metal cutting CNC machine tools in the American aircraft industry [8]
is the solution of practical problems associated with the showing that selected cutting speeds are far below the
efficient and precise removal of metal from workpiece. It optimal economic speeds.
has been recognized that the reliable quantitative The demand for high quality and fully automated
predictions of the various technological performance production focuses attention on the surface condition of
measures, preferably in the form of equations, are essential the product, especially the roughness of the machined
to develop optimization strategies for selecting cutting surface, because of its effect on product appearance,
conditions in process planning [3-5]. function, and reliability. For these reasons it is important
The progress in the development of predictive models, to maintain consistent tolerances and surface finish [9].
based on cutting theory, has not yet met the objective; the Also, the quality of the machined surface is useful in
most essential cutting performance measures, such as, tool diagnosing the stability of the machining process, where a
life, cutting force, roughness of the machined surface, deteriorating surface finish may indicate workpiece
energy consumption, … etc., should be defined using material non-homogeneity, progressive tool wear, cutting
experimental studies. Therefore, further improvement and tool chatter, etc.
optimization for the technological and economic The accelerated application of computer aided
performance of machining operations depend on a well- manufacturing (CAM) to machining by the use of CNC
based experimental methodology. Unfortunately, there is a machine tools has focused on developing reliable
lack of information dealing with test methodology and data machinery data systems, to ensure optimum production
evaluation in metal cutting experiments [6]. using expensive equipment. These computerized
machinability data systems have been classified into two

*
Corresponding author. e-mail: hayajneh@just.edu.jo
2 © 2007 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved - Volume 1, Number 1 (ISSN 1995-6665)

general types [10], namely database system and designs are appropriate when there are no restrictions on
mathematical model system. The database system uses the the order of the testing to avoid systematic biases error due
collection and storage of large quantities of data from to the wear of the cutting tool. The procedure to define a
experiments, and the mathematical models attempt to model of the process includes the following steps:
predict the optimum conditions [10]. 1. Selecting the factors to be involved in the
Among several industrial machining processes, milling process and choosing the levels of these
is a fundamental machining operation. End milling is the factors.
most common metal removal operation encountered. It is 2. Conducting the experimental at all possible
widely used in a variety of manufacturing industries factor level combinations randomly.
including the aerospace and automotive sectors, where 3. Analyzing the collected data using parametric
quality is an important factor in the production of slots and analyses of variance (ANOVA).
dies. The quality of the surface plays a very important role 4. Building the multiple regression model.
in the performance of milling as a good-quality milled 5. Validating of the model.
surface significantly improves fatigue strength, corrosion
resistance, and creep life. Surface roughness also affects 2.2. Experimental Procedure
several functional attributes of parts, such as wearing, heat
transmission, ability of holding a lubricant, coating, or This experiment employed a Bridgeport end-milling
resisting fatigue. Therefore, the desired finish surface is machine. Eight ¾-inch four-flute high-speed steel cutters
usually specified and the appropriate processes are were used. The experiment has been done under dry
selected to reach the required quality. Several factors machining environment. The experimental setup is shown
influence the final surface roughness in end milling in figure. 1. The cutting parameters were set as: four levels
operation [11]. Factors such as spindle speed, feed rate, of spindle speed (750, 1000, 1250, 1500 rpm), seven levels
and depth of cut that control the cutting operation can be of feed rate (150, 225, 300, 375, 450, 525, 600 mm/min),
setup in advance. However, factors such as tool geometry, and three levels of depth of cut (0.25, 0.75, 1.25 mm). The
tool wear, and chip formation, or the material properties of cutters used to execute the experiment were selected
both tool and workpiece are uncontrolled [12]. randomly. Surface roughness Ra measured in micro-meters
One should develop techniques to predict the surface was the response variable. Several variables were put
roughness of a product before milling in order to evaluate under close control including the machine on which
the robustness of machining parameters such as feed rate milling operation was performed (the same machine was
or spindle speed for keeping a desired surface roughness used for all experimental work), and the operator (the same
and increasing product quality. It is also important that the operator machined all specimens). The surface roughness
prediction technique should be accurate and reliable. data were collected randomly for each of the 84 machining
Researchers in this area attempt to develop models conditions defined by the levels of independent variables
which can predict surface finish of a metal for a variety of (4 spindle speeds ×7 cutting feeds × 3 depths of cut). The
machining conditions such as speed, feed, depth of cut, experiment was performed on aluminum workpieces.
etc. Reliable models would not only simplify
manufacturing process planning and control, but would
assist in optimizing machinability of materials. Therefore,
the purpose of this study is (1) to study the effect of
machining parameters on the surface quality of the
machined surfaces, (2) to develop one surface prediction
technique which is termed the multiple regression
prediction model and (3) to evaluate prediction ability of
model.

Figure 1: End milling operation

2. Experimental Setup and Procedure

2.3. Building the multiple Regression model

2.1. Experiment Design
The proposed multiple regression model is a two-way
Experiments have been performed in order to interaction equation:
investigate the effects of one or more factors of the process Y = C + B1 X 1 + B2 X 2 + B3 X 3 + B12 X 1 X 2 + B13 X 1 X 2 + B23 X 2 X 3
parameters (spindle speed, feed rate and depth of cut) on (1)
the surface finish of the machined surface. When an where
experiment involves two or more factors, the factors can Y: surface roughness in μm
affect the response individually or interactally. Often, the X1: spindle speed in rpm
experimental design does not give an idea about the X2: cutting feed in m/min
interaction effects of the factors as in the case of one factor X3: depth of cut in mm
at-a time experimentation, All possible factor level In this model, the criterion variable is the surface
combinations experiments conducted in completely roughness (Ra) and the predictor variables are spindle
randomized designs are especially useful for testing the speed, feed rate, and depth of cut. Because these variables
interaction effect of the factors. Completely randomized are controllable machining parameters, they can be used to
 © 2007 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved - Volume 1, Number 1 (ISSN 1995-6665) 3

predict the surface roughness in milling which will then profilometer to obtain the roughness average value Ra. All
enhance product quality. A commercial statistical package original 84 samples were randomly divided into two data
STATISTICA 6.0 was used to do the regression analysis. sets, training set and testing set. The training set contained
In order to judge the accuracy of the multiple regression 60 samples which were used to build the model and the
prediction model, percentage deviation φi and average testing set contained 24 samples which were used to test
percentage deviation φ were used and defined as the flexibility and the validity of the regression model as
Raim − Raip shown in Tables 1 and 2, respectively. The collected data
φi = × 100% (2) were analyzed using parametric analyses of variance
Raim
(ANOVA) with surface finish as the dependent variable
where φi: percentage deviation of single sample data.
and spindle speed N, Cutting feed F and depth of cut D as
Raim: measured Ra .
independent variables. The ANOVA model was modified
Raip: predicted Ra generated by a multiple regression
to include the main effects of the independent variables
equation.
n
and up to two-variable interactions only. The significance
∑φ i
(3)
level was based on the P-value from ANOVA [13] as
φ = i =1
Insignificant if P > 0.10
n
Mildly significant if 0.05 < P < 0.10
where φ : average percentage deviation of all sample data
and
n: the size of sample data
This method would test the average percentage Significant if P < 0.05 (4)
deviation of actual Ra (measured by an off-line stylus type A statistical model was created by regression function
profilometer) and predicted Ra (produced by the multiple in STATISTICA 6.0 from the training data set. The R
regression model). Square was 0.83879, which showed that 83.879 % of the
observed variability in Ra could be explained by the
independent variables. The Multiple R was 0.9158, which
3. Results and Discussion meant that the correlation coefficient between the observed
value of the dependent variable and the predicted value
After 84 specimens were cut for experimental purposes, based on the regression model was high.
they were measured off-line with a stylus type

Table 1 Effect of cutting parameters on the surface finish of the machined surfaces (training data set)
Cutting parameters R a, Cutting parameters Cutting parameters
R a, R a,
No. N F D μm No. N F D No. N F D
μm μm
rpm mm/min mm rpm mm/min mm rpm mm/min mm
1 750 525 1.25 3.7 21 1500 450 1.25 2.6 41 1000 600 0.75 4.0
2 1250 300 1.25 2.4 22 750 600 0.75 4.5 42 1250 150 1.25 1.7
3 1000 375 0.25 2.6 23 1000 525 0.25 3.8 43 1000 375 0.75 2.6
4 750 600 1.25 4.4 24 750 300 1.25 2.4 44 1250 300 0.75 2.5
5 750 300 0.75 2.6 25 1500 225 0.75 1.9 45 1000 225 0.75 2.4
6 1500 375 1.25 2.5 26 1250 150 0.25 1.2 46 1500 300 0.75 2.1
7 1250 450 1.25 2.3 27 1250 525 1.25 2.5 47 1000 525 0.75 3.9
8 1000 300 1.25 2.3 28 1250 375 1.25 2.5 48 1250 225 0.25 2.1
9 750 150 1.25 1.9 29 1000 225 0.25 2.3 49 1000 150 0.75 1.9
10 1500 600 0.75 2.6 30 1000 450 0.75 3.0 50 1250 375 0.75 2.5
11 1500 450 0.25 3.2 31 1000 600 0.25 4.1 51 1000 150 0.25 1.6
12 1000 450 0.25 4.0 32 1500 150 0.25 1.3 52 1000 225 1.25 2.7
13 750 375 0.75 3.1 33 750 375 1.25 2.6 53 750 225 1.25 2.5
14 1250 600 0.25 3.8 34 1500 525 1.25 3.0 54 1250 450 0.75 2.2
15 1250 225 0.75 2.1 35 1250 300 0.25 2.6 55 1500 300 0.25 2.3
16 1000 150 1.25 1.6 36 1000 300 0.25 3.1 56 750 450 0.25 4.8
17 1000 300 0.75 2.1 37 1500 225 0.25 1.4 56 1250 600 0.75 2.6
18 750 450 1.25 3.3 38 750 225 0.75 2.6 58 750 525 0.25 4.5
19 1500 600 0.25 3.2 39 750 150 0.75 1.7 59 1250 225 1.25 2.4
20 1250 525 0.75 2.5 40 750 525 0.75 4.0 60 1250 150 0.75 1.7
Table 2 Effect of cutting parameters on the surface finish of the machined surfaces (testing data set)
Cutting parameters R a, Cutting parameters Cutting parameters
R a, R a,
No. N F D μm No. N F D No. N F D
μm μm
rpm mm/min mm rpm mm/min mm rpm mm/min mm
1 1000 450 1.25 2.1 9 1500 450 0.75 2.3 17 1500 525 0.75 2.6
2 1500 150 1.25 1.5 10 750 600 0.25 4.7 18 1500 525 0.25 3.1
3 750 300 0.25 3.0 11 1500 375 0.75 2.1 19 1500 600 1.25 3.2
4 750 450 0.75 3.7 12 750 375 0.25 3.2 20 750 150 0.25 1.6
5 1000 375 1.25 2.6 13 1250 600 1.25 3.1 21 1250 450 0.25 2.5
6 750 225 0.25 2.1 14 1250 375 0.25 2.7 22 1500 150 0.75 1.4
7 1500 375 0.25 2.7 15 1000 600 1.25 2.1 23 1000 525 1.25 1.5
8 1250 525 0.25 3.1 16 1500 300 1.25 2.4 24 1500 225 1.25 1.8
4 © 2007 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved - Volume 1, Number 1 (ISSN 1995-6665)

p=.05
6.0
5.5 F 12.82938

5.0
N -7.05602
4.5
4.0
F*D -4.60321
Predicted Values

Factor
3.5
3.0 D -3.90378
2.5
2.0 N*F -3.24292

1.5
N*D 1.920406
1.0
0.5
0 2 4 6 8 10 12 14 16
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0 Effect Estimate (Absolute Value)

Observed Values Figure 3: Pareto Chart of Standardized Effects for surface

Figure 2:Observed vs. predicted values of surface roughness roughness Ra showing significant factors and interactions

The numerical estimates of the effects indicate that the

In Table 3, the coefficients for the independent effect of feed is the largest (12.82) and has positive
variables were listed in the column B. Using these direction. The positive direction means that the surface
coefficients, the multiple regression equation could be finish deteriorated with increasing the cutting feed. This is
expressed as: due to the increase in distance between the successive
R a = 1.178854 − 0.000492 N + 0.009897 F − 0.17625D grooves made by the tool during the cutting action, as the
− 0.000003N × F + 0.000811N × D − 0.003012 F × D (5) cutting feed increases.
The scatterplot between the observed Ra and the Figure 3 shows the effect of spindle speed (-7.05). The
predicted Ra of all 84 samples as shown in Figure 2 negative direction means that increasing the spindle speed
indicated that the relationship between the actual Ra and improves the surface finish. It is generally well known that
the predicted Ra was linear. an increase in cutting speed improves machineability. This
The result of average percentage deviation ( φ ) showed may be due to the continuous reduction in the build up
that the training data set (n=60) was 11.645% and the edge formation as the cutting speed increases.
testing data set (n=24) was 12.134%. This means that the The interaction between the cutting feed and depth of
statistical model could predict the surface roughness (Ra) cut significantly affects the surface roughness (-4.6). The
with about 88.355% accuracy of the training data set and interaction also suggests that to get a certain surface finish
approximately 87.866% accuracy of the testing data set. and maximum metal removal it is preferable to use a high
cutting feed associated with depth of cut.
Table 3: included in the multiple regression model (α=0.05) The depth of cut also has negative value (-3.9), which
Factor Effect Standard Regression P-value indicates that increasing the depth of cut improves the
(Beta) error Coefficient surface finish. The effect of the depth of cut is less
of effect (B) significant on the surface finish.
(SE Beta)
The interaction between the cutting feed and spindle
Intercept 1.178854 0.0000*
N -0.166194 0.173971 -0.000492 0.0000* speed is significantly affecting the surface roughness as
F 1.792732 0.242093 0.009897 0.0000* shown in Figure 3. The figure shows that increasing the
D -0.086891 0.258455 -0.176250 0.0002* spindle speed improves the surface finish as the cutting
N×F -0.803242 0.258455 -0.000003 0.0017* feed decreases. This supports the earlier discussion about
N×D 0.507176 0.242093 0.000811 0.0585** the effect of decreasing cutting speed on the surface
F×D -0.726007 0.173971 -0.003012 0.0000*
roughness of the machined workpieces.
* : Strongly significant
**: Mildly significant
The interaction between the depth of cut and spindle
speed is less significant as shown in Figure 3. The
interaction reveals that increasing the spindle speed and
The analysis indicated that all main factors and their
increasing the depth of cut deteriorates the surface finish.
interactions were highly significant (P<0.05). The
individual effects of various factors as well as their
interactions can be discussed from the Pareto chart
4. Conclusions
illustrated by Figure 3. The length of each bar in the
Pareto chart is proportional to the absolute value of its
A series of experiments has been conducted in order to
associated regression coefficient or estimated effect. The
begin to characterize the factors affecting surface
effects of all parameters and interactions terms, are
roughness for the end-milling process. The effect of
standardized (each effect is divided by its standard error).
spindle speed, feed rate, depth of cut on surface roughness
The order in which the bars are displayed corresponds to
of aluminum samples was studied. The model generated,
the order of the size of the effect. The chart includes a
which includes the effect of spindle speed, feed rate, depth
vertical line that corresponds to the 95% limit indicating
of cut, and the any two-variable interactions, predicts
statistical significance. An effect is, therefore, significant if
surface roughness reasonably well. The deviation between
its corresponding bar crosses this vertical line.
predicted and measured surface roughness values was
within an error band of about 12%. The machining
parameters investigated influenced the surface finish of the
 © 2007 Jordan Journal of Mechanical and Industrial Engineering. All rights reserved - Volume 1, Number 1 (ISSN 1995-6665) 5

machined workpiece significantly. In general, the study

shows that cutting feed is by far the most dominant factor
of those studied. The most important interactions, that References
effect surface roughness of machined surfaces, were
between the cutting feed and depth of cut, and between [1] J. C Chen and R. A. Smith: Journal of Industrial
cutting feed and spindle speed. Technology, vol. 13, No.3, 1997, 15-19.
[2] J. T. Black: Journal of Engineering for Industry, vol.
101, No 4, 1979, 403-415.
[3] P. L. B. Oxley: The Mechanics of Machining An
Acknowledgment
Analytical Approach to Assessing Machinability, Ellins
Horwood Ltd., Chichester West Sussex, England, 1989.
Machine shop and the laboratory facilities at Jordan [4] E. M. Trent: Metal Cutting, Butterworth-Heinemann
University of Science and Technology and the Deutsche Ltd., Oxford. England, 1991.
Forschungsgemeinschaft ‘DFG, are gratefully [5] V. P. Astakhov and M. O Osman: Journal of Materials
acknowledged. Processing Technology, vol. 62, No 3, 1996, 175-179.
[6] V. P. Astakhov: Metal Cutting Mechanics, CRC press,
1999.
[7] E. J Armarego: CSME Forum SCGM A Manufacturing
Science and Engineering, Hamilton, Canada, 1996, 18-
27.
[8] E. J Armarego and R. H. Brown: The Machining of
Metals, Prentice Hall, 1969.
[9] N Tabenkin: Carbide and Tool, vol. 21, 1985. 12-15.
[10] P. Balakrishnan and M. F. De Vries: Analysis of
Mathematical Model building Techniques Adaptable to
Machinability Data Base System, Proceeding of
NAMRC-XI, 1983.
[11] G. Boothroyd and W. Knight: Fundamentals of
Machining and Machine Tools. Second Edition, Marcel
Dekker Inc., New York. 1989.
[12] V. M. Huynh and Y. Fan: The International Journal of
Advanced Manufacturing Technology, vol. 7, 1992. 2-
10.
[13] J. Neter, W. Wasserman and M. Kutner: Applied Linear
Statistical Models, Irwin, Boston, MA, 1990
View publication stats