OPTIMIZATION OF CUTTING PARAMETERS TO

MINIMIZE SURFACE ROUGHNESS OF C45 STEEL IN

TURNING PROCESS ON S530 X 1000 LATHE USING
TAGUCHI METHOD

Written by:
ACHMAD ABIMANYU DJUNAIDI
1201154066

INDUSTRIAL ENGINEERING STUDY PROGRAM

INDUSTRIAL ENGINEERING FACULTY
TELKOM UNIVERSITY
2020
 APPROVAL SHEET

Given Final Project with Title:

OPTIMIZATION CUTTING PARAMETERS TO MINIMIZE

SURFACE ROUGHNESS OF DURAL AND C45 STEEL IN
TURNING FACING PROCESS ON S530 X 1000 LATHE USING
TAGUCHI METHOD

By:
Achmad Abimanyu Djunaidi
1201154066

Has been approved and ratified on the Final Session of Final Project in Industrial
Engineering Bachelor Degree, Industrial Engineering Faculty, Telkom University.

Bandung, 15 January 2020

Approved by,

Advisor 1 Advisor 2

Haris Rachmat, S.T., M.T. Denny Sukma E. A., S.T., M.Sc.

i
 INTELLECTUAL PROPERTY STATEMENT FORM

Name : Achmad Abimanyu Djunaidi

NIM : 1201154066
Address : Taman Puri Bintaro PB 39 No.23,
Bintaro Jaya Sektor 9. Tangerang
Selatan, Banten.
Phone : 081286515740
Email : rakusabie@gmail.com

I declare that this final project is my own original work. Based on this statement, if
there is proof of plagiarism or academic offense, I am ready to burden the
consequences and punishment given to me.

Bandung, 15 January 2020

Achmad Abimanyu Djunaidi

ii
I offer this final project to my father, mother, brothers, and family as a symbol of
appreciation for their support and prayers over the years.

iii
 ABSTRACT

This paper presents the results of experimental work carried out in the process of
turning on the S530 X 1000 Lathe with research objects are C45 Steel. This study
aims to optimize the cutting parameters of the research object in order to minimize
surface roughness using the Taguchi method. In this study, the orthogonal array
used 𝐿9 (33 ) array with 3 levels and 3 factors. The factors used are derived from
cutting parameters such as spindle speed, feed rate, and depth of cut. In this study,
using cutting parameters on C45 Steel objects such as the following: spindle speed
120, 280 and 440 rpm, feed rates 0.05, 0.10, 0.21 mm / rev, and depth of cut 0.5,
0.75, and 1 mm. In this study, using cutting parameters on C45 Steel objects as
follows: spindle speed 855, 1350, and 2000 rpm, feed rates 0.13, 0.17, and 0.21
mm/rev, and depth of cut 0.3, 0.4, and 0.5 mm. The use of Signal to Noise Ratio (S
/ N Ratio) uses a "smaller is better" approach to find optimal experimental results
in minimizing surface roughness. After that, an analysis of variance (ANOVA) test
is performed in order to prove that there are significant differences in each
parameter. The researcher will test a number of experiments, an analysis will be
carried out on each experiment that has been tested and then one of the optimal
experiments will be chosen to minimize surface roughness.

Keywords: Optimize, Taguchi method, Surface roughness, Signal to Noise Ratio,

ANOVA.

iv
 PREFACE

Praise and Gratitude the presence of Allah SWT for all His grace and gifts
so that the writer is given the opportunity to complete this final project. And the
writer also thanks parents, family, friends who have provided assistance in the form
of support and prayer. And the writer thanks to the lecturers and instructors who
have guided the writer in completing the final project, and have provided useful
knowledge. The writer also wants to thank to:

1. Mr. Haris Rachmat S.T., M.T as first mentor who has guided and gave
knowledge in completing the final project.
2. Mr. Denny Sukma Eka Atmaja S.T., M. Sc as second mentor who has
guided and gave knowledge and advice in completing the final project.
3. Mr. Yulis who has given a lesson, knowledge and guidance in completing
the final project.
4. Mr. Rino Andias who have lent a measuring instrument so that the writer
can complete the final project.

The author realizes that in this thesis there are some shortcomings and
errors. So, the authors apologize for the mistakes that have been made. The author
is very open to suggestions and corrections in this thesis so that the final project
becomes better. Hopefully, this final project is not only useful for Telkom
University students, but also for the people of Indonesia.

Bandung, 15 January 2020

Achmad Abimanyu Djunaidi

v
 TABLE OF CONTENT
APPROVAL SHEET ............................................................................................... i
INTELLECTUAL PROPERTY STATEMENT FORM ........................................ ii
ABSTRACT ........................................................................................................... iv
PREFACE ............................................................................................................... v
TABLE OF CONTENT ......................................................................................... vi
LIST OF FIGURES ............................................................................................. viii
LIST OF TABLES .................................................................................................. x
GLOSARRY .......................................................................................................... xi
LIST OF APPENDIXES ...................................................................................... xiii
CHAPTER I INTRODUCTION ............................................................................. 1
I.1 Research Background .................................................................................... 1
I.2 Problem Formulation ..................................................................................... 4
I.3 Research Objectives ....................................................................................... 4
I.4 Research Limitation ....................................................................................... 5
I.5 Benefits of Research ...................................................................................... 5
I.6 Writing Systematics ....................................................................................... 5
CHAPTER II LITERATURE REVIEW ................................................................. 7
II.1 Design of Experiment ................................................................................... 7
II.1.1 Design of Experimental Definition ........................................................ 7
II.1.2 Taguchi Method ..................................................................................... 7
II.2 Parametric Test ........................................................................................... 14
II.2.1 Normality Test ..................................................................................... 15
II.3 Non-Parametric Test ................................................................................... 15
II.3.1 Kruskal-Wallis Test ............................................................................. 16
II.4 Surface Roughness ...................................................................................... 16
II.4.1 Factors That Cause Surface Roughness ............................................... 18
II.5 Turning Process .......................................................................................... 18
II.6 Bench Lathe ................................................................................................ 19
II.7 Predecessor Research.................................................................................. 20
CHAPTER III RESEARCH METHOD ............................................................... 23
III.1 Conceptual Model ..................................................................................... 23
III.2 Problem Solving Systematic ..................................................................... 24
III.2.1 Initial Stage of Research ..................................................................... 25
III.2.2 Data Collection ................................................................................... 27

vi
 III.2.3 Data Processing Stage......................................................................... 29
III.2.4 Analysis and Conclusion .................................................................... 33
CHAPTER IV COLLECTING AND DATA PROCESSING .............................. 34
IV.1 Data Collecting .......................................................................................... 34
IV.1.1 Product Design ................................................................................... 34
IV.1.2 Machining Process .............................................................................. 38
IV.1.3 Parameter Experiment Using C45 Steel as an Object on S530 X 1000
Lathe .............................................................................................................. 43
IV.1.4 Performance Characteristic Identification .......................................... 43
IV.1.5 Surface Roughness Measurement Result ........................................... 44
IV.2 Processing Stage........................................................................................ 45
IV.2.1 Orthogonal Array Determination........................................................ 45
IV.2.2 Signal to Noise Ratio Calculation ...................................................... 46
IV.2.3 Normality Test .................................................................................... 48
IV.2.4 Kruskal-Wallis Test ............................................................................ 50
IV.2.5 Analysis of Variance Calculation ....................................................... 51
IV.2.6 Tukey Comparison Test...................................................................... 53
IV.2.7 Optimum Condition Confirmation ..................................................... 58
CHAPTER V ANALYSIS .................................................................................... 59
V.1 Analysis of DOE Result ............................................................................. 59
V.1.1 Analysis of Signal to Noise Ratio ........................................................ 59
V.1.2 Analysis of Normality Test .................................................................. 59
V.1.3 Analysis of Kruskal-Wallis Test .......................................................... 60
V.1.4 Analysis of ANOVA ............................................................................ 60
V.1.5 Analysis of Tukey Pairwise Comparison ............................................. 60
V.1.6 Analysis of Optimal Condition ............................................................ 60
CHAPTER VI CONCLUSION............................................................................. 62
VI.1 Conclusion................................................................................................. 62
VI.2 Suggestion ................................................................................................. 63
REFERENCES...................................................................................................... 64

vii
 LIST OF FIGURES
Figure I.1 Taguchi and conventional loss functions ............................................... 3
Figure I.2 S530 X 1000 Lathe in Manufacturing Building at Telkom University .. 4
Figure II.1 Taguchi Design Procedure (Zhang et al., 2007) ................................... 8
Figure II.2 Auxiliary diagrams for determination of values of a) Ra and b) Rz
(Burakowski and Wierzchoń, 1999) ..................................................................... 17
Figure II.3 Turning Operation (Groover, 2010) .................................................... 19
Figure II.4 Basics of the turning process normally done on a lathe (Black and
Kohser, 2008) ........................................................................................................ 19
Figure II.5 Bench Lathe Machine (Singh, 2006) .................................................. 20
Figure III.1 Conceptual Model .............................................................................. 23
Figure III.2 Problem Solving Systematics ............................................................ 24
Figure III.2 Problem Solving Systematics (Continue) .......................................... 25
Figure III.3 Surface Roughness Measurement ...................................................... 30
Figure IV.1 Workpiece Size details ...................................................................... 34
Figure IV.2 Workpiece Object .............................................................................. 34
Figure IV.3 C45 Steel Round Bar ......................................................................... 35
Figure IV.4 Mitutoyo SJ-410 Series ..................................................................... 35
Figure IV.5 S530 X 1000 Winho High-Speed Precision Lathe ............................ 36
Figure IV.6 Insert Tools Walter VBMT160408-MM4 WSM20S ........................ 37
Figure IV.7 Product Drawing................................................................................ 37
Figure IV.8 Machining Process ............................................................................ 38
Figure IV.9 Put the raw material (C45 Steel) into the Spindle ............................. 39
Figure IV.10 Facing process ................................................................................. 39
Figure IV.11 Mark off the center of material in the front face of the tailstock .... 40
Figure IV.12 Turning process ............................................................................... 40
Figure IV.13 Cut the material into a specific size ................................................. 41
Figure IV.14 Cut objects that have been processed with a hacksaw .................... 41
Figure IV.15 C45 Steel part .................................................................................. 42
Figure IV.16 Cutting Parameters on S530 X 1000 Lathe ..................................... 42
Figure IV.17 Cutting process and parts that have been cut................................... 43
Figure IV.18 Main Effects Plot for SN Ratios ...................................................... 47

viii
Figure IV.19 Probability Plot of Spindle Speed ................................................... 48
Figure IV.20 Probability Plot of Feed Rate .......................................................... 49
Figure IV.21 Probability Plot of Depth of Cut ...................................................... 50
Figure IV.23 Interval Plot of SNRA1 vs Feed Rate.............................................. 52
Figure IV.24 Interval Plot of SNRA1 vs Depth of Cut ......................................... 53
Figure IV.25 Differences of Means for SNRA1 (Spindle Speed) ........................ 54
Figure IV.26 Differences of Means for SNRA1 (Feed Rate) ............................... 56
Figure IV.27 Differences of Means for SNRA1 (Depth of Cut)........................... 57

ix
 LIST OF TABLES
Table II.1 𝐿9 (34) Orthogonal Array ................................................................... 10
Table II.2 Surface roughness values produced by the various manufacturing
processes (Groover, 2010) .................................................................................... 17
Table II.3 Similar Research................................................................................... 22
Table IV.1 Product Details.................................................................................... 37
Table IV.2 Cutting Parameters and Levels ........................................................... 44
Table IV.3 Surface Roughness Measurement Result ............................................ 44
Table IV.4 Taguchi’s Orthogonal Array L_9 Design ........................................... 45
Table IV.5 Orthogonal Array with Factors ........................................................... 45
Table IV.6 Signal to Noise Ratio Calculation ....................................................... 46
Table IV.7 Response Table for Signal to Noise Ratios ........................................ 47
Table IV.8 Descriptive Statistics .......................................................................... 51
Table IV.9 Kruskal-Wallis Test Result ................................................................. 51
Table IV.10 ANOVA One-way Test SNRA1 Versus Feed Rate ......................... 52
Table IV.11 ANOVA One-way Test SNRA1 Versus Depth of Cut ..................... 53
Table IV.12 Grouping Information Using Tukey Method and 95% Confidence
(Spindle Speed) ..................................................................................................... 54
Table IV.13 Tukey Simultaneous Test for Differences Means (Spindle Speed) .. 54
Table IV.14 Grouping Information Using Tukey Method and 95% Confidence
(Feed Rate) ............................................................................................................ 55
Table IV.15 Tukey Simultaneous Test for Differences Means (Feed Rate) ......... 55
Table IV.16 Grouping Information Using Tukey Method and 95% Confidence
(Depth of Cut) ....................................................................................................... 56
Table IV.17 Tukey Simultaneous Test for Differences Means (Depth of Cut) .... 57
Table IV.18 Optimum Condition of Cutting Parameters ...................................... 58
Table IV.19 Experiment and Prediction Condition .............................................. 58

x
 GLOSARRY

ANOVA : Analysis of Variance

DOE : Design of Experiment

H0 : Null hypothesis

H1 : Alternative Hypothesis

F-value : statistic test used to determine whether the term is

associated with the response.

P-value : a probability that measures the evidence against the null

hypothesis. Lower probabilities provide stronger evidence
against the null hypothesis.

𝑅𝑎 : Average roughness of a surface

𝑅𝑧 : The difference between the tallest "peak" and the deepest

"valley" in the surface.

T−Value : measures the size of the difference relative to the variation

in your sample data.

H-Value : test statistic

Z-Value : The z-value indicates how the average rank for each group
compares to the average rank of all observations.

Mean Rank : the average of the ranks for all observations within each
sample.

Median : the midpoint of the data set.

Mean : the mean of the observations within each group

Adjusted P-Value : The adjusted p-value indicates which pairs within a family
of comparisons are significantly different.

Difference of Means : This value is the difference between the sample means of
two groups.

xi
ABBREVIATION

rpm : Revolutions per minute

mm/rev : Millimeter per revolution

mm : Millimeter

n : Spindle speed

f : Feed rate

ap : Depth of cut

SE : Standard Error

95% Cl : Confidence Interval of 95%

AD : Anderson-Darling

N : the number of nonmissing observations in a sample.

StDev : Standard Deviation

Adj MS : Adjusted Mean Squares

Adj SS : Adjusted sums of squares

DF : Degrees of Freedom

Seq SS : Sequence Sum of Squares

SE of Difference : The standard error of the difference

xii
 LIST OF APPENDIXES
APPENDIX A ....................................................................................................... 68
APPENDIX B ....................................................................................................... 71
APPENDIX C ....................................................................................................... 73
APPENDIX D ....................................................................................................... 75
APPENDIX E ....................................................................................................... 78
APPENDIX F ........................................................................................................ 82
APPENDIX G ....................................................................................................... 84

xiii
 CHAPTER I INTRODUCTION

I.1 Research Background

Manufacturing became a capital-intensive operation in the early part of the 20th
century. A hierarchical system of mass production has substituted more small
batches and make-to-order manufacturing of goods. The 1920s was a turning point.
With increased household income in North America and Europe, large-scale
household appliances and motor vehicles were produced (Benhabib, 2003).
Manufacturing is an important commercial activity carried out by companies that
sell products to customers. The form of the processing carried out by a company
depends on the type of commodity it provides. Let us discuss this partnership by
analyzing the manufacturing industries and the goods they produce (Groover,
2010).

The manufacturing world is not separated by machines. The machine is a valuable

entity that is very important to run the production system in manufacturing.
Machines have an important role in running production, without machinery, the raw
material will not be a product. According to Walsh and Cormier (2006), in
machining, there are machining processes such as drilling, milling, sawing, and
many others. One of them is the turning process, turning is the procedure used to
create existing cylindrical shapes, generally with a one-point cutting tool, by
removing material. Boring is mainly an internal adjustment for the creation of
internal forms. The through rotating machines are engine lathes, single spindle
automatic lathes, horizontal-turret lathes, automatic screw machines, Swiss-type
automatic screw machines, multiple-spindle automatic bar and chucking machines,
and computer-controlled automatic turning centers.

Every factory and company must avoid the production that has defects product that
will become reject items which will harm the company both in terms of material
and cost. A defect is linked to a quality feature that fails to meet certain standards.
In addition, this fact may lead to the unacceptable (or defective) accuracy of one or
more defects in a product or service (Amitava, 2016).

1
There are various kinds of defects that exist in the product, one of which is surface
roughness. Surface roughness is a commonly used product quality index and is a
technical prerequisite for mechanical products in most cases. For the functional
behavior of a part, the achievement of desired surface quality is of great importance.
On the other hand, a simple solution is almost impossible to achieve by the process-
dependent existence of the surface roughness formation mechanism together with
the many uncontrollable factors that influence related phenomena (Benardos and
Vosniakos, 2003).

A product has features that define their quality in relation to customer expectations
or specifications. In these attributes, the performance of a material is calculated
(Ross, 1996). In selecting competing products and services, quality has become one
of the most important factors for consumers. Whether it is individual, industrial,
retail, banks, financial institutions or military advocacy programs. The
understanding and enhancement of quality are therefore key factors that contribute
to business success, growth and increased competitiveness (Montgomery, 1985).

The quality of the products can be seen from a number of aspects, one of which is
surface roughness. According to Ribeiro et al. (2017), the value of the machined
surface is measured by the surface roughness of the machined component, which is
the most important quality characteristic. Nonetheless, the application of
optimization techniques could be an interesting solution for reducing the number of
variations of experimental experiments. Each combination of cutting parameters
can result in different surface roughness and tool life. Nonetheless, it is very
difficult to define the best combination that gives lower surface roughness and total
tool life for many different parameters to monitor with multiple levels for each one.
For the manufacturing industry, surface quality, tool life, and production costs are
the most important features in a certain combination of machining parameters.

2
According to Roy (2010), Figure I.1 displays the conventional method and the loss
function view of Taguchi. The graph shows the loss function in order to deviate
from an ideal or goal value of a particular design parameter. Here is the objective
value or the most desirable value of the considered variable. This parameter could
be a vital aspect, material color, surface finish or any other function contributing to
the customers.

Figure I.1 Taguchi and conventional loss functions

UAL and LAL describe in Figure I.1 the upper and lower acceptable limits of design
parameters. The product is typically theoretically suitable if the parameter value is
within UAL to LAL limits. Nonetheless, the functional failure happens 100%
outside of the cross-country limits as shown, and this element is either discarded,
reworked or saved. This effort is made to monitor the production process in order
to maintain the product within acceptable limits. This function is continuous in
Figure I.1, as shown in the dotted line. If development parameters differ from ideal
or target quality, product performance begins to be affected. High customer
satisfaction will be achieved by producing products that consistently exceed the
target price. It may be worth noting that Taguchi causes more than 100 percent
damage to a customer. If the entire system fails or a system fails catastrophically,
such cases may occur. Therefore, the unique feature of Taguchi's quality control
philosophy is to reduce the gap around the target value (Roy, 2010).

3
In Figure I.2, it is shown a manual turning lathe in the Telkom University
manufacturing building

Figure I.2 S530 X 1000 Lathe in Manufacturing Building at Telkom University

In this study, the experiment using the Taguchi method to minimize surface
roughness for C45 Steel in the turning process. The machines used are in the
manufacturing building, Telkom University. Taguchi Method using several cutting
parameters from S530 x 1000 lathe which parameters will be carried out
experiments to optimize the cutting parameters in minimizing surface roughness.

I.2 Problem Formulation

The problem to be discussed in this project is how to determine the optimal level of
cutting parameters in the turning process on the S530 X 1000 Lathe as to minimize
surface roughness on C45 Steel?

I.3 Research Objectives

The objective of this research is finding the optimum condition of cutting
parameters in S530 X 1000 lathe in order to minimize the surface roughness. The
way to get the optimum conditions is by optimizing the cutting parameters so as to
produce the smallest roughness possible from several experiments conducted.

4
I.4 Research Limitation
1. Minitab 18 as processing statistical data software.
2. Solidworks 2015 as CAD software.
3. The workpiece material is C45 Steel.
4. The output is this research is to determine the optimal choice of experiments
to minimize the surface roughness of the workpieces.

I.5 Benefits of Research

1. As research for Telkom University to make improvements on S530 X 1000
Lathe.
2. By implementing the automation system on S530 X 1000 Lathe, an operator
can produce quality products and reduce defective products.
3. As an input for the industry, specifically in the process of turning on manual
lathe machines.
4. As learning material for students and readers about machining.

I.6 Writing Systematics

The systematics of writing in this study are:
Chapter I Introduction
This chapter contains an explanation of the background of this
research. Furthermore, from the background of this problem a
problem definition, research objectives, research limitation and
benefits of the research.

Chapter II Literature Review

This chapter contains explanations and explanations related to this
research to support research in developing scientific assignments.
Literature studies obtained come from books, journals, and previous
research.

5
Chapter III Research Method
This chapter explains the outline of the research from start to finish
which is described visually as a conceptual model. Then, there is
systematic problem solving contains stages of how researchers do
problem-solving which contain the process of research.

Chapter IV Collecting and Data Processing

In the data, the processing chapter contains data needed to meet
research requirements. In data processing, data processing occurs
from the data that has been collected and then produce experimental
results that will be considered to find the optimal experiment

Chapter V Analysis
This chapter contains the analysis of research based on the Taguchi
experimental results using the ANOVA test and signal to noise ratio.
From all the experimental results, one optimal experiment was
chosen to minimize surface roughness.

Chapter VI Conclusion
This chapter contains a summary of the result of an experiment that
used Taguchi method to find the optimal result of several
experiments that took a variable of parameters n S530 X 1000 Lathe
to be optimized in order to minimize or reduce the surface
roughness.

6
 CHAPTER II LITERATURE REVIEW

II.1 Design of Experiment

II.1.1 Design of Experimental Definition
Experimental design model for agricultural experiments developed by Ronald A.
Fisher from England in 1920. That method then extends in the fields of medicine
and biology. Throughout biology, the characteristics produced by individuals are
important. The contents of the experimental design are therefore a means for
variations to be represented and how these variations can be distinguished. It
includes blocking independent conditional differences, randomization of sample
sequences, repeated experiments, distribution, and sense screening, among others.
The purpose of the experimental design is to find links between various factors
(variables) and their responses (Taguchi et al., 2005).

II.1.2 Taguchi Method

The Taguchi method was later developed by Genichi Taguchi in Japan to improve
the application of total offline quality control. This method deals with finding the
best value of a control factor to make a problem less sensitive to variations in
uncontrolled factors. This kind of problem is called Taguchi robust design problem
(Negrete et al., 2015).

The Taguchi technique is a powerful tool in designing high-quality systems. The

advantage of the Taguchi technique is that it provides a simple, efficient and
systematic approach that aims to optimize a design for performance, quality, and
cost (Basavarajappa, 2007). Taguchi technology helps optimize parameters with the
least number of experimental experiments and the advantage is that research time
is efficient with less effort to provide the best response from input data (Trehan et
al., 2013).

7
The Taguchi method is now widely used in all applications as a powerful tool in
analyzing cutting parameters. This was proven to be successfully used in the
unconventional machining process by Lin et al. A specially designed Taguchi
orthogonal array used in this study aims to investigate the effects of machining
parameters through a small amount of experiment and only a small amount of time
is spent for an experimental investigation (Murthy et al., 2016).

Based on Zhang et al. (2007), the Orthogonal array that exists in the Taguchi
method is used to study the entire experimental process with little experimental
output. The Taguchi design method consists of three stages: system design,
parameter design, and tolerance design.

In Figure II.1, the system design phase consists of determining suitable working
levels of the design factors. The next phase is parameter design, in that phase consist
of five processes such as selecting proper orthogonal array (OA), run experiments,
analyze data, identify optimum condition, and confirmation runs. In the last phase,
there is tolerance design that consists of determining the result of parameter design.

Figure II.1 Taguchi Design Procedure (Zhang et al., 2007)

8
II.1.2.1 Orthogonal Array
In experimental design, orthogonal means balanced. The main features of
orthogonal arrays are flexibility and the ability to set a number of variables. There
is also an important feature of orthogonal arrays is the reproducibility of the
conclusions drawn from small-scale experiments in research and development for
design processes. In Taguchi's methodology, the main role of the orthogonal array
is to allow engineers to evaluate the design of the product in terms of robustness.
Orthogonal means balanced in the design of the experiments. The versatility and
willingness to delegate a number of variables to orthogonal arrays is a key feature
of orthogonal array use. In addition, the reproducibility or repeatability of
conclusions drawn from small-scale trials in research and development based on
process design (Negrete et al., 2015).

According to Asfar et al. (2018), the advantages of orthogonal arrays is that they
can use a variety of factors, use unnecessary experiments and reduce the time and
cost of research. The goal of orthogonal arrays is to support a variety of factors with
a minimum number of tests. The procedure in determining the appropriate
orthogonal adjustment matrix is as follows:

1. Define the Number of Factors and Levels

The stage of determining the number of factors and levels is carried out. In
analyzing parameters in the production process, known parameters and level
measurements are carried out for each factor that exists. Everything happens to
every individual. Interaction can also affect the number of degrees of freedom.

2. Determine the Degree of Freedom

The level of freedom is a concept that describes the research conducted and also the
information obtained from experiments conducted. The form of equality in the
degree of orthogonal matrix freedom, in determining the number of experiments to
be chosen is:

𝑉𝑂𝐴 = (number of trials - 1)

The level of freedom (𝑉𝑓𝑙 ) of a factor is used in calculating the number of levels
agreed or determined by a factor, the form of the equation is:

9
 𝑉𝑓𝑙 = (number of levels - 1)

The equality of degrees of freedom is to study the freedom of an experiment or the

total degrees of freedom are:

Total 𝑉𝑓𝑙 = (many levels). (𝑉𝑓𝑙 )

3. Choosing the Orthogonal Matrix Table

In choosing the right orthogonal matrix array table, we need an orthogonal matrix
equation that contains the number of factors, level, and a number of assessments.
The general form of the orthogonal matrix model is:

𝐿𝑎 (𝐵 𝐶 )

Where is the Latin square, is a lot of experiments (lines), b is many levels, and c is
many columns (factors).

According to Mason et al. (2003), in experiments where 4 factors of 3 levels (1, 2,

and 3) must be investigated each, the 𝐿9 (34 ) array in Table II.1 can be used. There
are only nine test runs in the experimental design. It is an experiment with 34−2
fractions and Resolution-III. These layouts are usually labeled as 𝐿𝑛 (3𝑘 ), where
the number of test runs n is about p of 3, and k = (n − 1)/2 is the greatest number of
factors to be examined using the design layout.

Table II.1 𝐿9 (34 ) Orthogonal Array

Run No. Factor A Factor B Factor C Factor D

1 1 1 1 1
2 1 2 2 2
3 1 3 3 3
4 2 1 2 3
5 2 2 3 1
6 2 3 1 2
7 3 1 3 2
8 3 2 1 3
9 3 3 2 1

10
II.1.2.2 Signal Noise to Ratio
According to Singh et al. (2016), the Signal-to-noise ratio (summarized SNR or S /
N) is a metric that contrasts a target signal with a background noise level in science
and engineering. A strength metric used to recognize control factors (signal factors)
that minimize product or system variance by reducing the effects of uncontrollable
factors (noise factors, such as voltage fluctuations, fatigue, temperature, humidity)
is used in Taguchi designs.

Taguchi's experimental design uses these signal-to-noise ratios to classify the

control factors minimizing product or system variance by reducing the impact of
uncontrollable (i.e. noise factors). During the experiment, the noise factor is
controlled to force variance and the results define optimal control factors that
robustly resist variations of the noise factor in the system or in the material.

The three signal-to-noise ratios that are commonly used in optimization process are:

1. Smaller the better: It is used where smaller value of the response is desired.
1
S/N ratio (ƞ) = − 10log10 𝑛 ∑𝑛𝑖=1 𝑦𝑖 2

2. Larger the better: It is used where larger value of the response is desired.
1 1
S/N ratio (ƞ) = − 10log10 𝑛 ∑𝑛𝑖=1 𝑦𝑖 2

3. Nominal the best: It is used where the nominal or target value and variation
about that value is minimum.
𝜇2
S/N ratio (ƞ) = − 10log10 𝜎2

Where, 𝛾 = 𝑖 𝑡ℎ observed response value and 𝑛 = number of observations

in a trial, 𝜇 = average of observed response values and 𝜎 = variance (Mazumder et
al., 2016).

11
II.1.2.3 Analysis of Variance (ANOVA)
Based on Sutoni et al. (2018), Variance analysis was first introduced by Sir Ronald
Fisher, a British statistician. Variance analysis is a method that divides variance
into a source of variance that can be identified and is a collection of degrees of
freedom in experiments. The data were taken, both the actual condition data and the
experimental data in robust design can be divided into three types, namely:

1) Variables, namely data that can be accounted for during measurements

on a continuous scale.

2) Attributes, namely data from experiments that have characteristics that

are not continuous but can be classified on a discrete scale.

3) Digital, which is data that has a value of 0 or 1.

ANOVA in the Taguchi method is used as a statistical method for interpreting

experimental data. Analysis of variance is a calculation technique with quantitative
methods to estimate the contribution of each factor to each response that has been
measured which helps the process of identifying the contribution of factors so that
the estimated accuracy of the model can be determined. The classification of
experimental results is carried out statistically according to the source of variation.
ANOVA for orthogonal matrices is based on the calculation of the number of
squares for each column (Asfar et al., 2018).

Based on Ross (1996), one-way ANOVA is the next most complicated ANOVA to
be used, in comparison to no way ANOVA, where parameters have not been
managed, the influence of one controlled variable on the output of a ma terial or
system.

According to Murray (2017), one-way analysis of Variance or One-way ANOVA

is a tool to figure out whether there is a variance in meaning between two or more
independent groups when the outcomes for a particular categorical factor are
specified in groups. In its aim, it is essential to extend the independent sample test
by means of a difference, extended to more than two groups. Assume the number
of groups is K bis 2. In the null hypothesis, the outcomes of all groups are the same.

12
At least one method varied from another medium according to the alternative
hypothesis.

The following assumptions are necessary for the one-way ANOVA test:
1. Random assignment: The classes to which the results belong are determined
independently of the parameter of the result.
2. Independent groups: Findings of one group are independent of findings in
all other groups. Therefore, there must be separate sampling units in each
band.
3. Homogeneity of variance: The sample variants of the result factor for the
different groups are the same, although the sample mean may be different.
4. Central Limit Theorem assumptions: The scale of the test is sufficiently
large and/or the population distribution is standard for the result parameter.

II.1.2.4 Tukey Method

Based on Dubitzky et al (2013), the Tukey method defines the particular means that
are significantly different from the set of ways. The Tukey test is a conditional
reference measure and is valid when more than two measures are measured (for two
ways, use the t scale). Usually, the Tukey method is used when (Variance Analysis)
has shown that there is a significant difference and defines where the difference
exists. The Tukey check is determined by contrasting all factors in pairs. Clearly, a
significant difference is shown when the pairwise discrepancy between the two
measures exceeds the value measured as:

𝑀𝑆
𝐻𝑆𝐷 = 𝑞 √
𝑛

Tukey confidence intervals are a group of parallel confidence intervals that carry
up to 1 − 𝛼. Tukey's approach is a very restrictive technique relative to Fisher's
LSD because it needs a greater observable disparity in patient rates in addition to
disable a particular mean set. Minitab applies the Tukey protocol and records the
effects on the confidence interval. Remember that, like Fisher's LSD, Tukey's

13
approach shows that all mean pairs are distinct except at 10% and 15%
concentrations (Montgomery and Runger, 2003).

II.1.2.5 Degree of Freedom

According to Ross (1996), degrees of freedom are defined as the number of
relationships between process parameters needed to determine the best level of an
experiment. Degrees of freedom are defined as the number of process parameters
minus one which in this analysis three-level process parameters are calculated for
two degrees of freedom because interactions between cutting parameters are
ignored. To gain an understanding of the relation between a variable with different
quality characteristics rate, the measurement of degrees of freedom takes place.
This correlation alone provides information on variables and rates that affect
qualitative features considerably.

Experiments can extend the definition of the degree of freedom; n x r studies have
been carried out in many experiments and other trials. The number of freedom
degrees is:

Db = nr – 1

With Db is the degree of freedom, n is the number of experiments, and r is the

number of replications (Asfar et al., 2018).

II.2 Parametric Test

Parametric test is more reliable and mostly require fewer data than nonparametric
tests to draw a stronger conclusion. However, 3 data parameters must be true or
assumed in order to use a parametric test. First of all, the data must usually be
distributed, so all the data points must adopt a bell-like curve without skewing
above or below the typical results. The levels of Ca-125 are a non-normal example
of data. Normal Ca-125 values range between 0 and 40 in the general population.
The mean is 15, contributing to a skewed distribution rather than to a natural one.
The data must also have the same variance and the same default (Kaur and Kumar,
2015).

14
II.2.1 Normality Test
Normality tests are analogous to the statistical measurement of normality. The core
criteria for the normality determination are the Kolmogorov-Smirnov (K-S) test,
Lilliefors corrected K-S test, Shapiro-Wilk test, Anderson-Darling test, Cramer-von
Mises test, D’Agostino skewness test, Anscombe-Glynn kurtosis test, D’Agostino-
Pearson omnibus test, and the Jarque-Bera test. The experiments described above
equate the scores in the study to a normally distributed range of scores with the
same mean and standard deviation; the null hypothesis is that "the distribution of
the sample is normal." If the result is relevant, the distribution is not normal. For
small sample sizes, normality checks have little power to reject the null hypothesis
and therefore small samples usually pass normality tests. The Shapiro-Wilk test is
based on the similarity between the results and the resulting standard scores and
provides better control than the K-S method even after the Lilliefors correction.
Power is the most common indicator of the importance of the normality test— the
ability to detect whether a sample originates from a non-normal distribution. Some
researchers are recommending the Shapiro-Wilk test as the best test option for
either (Ghasemi and Zahediasl, 2012).

II.3 Non-Parametric Test

Based on Kaur and Kumar (2015), Non-parametric covers techniques not dependent
on data from any given distribution. In other terms, a feature on a sample that is not
contingent on the parameter can be alluded to in non-parametric experiments, the
definition of which relies on the population following certain parametrical
distributions. For statisticians and decision-makers, non-parametric research plays
a central role. Statistical hypotheses include the conduct of observable random
variables among various notable researchers. For example, the hypothesis that (a)
the normal distribution has a specified mean and variance is statistical; (b) that there
is a specified mean but an unknown variance; (c) that a distribution is of normal
form, and that the mean as well as variance are unspecified.

15
II.3.1 Kruskal-Wallis Test
The seminal work of Kruskal and Wallis (1952) provided us with a rigorous rank-
based test for the k-sample problem, complementing parametric methods such as
the one-way variance analysis (ANOVA). In the classical K-sample problem, the
data is well categorized into different categories or classes. Nevertheless, in many
current scientific experiments, the categorical variables are not inherently
deterministic, and the complexities are quantitatively represented by the
distribution of probabilities across attributes. These classification issues also occur
in biomedical and bioinformatics applications where data mining methods and
classification algorithms are used to achieve membership level probabilities (Acar
and Sun, 2012).

II.4 Surface Roughness

The roughness of the surface is a mode of inequalities which normally differs less
from peaks and valleys than in waves. Roughness is defined as a set of asperities of
the actual surface, conventionally described as deviations of the measured profile
from the reference line within the limits of the length of which the wavelength is
not taken into account. The following surface roughness parameters by ISO 4287/1-
1984 (E/F/R) are defined. where represents absolute values of distances between
profile points and the centerline along a length L of the measured surface.

16
 Figure II.2 Auxiliary diagrams for determination of values of a) Ra and b) Rz
(Burakowski and Wierzchoń, 1999)

𝒊=𝒏
𝟏
𝑹𝒂 = ∑|𝒚𝒊 |
𝒏
𝒊=𝟏

The middle line m of a profile is known as the roughness profile line so that the
sum of the squares is minimal; this line is oriented according to the general profile
directions; – 10 points of roughness 𝑅𝑧 , the mean distances of five peaks at five
lowest valleys at length l of a simple interval, are, I am, meaning the roughness
profile of the graph (Burakowski and Wierzchoń, 1999).

Based on Groover (2010), the surface roughness expected from different

manufacturing processes is indicated in Figure II.3. The manufacturing process
defines the surface finish and the quality of the product. Many systems are able to
produce stronger surfaces than others. In addition, processing costs rise with better
surface finishing. This is due to additional operations and more time to get ever
better surfaces. The processes used for superior finishing include finishing,
finishing, polishing, and superfinishing.

Table II.2 Surface roughness values produced by the various manufacturing

processes (Groover, 2010)

17
II.4.1 Factors That Cause Surface Roughness
According to Khandey (2009), when two surfaces come into contact, the quality of
the assemblies plays an important role in the performance and wear of the assembly
parts. The height, shape, arrangement, and direction of these surface irregularities
on the workpiece depend on a number of factors, such as:

A. The machining factors that are used:

1. Cutting speed
2. Feed, and
3. Depth of Cut

B. The geometry of the tool:

1. Nose radius
2. Rake angle
3. Side cutting edge angle, and
4. Cutting edge

C. Combination of workpiece and tool material and mechanical properties

D. Quality and type of the machine tool used,
E. Auxiliary tooling, and lubricant used, and
F. Vibrations between the workpiece, machine tool, and cutting tool.

II.5 Turning Process

According to Groover (2010), Turning is a machining process in which a single-
point tool removes material from the surface of a rotating workpiece. The device is
fed in a linear direction to create a cylindrical shape, parallel to the rotation axis.
Twisting and other machining activities using single point machines. Turning is
traditionally done using a machine tool called a lathe, which gives the power to turn
the part at a certain turning speed and feeding the tool to a specified speed and
cutting depth. It is shown in Figure II.3.

18
 Figure II.3 Turning Operation (Groover, 2010)

Figure II.4, it showed the picture of the turning process normally done on a lathe.
Turning is the machining method of outside conical and cylindrical surfaces.
Usually, it is done on a lathe called machine tool. The workpiece is turned, the one-
point cutting machine is fed into the workpiece longitudinally and then shifts
parallel to the direction of rotating the workpieces to the diameter by the cutting
length. The tool feeds at a rate of fr, speed of cutting, V, depending on the turning
per minute (rpm) and the workpiece diameter (Black and Kohser, 2008).

Figure II.4 Basics of the turning process normally done on a lathe (Black and
Kohser, 2008)

II.6 Bench Lathe

This is a small lathe that is normally placed on a desk. It has almost all the parts of
the engine lathe or speed lathe, and it conducts almost all the operations. This is
used for small and detailed research (Singh, 2006).

19
 Figure II.5 Bench Lathe Machine (Singh, 2006)

II.7 Predecessor Research

According to Saravanakumar et al. (2016) Throughout his work on the optimization
of machining composite parameters using the Taguchi Surface Roughness process,
the aim is to reduce or remove surface roughness by optimizing cutting parameters
such as spindle speed, feed rate, and wt during drills for the Al6063/Al2O3/Gr
hybrid composite.

Moreover, in the research conducted by Ribeiro et al. (2017) entitled Optimization

of Cutting Parameters to Minimize the Surface Roughness at the End Milling
Process Using the Taguchi Method, which is intended to optimize the quality of the
surface in the CNC final milling operation. This research used the Taguchi approach
to determine an appropriate range of cutting parameters combinations, such as
spindle speed, feed rate and depth of cut on the object steel block with tungsten
carbide coated equipment.

Further Research is Osman et al. (2017) entitled Effect of Cutting Parameters on

Surface Roughness in Dry Drilling of AISI D2 Tool Steel by Using Taguchi Method
by Using Taguchi Method, shows the effect of drilling equipment using the Taguchi
process that used drilling parameters to achieve minimum surface roughness in dry
drilling.

20
Ibrahim et al. (2017) entitled Analysis of Surface Roughness Value When Drilling
Magnesiumaz31 Using Taguchi Method, the aim of this paper is to analyze the
roughness of the surface. Experimental trials were conducted at rotational cutting
rates of 635, 970 and 1420 rpm, feed rates of 0.10, 0.18 and 0.24 mm / rev, tool
diameters of 10, 12 and 14 mm. At a cutting rotation of 970 rpm, a feed rate of 0.10
mm / rev and a tool diameter of 14 mm, the minimum surface roughness was
achieved. It can therefore be stated that the selection of a low feed rate factor
produced a low surface roughness value. Another hand, using a high rotation of
cutting, resulted in a low surface roughness.

Zerti et al. (2017) entitled Taguchi Design of Experiments for Optimization and
Modeling of Surface Roughness When Dry Turning X210Cr12 Steel, Approach for
the determination of optimum machining parameters leading to minimum surface
roughness by the Taguchi method. The research used a mixed orthogonal array for
turning operations. The optimization results showed that the best surface roughness
is obtained by using a low feed rate and a broad nose radius centered on the
"smaller-is-the-better" method. The results of this study reveal that the feed
intensity (f) and the nose radius (r) have the most significant effect accompanied by
the surface roughness relationship (f x ap).

Selvaraj (2017) entitled Optimization of Surface Roughness of Duplex Stainless

Steel in Dry Turning Operation Using Taguchi Technique, It illustrates the work
that optimizes surface roughness in the dry-turn process of ASTM A 995 Grade 5A
stainless steel duplex alloy steel. In this analysis, the chosen cutting parameters are
the following: cutting speed, feed rate and cutting depth. Scientists use the Taguchi
approach to evaluate test data in order to consider suitable experimental methods
for reducing surface roughness.

21
 Table II.3 Similar Research
Researcher

A. Saravanakumar, et M.H. Osman, et Gusri Akhyar Oussama D, Philip Research

Characteristic João Ribeiro, et al
al al Ibrahim, et al Zerti, et al Selvaraj Now

2016 2017 2017 2017 2017 2017 2019

Analysis - - v v - v -
Topic Optimization v v - - v v -
Design of Experiment v v v - v v v
SPSS - - - - - - -
Matlab - - v - - - -
Software
Minitab v - - v - - v
Qualitek-4 - - - - - v -
Taguchi v v v v v v v
ANOVA v v v v v v v
Method Signal Noise to Ratio v v v v v v v
Orthogonal Array v v v v v v v
Regression Analysis - - - - v - -
Optimization v v - v v v v
Identification - - - v - - -
Objective
Minimize Defect - v - - v v v
Quality Product - v - - - - v
Feed Rate v v v v v v v
Spindle Speed v v v v v - v
Cutting Speed - - - - - v -
Parameter
Depth of Cut - v - - - v v
Drilling Tools - - v - - - -
Diameter of drill - - - v v - -
CNC - v - - - v -
Lathe Machine - - - - - - v
Drilling Tools - - v - - - v
Object
Composite v - - - - - -
Steel - - v - v v v
Magnesium - - - v - - -

22
 CHAPTER III RESEARCH METHOD

III.1 Conceptual Model

To understand more deeply about this research, a conceptual model is needed which
contains abstract and general concepts. The conceptual model describes the outline
from input to output which contains a combination of several concepts into a
scheme or diagram. in chapter III this conceptual model is described in the form of
a diagram of the outline of this research in Figure III.1.

Figure III.1 Conceptual Model

23
III.2 Problem Solving Systematic
In this chapter, systematic steps for solving problems can be seen in Figure III.2
which contains several stages, namely the initial stage research and data collection
and processing stages. At the collection stage and data processing consists of 3
stages, namely; initial Stage of Research, Data Collection and Data Processing
Stage, Analysis and Conclusion.

Figure III.2 Problem Solving Systematics

24
 Figure III.2 Problem Solving Systematics (Continue)

III.2.1 Initial Stage of Research

The first stage in problem-solving systematics consists of 5 stages, namely the
formulation of the problem, determination of research objectives, determination of
the problem boundaries, selection of research methods, and design of the conceptual
model. The following is an explanation of each stage:

25
III.2.1.1 Formulation of the Problem
In this stage, the researcher formulates the problem that will be discussed in this
study in which the problem under study is a problem that should be used as research
material or not. In the formulation of the problem, the problem to be raised is the
surface roughness of the C45 Steel workpiece on the S530 X 1000 Lathe

III.2.1.2 Determination of research objectives

In determining research objectives must be based on the issues raised. This stage
explains the purpose of the research which aims to provide a solution to the
problem. In this research, the purpose carried out in this study was to reduce defects
in C45 Steel. The product defect is surface roughness. The researcher used the
Taguchi method to solve problems.

III.2.1.3 Determination of the research limitations

In determining the problem boundary, the boundary used must relate to the problem
discussed. Limitations of the problem must be specific and accurate to the research
problem so that with specific limitations it can support research in solving existing
problems. In applying problem boundaries, research must focus on the main
problem and the problem object does not widen.

III.2.1.4 Selection of research methods

At this stage, the selection of research methods is not arbitrary in choosing the
method. The method chosen must be able to provide a solution to the problem and
the researcher understands the method chosen. In the selection of research methods,
which are useful in helping to solve problems and provide solutions in compiling
research assisted by data that has been collected and several references.

III.2.1.5 Design of conceptual model

The conceptual model must contain the parameters that have been selected, then
proceed with the method used and the processes that exist in that method until a
solution is obtained. In designing conceptual models, it serves to display visual
models in the form of diagrams that explain the outline of the research.

26
III.2.2 Data Collection
At this stage, data collection is carried out both from books, journals, and
experiments in order to obtain the data needed to conduct research. the data that has
been collected will be used to be processed into a material to find solutions to the
problems that exist in this study. the data needed include cutting parameters,
roughness averages (𝑅𝑎 ), and others.

III.2.2.1 Product Design

The process of integrating all specific design activities spanning from the
identification of product specifications to the creation of the model, and eventually
the conversion of the idea into a production system (Risitano, 2011). At this stage,
designing products with C45 Steel material in order to get a surface size that will
later be measured surface roughness on objects. In designing this product, it was
assisted with the Solidworks 2015 software. At this stage consists of research
materials and research equipment. In the research material contains materials or
product objects, namely C45 Steel, then the resarch equipment describes the
equipment used in research.

III.2.2.2 Machining Process

At the machining stage, the C45 Steel material which is still in the form of a
cylindrical rod is first carried out turning process with the aim of cleaning the outer
surface of the material. After that, the material is cut into 9 parts. Furthermore, the
parts that have been cut are machined to a size that has been set to produce a
measurement area for surface roughness later. Insert tools used in the machining
process are Walter VBMT160408-MM4 WSM20S which specialized in turning
process.

III.2.2.3 Parameters Selection and Their Levels

The selection of the parameters to be examined relies on the characteristics of the
material or system quality or the response(s) of concern. The consumer who
actually utilizes the item wants or requires some kind of product feature. Several
methods are useful for determining which factors should be included in the initial

27
experiments like brainstorming, flowcharting, and cause-effect diagrams. The level
numbering system is not so urgent, two solutions may be regarded. In the case of
continuous factors, the first level may be the lower of the two values being tested
and the second level the higher. It provides an implicit connection between the
quality of the variable or function and the levels (Ross, 1996).

III.2.2.4 Parameter Experiment Using C45 Steel as an Object on S530 X 1000

Lathe
At this stage, an experiment with cutting parameters was carried out. This is done
with the aim of getting a smooth surface that can be measured by a surface
measurement detector. Because if the result is still rough, it cannot be measured by
a roughness gauge. If the parameters can’t be measured, then the experiment must
be repeated using different parameters. This experiment was carried out on c45 steel
as an object with the S530 X 1000 Lathe. The parameters tested were spindle speed,
feed rate, and depth of cut. The three parameters are adjusted to the cutting
parameters of the S530 X 1000 lathe.

III.2.2.5 Performance Characteristics Determination

The next stage is to identify the performance characteristics of the S530 X 1000
Lathe. The number of variable factors depends on the variable selected. In this
study, the factors or parameters chosen were depth of cut, spindle speed, and feed
rate. This study uses 3 levels and 3 Taguchi factors. The following are the selected
cutting parameters:

1. Spindle Speed
Spindle speed chosen by researchers is 855, 1350, and 2000 rpm. 855 rpm
for level 1, 1350 for level 2, and 2000 for level 3. The reason researchers
chose these three numbers is because the researchers wanted to produce a
surface with minimal roughness so that the choice of spindle speed number
was large. These three numbers were chosen because they have a surface
that is still relatively smooth and can be measured.

2. Feed Rate

28
 Feed rate chosen by researchers is 0.13, 0.17 and 0.21 mm/rev. 0.13 mm/rev
for level 1, 0.17 mm/rev for level 2, and 0.21 mm/rev for level 3. The reason
researchers chose these three numbers is because the researchers wanted to
produce a surface with minimal roughness so that the choice of feed rate
number was not too large. These three numbers were chosen because they
have a surface that is still relatively smooth and can be measured.

3. Depth of Cut
The depth of cut chosen by researchers is 0.5, 0.4 and 0.3 mm. 0.5 mm for
level 1, 0.4 mm for level 2, and 0.3 mm for level 3. The reason researchers
chose these three numbers is because the researchers wanted to produce a
surface with minimal roughness so that the choice of depth of cut number
was not too large. These three numbers were chosen because they have a
surface that is still relatively smooth and can be measured.

III.2.3 Data Processing Stage

At this stage of the data processing stage, the stages of the research will be discussed
starting from orthogonal array identification to optimum condition confirmation.

III.2.3.1 Orthogonal Array Identification

Orthogonal arrays are used to determine the number of experiments involving
several variable factors that affect parameters. The way orthogonal arrays work lies
in the combination of levels of the input variables that have been selected for each
experiment. Researchers used 3 levels and 3 factors (spindle speed, feed rate, and
depth of cut) to produce an array of 33 , so that it results in 9 experiments or also
called 𝐿9 orthogonal arrays.

III.2.3.2 Surface Roughness Measurement

After turning the C45 Steel and selecting cutting parameters, the next step is to
measure surface roughness using Mitutoyo SJ-410 Series Portable Surface
Roughness Tester that shown in Figure IV.4. Data measurements were performed

29
9 times in accordance with 𝐿9 orthogonal arrays. The following are the steps for
measuring surface roughness on C45 Steel:

Figure III.3 Surface Roughness Measurement

1. Put parts on a stand so that the surface is flat

2. Position Mitutoyo tester parallels to the part. Position the measuring
needle with the surface of the part with precision
3. On the Mitutoyo monitor, move the button until the green mark on the
monitor. Then run measurements on the monitor Mitutoyo roughness
tester
4. After that, a graph and roughness average (Ra) appears on the monitor
screen.

III.2.3.3 Signal to Noise Ratio Calculation

In this segment, we calculate the Signal to Noise Ratio for each roughness criterion
based on cutting parameters. According to Zerti et al. (2017), the S/N ratio is
defined as the desired signal relative to the experimental data's unwanted and
random noise quality. Three types of performance analysis in the S / N ratio are
carried out with the smaller the better in order to minimize the surface roughness of
C45 Steel. The optimum level of the system variable for each of these classes is the
level that provides the biggest comparison of values S / N.

30
III.2.3.4 Normality Test
Based on Ghasemi and Zahediasl (2012), the experiments described above equate
the scores in the experiment to a normally distributed collection of scores with the
same mean and standard deviation; the null hypothesis is that "the distribution of
the sample is usual." If the result is relevant, the distribution is not regular. With
small sample sizes, normality checks have little power to reject the null hypothesis
and therefore small samples are most likely to pass normality tests. In the case of
large sample sizes, significant results would be obtained even in the case of a slight
deviation from normality, although this small deviation will not impact the findings
of the parametric analysis. The K-S experiment is an empiric distribution function
(EDF) in which the hypothetical cumulative distribution function of the test
distribution is contrasted with the EDF results.

III.2.3.5 Parametric
According to Mircioiu and Atkinson (2017), parametric methods are based on a
normal or Gaussian distribution, characterized by mean and standard deviations.
The distribution of the tests was symmetrical around the average, with 95% of the
findings within two standard deviations of the norm. Parametric metrics are used
with constant, period data showing equivalent periods or variations.

III.2.3.6 Non-Parametric
According to Mircioiu and Atkinson (2017), nonparametric statistics are not
focused on such parameterized probability distributions or on any conclusions
about the probability distribution of the results. The distribution of the tests was
symmetrical around the average, with 95% of the findings within two standard
deviations of the mean. Non-parametric approaches are used for ordinal data, such
as Likert scale results, which includes the calculation of "larger" or "smaller," i.e.
value classification.

31
III.2.3.7 Kruskal-Wallis Test
Based on Ostertagová et al. (2014), the Kruskal-Wallis method (Kruskal and Wallis
1952, 1953) is a non-parametric version to one-way ANOVA and is used to check
whether the specimens come from the same distribution. The statistical result for a
one-way variance calculation is determined as the proportion of the procedure sum
of the squares to the residual total of the squares. The Kruskal-Wallis experiment
follows the same approach but, as with many nonparametric experiments, the
information rankings are used instead of the raw data.

III.2.3.8 Analysis of Variance Calculation

According to Rutherford (2011), in terms of test conditions (model) and error,
ANOVA attempts to explain data (dependent variable scores). The researcher who
uses ANOVA usually wants to determine the variable score dependent on the
experimental condition. There is also interest in how much of the variation can be
attributed, as defined by the Independent Variable(s), to differences between
specific experimental groups or conditions. Type of Anova test used in research is
One-Way Anova test that compares the methods obtained from various groups of
dependent variable scores.

III.2.3.9 Tukey Comparison Test

The study measures the gap between each pair of means with an acceptable
modification for multiple tests. The findings are displayed as a matrix that displays
the outcome for each set, either as a P-value or as a confidence interval. The Tukey
multiple comparison method, like both the t-test and the ANOVA, suggests that
results from different groups derive from samples where the findings have a normal
distribution and the standard deviation is the same for each group. Most statistical
services provide a Tukey conditional comparison check as an alternative while
running a one-way ANOVA study, e.g. this method is included in SPSS and
Minitab (Olleveant et al., 1999).

32
III.2.3.10 Optimum Condition Confirmation
After the optimal level, the process parameters are selected. The final step is
confirming the optimal conditions of experiments that have passed the Signal to
Noise ratio test and the One-Way Anova test. Under optimal conditions, cutting
parameters and their levels will be confirmed to produce minimal surface
roughness.

III.2.4 Analysis and Conclusion

At this stage, leave conclusions from the results of research that has been done and
suggestions for research.

III.2.4.1 Analysis of DOE Result

At this stage, an analysis of the experimental results that have been tested and
confirmed is done. The results of the analysis are used to find out whether there are
still deficiencies and also the impacts resulting from the study. If there are still
deficiencies, then improvements are made.

III.2.4.2 Conclusion of the Result

The last stage, it is the conclusion of the results of the experiment. The conclusion
contains an overview of the results of experiments that have been selected and
confirmed, how these results occur until it can be confirmed and applied to the
machine.

33
 CHAPTER IV COLLECTING AND DATA PROCESSING

IV.1 Data Collecting

IV.1.1 Product Design
The final product is as shown above. In figure IV.6, showing the size of parts with
an outer diameter of 38.40 mm and an inner diameter of 32.60 mm and a length of
parts of 62.80 mm. In Figure IV.7, it has shown the 3-dimension model of part of
C45 Steel that has been done turning process.

Figure IV.1 Workpiece Size details

Figure IV.2 Workpiece Object

IV.1.1.1 Research Material

1. C45 Steel Round Bar
The workpiece material in Figure IV.3 is C45 Steel Round Bar. This object
is still a raw material with specifications in Figure III.2 is 50 mm in
diameter, with a length of 32,5 cm.

34
 Figure IV.3 C45 Steel Round Bar

IV.1.1.2 Research Equipment

Equipment as research support tools that are used are as follows:
1. Minitab 18 Software
Minitab software is a data processing application that deals with statistics
used by researchers in analyzing research data using the Taguchi, Signal to
Noise Ratio and Anova Test methods.

2. Mitutoyo SJ-410 Series Portable Surface Roughness Tester

Mitutoyo SJ-410 Series shown in Figure IV.4 is a surface roughness
measuring device used to measure the roughness of a surface on the object.
This portable surface roughness tester has been equipped with several
features, such as analysis functionality, Color-graphic LCD, High accuracy
measuring etc.

Figure IV.4 Mitutoyo SJ-410 Series

35
 The measuring distance on this device is around 800μm/0.01μm,
80μm/0.001μm, and 8μm/0.0001μm. This tool has a traverse length measure
with a size of 0.3μm/25mm.

3. Solidworks 2015
Solidworks is a CAD software used to draw a part by displaying a precise
size with a predetermined unit. The appearance of parts in Solidworks can
be 2 dimensions (2D) and can also be 3 dimensions (3D) with ANSI
(American National Standards Institute) standards or others.

4. S530 X 1000 Winho High-Speed Precision Lathe

The machine used in the study was the S530 X 1000 High-Speed Precision
Lathe shown in Figure IV.5, with the Winho brand. This machine is in the
Telkom University manufacturing process building. The S530 X 1000 has
a spindle speed specification of 40 ~ 2000 rpm, and is used for turning,
facing, threading, chamfering, etc.

Figure IV.5 S530 X 1000 Winho High-Speed Precision Lathe

5. Insert Tools Walter VBMT160408-MM4 WSM20S

Insert tool shown in Figure IV.6 used in this study is Walter made in
Germany with series VBMT160408-MM4 WSM20S. This insert tool
specializes in turning process. The tools have a 35-degree angle and it is
suitable for iron, stainless steel, and high temp alloys.

36
 Figure IV.6 Insert Tools Walter VBMT160408-MM4 WSM20S

The final product is as shown in Figure IV.7, showing the size of parts
with an outer diameter of 38.40 mm and an inner diameter of 32.60 mm
and a length of parts of 62.80 mm.

Figure IV.7 Product Drawing

Table IV.1 shows the details on the product insert tools which contains a
description of the product size. This detail consists of cutting-edge length,
corner radius, feed per revolution, and depth of cut which each description
contains a value in millimeter.

Table IV.1 Product Details

Description Symbol Value

Cutting edge
𝑙 16.61 mm
length

Corner radius r 0.8 mm

0.12 - 0.3
Feed per revolution f
mm
Depth of cut 𝘢𝑝 0.5 - 2 mm

37
IV.1.2 Machining Process
Before proceeding to the data processing step, the required research objects needed
in processing data. In Figure IV.8, the raw material that has been collected is first
processed into materials needed by researchers. Then, run the facing process and
mark off the front of tailstock to make a center hole with center drill. After that, cut
material in lathe but not properly because of limited tools. The next step is to cut
material with a hacksaw to cut the rest of the material, so that it can separate from
other parts. Furthermore, run turning process again to leave a surface on parts. Then,
set up cutting parameters on S530 x 1000 lathe with experiment parameters to
produce a smooth surface of parts. If the surface is still rough which is touched by
hand then the cutting parameters must change and if surface is smooth so can
continue to surface roughness measurement.

Figure IV.8 Machining Process

38
The following are the steps in the machining process:

1. Put the raw material (C45 Steel) into the Spindle, then clamp it with chuck.
At this step, a workpiece or raw material is clamped with spindle chuck on
the edge of the material. It is shown in Figure IV.9.

Figure IV.9 Put the raw material (C45 Steel) into the Spindle

2. Run the facing process

At this step, run the facing process of material with Insert Tools Walter
VBMT160408-MM4 WSM20S to create a smooth surface. It is shown in
Figure IV.10.

Figure IV.10 Facing process

39
3. Mark off the center of material in the front face of the tailstock
After the workpiece is clamped, then mark the front face of material with
center drill, it is shown in Figure IV.11.

Figure IV.11 Mark off the center of material in the front face of the tailstock

4. Run the turning process

At this phase, running the turning process with lubricant along the
workpiece. It is shown in Figure IV.12.

Figure IV.12 Turning process

40
5. Cut the material into a specific size
After the turning process is done, cut off the material into 9 parts. But this
cutting process was not cut properly because of the limited distance of the
insert tool to the material. It is shown in Figure IV.13.

Figure IV.13 Cut the material into a specific size

6. Cut objects that have been processed with a hacksaw

At this step, material that has not been cut properly must do the cutting
process again with a hacksaw, so that the material is really cut off. It is
shown in Figure IV.14.

Figure IV.14 Cut objects that have been processed with a hacksaw

41
7. Run the turning process to leave a surface
After the material has been cut, next is run again turning process with a
determined size in order to leave a surface where the surface will be
measured with roughness tester later, the results are in Figure IV.15.

Figure IV.15 C45 Steel part

8. Set up cutting parameters

Before carrying out the turning process, set the cutting parameters according
to the parameters previously selected. The cutting parameter is shown in
Figure IV.16.

Figure IV.16 Cutting Parameters on S530 X 1000 Lathe

42
 9. Run the turning process with different parameters
At this step, a workpiece that has been cut into nine parts must do the turning
process 9 times with cutting parameters which already determined. It is
shown in Figure IV.17.

Figure IV.17 Cutting process and parts that have been cut

IV.1.3 Parameter Experiment Using C45 Steel as an Object on S530 X 1000

Lathe
The process parameters chosen in the turning process are 3 parts, like depth of cut,
spindle speed, and feed rate (𝑛, 𝑓, 𝑎𝑝) that are shown in Table IV.2. The Taguchi
method is used to optimize the parameters which can minimize the surface
roughness. The reason researchers conducting experiments on parameters is to
produce a smooth surface and can still be measured by a Roughness Tester.
Previously, researchers used a spindle speed below 855 with a large feed rate and
depth of cut resulting in a rough surface and could cause damage to the Roughness
Tester.

IV.1.4 Performance Characteristic Identification

After experimenting with parameters, finally found cutting parameters that produce
a relatively smooth surface and can be measured by a Roughness Tester. The
following are the cutting parameters selected in table IV.2.

43
 Table IV.2 Cutting Parameters and Levels

Levels
Cutting Parameters
(Factors) Level 1 Level 2 Level 3

Spindle Speed (𝑛) 855 1350 2000

Feed Rate (𝑓) 0.13 0.17 0.21
Depth of Cut (𝑎𝑝 ) 0.4 0.3 0.5

IV.1.5 Surface Roughness Measurement Result

After selecting the parameters and determining the orthogonal array, then the
surface roughness measurements of 9 parts are carried out using Mitutoyo SJ-410
Series. The result is shown in Table IV.3.

Table IV.3 Surface Roughness Measurement Result

Spindle Feed Depth

Ra
Speed Rate of Cut
855 0.13 0.4 2.474
855 0.17 0.3 2.957
855 0.21 0.5 2.404
1000 0.13 0.4 1.462
1000 0.17 0.3 1.342
1000 0.21 0.5 1.548
1200 0.13 0.4 0.86
1200 0.17 0.3 1.102
1200 0.21 0.5 1.463

44
IV.2 Processing Stage
IV.2.1 Orthogonal Array Determination
It is important to determine the level to evaluate several factors by the number of
tests. In this study, 3 level arrays with 3 parameter variables have been chosen to
get the Taguchi array with 𝐿9 . To determine the Orthogonal Array needed, three
factors with three levels are needed, so an array with 𝐿9 is generated. Table IV.4
shows Taguchi’s Orthogonal Array 𝐿9 .

Table IV.4 Taguchi’s Orthogonal Array L_9 Design

Number of Cutting Depth of Feed

Experiment Speed Cut Rate
1 1 1 1
2 1 2 2
3 1 3 3
4 2 1 3
5 2 2 1
6 2 3 2
7 3 1 2
8 3 2 3
9 3 3 1

After the orthogonal array is generated, the numbers of the three parameters
(spindle speed, feed rate, and depth of cut) are filled into the table according to their
level that shown in Table IV.5.

Table IV.5 Orthogonal Array with Factors

Number of Cutting Depth Feed

Experiment Speed of Cut Rate
1 855 0.13 0.4
2 855 0.17 0.3
3 855 0.21 0.5
4 1000 0.13 0.4
5 1000 0.17 0.3
6 1000 0.21 0.5
7 1200 0.13 0.4
8 1200 0.17 0.3
9 1200 0.21 0.5

45
IV.2.2 Signal to Noise Ratio Calculation
The signal to noise ratio used in this study is smaller is better, because the aim is to
minimize surface roughness. According to Mazumder et al. (2016) the formula of
smaller is better is:

1
S/N ratio (ƞ) = − 10log10 𝑛 ∑𝑛𝑖=1 𝑦𝑖 2

Information:
Where, 𝛾 = 𝑖 𝑡ℎ observed response value
𝑛 = number of observations in a trial,
𝜇 = average of observed response values

In Table IV.6, Signal to Noise Ratio has been calculated with surface roughness
(Ra) as response data on all three factors and levels. can be seen in the table,
generated analysis numbers with 9 pieces of data because the orthogonal array is
𝐿9 . which can be seen in the relationship of parameters with surface roughness.

Table IV.6 Signal to Noise Ratio Calculation

Spindle Depth of
Feed Rate Ra SNRA1
Speed Cut
855 0.13 0.4 2.474 -7.868
855 0.17 0.3 2.957 -9.417
855 0.21 0.5 2.404 -7.6187
1350 0.13 0.3 1.462 -3.2989
1350 0.17 0.5 1.342 -2.5551
1350 0.21 0.4 1.548 -3.7954
2000 0.13 0.5 0.860 1.3100
2000 0.17 0.4 1.102 -0.8436
2000 0.21 0.3 1.463 -3.3049

In the Figure IV.18, it is known that the largest influence of spindle speed parameter
is at 2000 rpm with level 3, then the optimum feed rate is at 0.13 mm/rev with level
1, and at depth of cut, the optimal result is at 0.5 mm with level 3.

46
 Figure IV.18 Main Effects Plot for SN Ratios

In Table IV.7 it is known that the highest results are found in the spindle speed
parameter with delta number 7.351 of rank 1, so it can be said that the spindle speed
parameter has a large effect on minimizing surface roughness. Delta is obtained
from the difference in numbers between the highest and lowest factors.

Table IV.7 Response Table for Signal to Noise Ratios

Spindle Depth of
Level Feed Rate
Speed Cut
1 -8.3012 -3.2856 -5.3403
2 -3.2165 -4.2719 -4.1690
3 -0.9462 -4.9063 -2.9546
Delta 7.3551 1.6207 2.3857
Rank 1 3 2

47
IV.2.3 Normality Test
At this stage, normality tests are performed on the spindle speed, feed rate, and
depth of cut variables. This normality test used the Anderson-Darling Test and the
software used is Minitab 18.

The following is a research hypothesis on the normality test:

H0 : factors normal distribution

H1 : factors not normal distribution
𝛼 : 0.05
Criteria : Reject H0 if the number of P-Value is lower than the
significance level 0.05 and Reject H1 if the number of P-
Value is greater than the significance level 0.05.

IV.2.2.1 Normality Test of Spindle Speed

In Figure IV.19, it is known that the AD value is 0.681 with N is 9, Mean is 1402,
the standard deviation is 497.3, and the P-Value is 0.049 which is lower than 0.05.
It can be concluded that the results reject H0, so the variable not normal distribution.

Figure IV.19 Probability Plot of Spindle Speed

48
IV.2.2.2 Normality Test of Feed Rate
In Table Figure IV.20, it is known that the AD value is 0.657 with N is 9, Mean is
0.17, the standard deviation is 0.03464, and the P-Value is 0.057 which is more than
0.05. It can be concluded that the results accept H0, then the variable can be stated
as normal distribution.

Figure IV.20 Probability Plot of Feed Rate

IV.2.2.3 Normality Test of Depth of Cut

In Table Figure IV.21, it is known that the AD value is 0.657 with N is 9, Mean is
0.4, the standard deviation is 0.08660, and the P-Value is 0.057 which is more than
0.05. It can be concluded that the results accept H0, then the variable can be stated
as normal distribution.

49
 Figure IV.21 Probability Plot of Depth of Cut

IV.2.4 Kruskal-Wallis Test

IV.2.4.1 Kruskal-Wallis Test Spindle Speed
At this stage, Kruskal-Wallis Test are performed on the spindle speed variables.
The Kruskal-Wallis test is performed because the P-Value of the spindle speed
variable is not normally distributed.

Research hypotheses used are:

H0 : All medians are equal
H1 : At least one median is different
𝛼 : 0.05
Criteria : Reject H0 if the number of P-Value is greater than the
significance level 0.05 and Reject H1 if the number of P-
Value is lower than the significance level 0.05.

Table IV.8 shows a statistical description of the Kruskal-Wallis, the N number in

the spindle speed parameter 855, 1350, and 2000 is 3, then the Median at 855 is (-
7.86799), at the speed of 1350 with a median (-3.29895) and at a speed of 2000
with a median (-0.84363). Mean Rank at 855 is 2, then 1350 is 5.7, and 2000 is 7.3.
Z-Value results at a speed of 855 of (-2.32), then at a speed of 1350 of 0.52 and at
a speed of 2000 of 1.81.

50
 Table IV.8 Descriptive Statistics

Spindle
N Median Mean Rank Z-Value
Speed
855 3 -7.86799 2 -2.32
1350 3 -3.29895 5.7 0.52
2000 3 -0.84363 7.3 1.81
Overall 9 5

In Table IV.9, the results obtained from the Kruskal-Wallis test where the degree
of freedom (DF) is 2, followed by an H-Value of 5.96, and a P-Value of 0.051. The
result of the P-Value 0.051 is greater than the significant value 0.05, so it can be
concluded that decline the null hypothesis (H0) which means at least one median is
different or accept alternative hypothesis (H1).

Table IV.9 Kruskal-Wallis Test Result

DF H-Value P-Value
2 5.96 0.051

IV.2.5 Analysis of Variance Calculation

After calculating the signal to noise ratio, the next step is to carry out the Anova
One-Way test.

Research hypotheses used are:

H0 : factors don’t have a significant influence
H1 : factors have a significant influence
𝛼 : 0.05
Criteria : Reject H0 if the number of P-Value is lower than the
significance level 0.05 and Reject H1 if the number of P-
Value is greater than the significance level 0.05.

51
IV.2.4.1 ANOVA One-Way Test Feed Rate
After a one-way ANOVA test ran, data shown in Table IV.10 that degree of
freedom (DF) is 2 with a contribution is 4.06%, sequence sum of squares (Seq SS)
is 4.002, adjusted mean squares (Adj MS) is 2.001, then F-Value is 0.13 and P-
Value is 0.883.

Table IV.10 ANOVA One-way Test SNRA1 Versus Feed Rate

Source DF Seq SS Contribution Adj SS Adj MS F-Value P-Value

Feed
2 4.002 4.06% 4.002 2.001 0.13 0.883
Rate
Error 6 94.447 95.94% 94.447 15.741
Total 8 98.449 100.00%

In Figure IV.22, can be seen graphs showing the relationship between the feed rate
factors 0.3, 0.4, and 0.5 and measured surface roughness. At 0.13 mm/rev, it has a
mean number (-3.29), followed by 0.15 mm/rev has a mean number (-4.27), and at
0.17 mm/rev has a mean number (-4.91).

Interval Plot of SNRA1 vs Feed Rate

95% CI for the Mean
4

-2
SNRA1

-4

-6

-8

-10

-12
0.13 0.17 0.21
Feed Rate
The pooled standard deviation is used to calculate the intervals.

Figure IV.22 Interval Plot of SNRA1 vs Feed Rate

IV.2.4.2 ANOVA One-Way Test Depth of Cut

After a one-way ANOVA test ran, data shown in Table IV.11 that degree of
freedom (DF) is 2 with a contribution is 8.67%, sequence sum of squares (Seq SS)
is 8.538, adjusted mean squares measure (Adj MS) is 4.269, then F-Value is 0.28
and P-Value is 0.762.

52
 Table IV.11 ANOVA One-way Test SNRA1 Versus Depth of Cut

Source DF Seq SS Contribution Adj SS Adj MS F-Value P-Value

Depth of
2 8.538 8.67% 8.538 4.269 0.28 0.762
Cut
Error 6 89.910 91.33% 89.910 14.985
Total 8 98.449 100.00%

In Figure IV.23, can be seen graphs showing the relationship between the three
depth of cut factors 0.4, 0.3, and 0.5 and measured surface roughness. At 0.3 mm
has a mean number (-5.34), followed by 0.4 mm has a mean number (-4.17), and at
0.5 mm has a mean number (-2.95).

Interval Plot of SNRA1 vs Depth of Cut

95% CI for the Mean
4

-2
SNRA1

-4

-6

-8

-10

-12
0.3 0.4 0.5
Depth of Cut
The pooled standard deviation is used to calculate the intervals.

Figure IV.23 Interval Plot of SNRA1 vs Depth of Cut

IV.2.6 Tukey Comparison Test

IV.2.6.1 Tukey Comparison Test Spindle Speed
In Table IV.12, it shows that group A contains 2000 and 1350 spindle speed, and
group B contains 855 spindle speed with a number of observations in each group is
3. The higher results of differences between means are 2000 spindle speed with (-
0.95) and the lowest is 855 spindle speed with (-8.301). The 2000 and 1350 not
significantly different means while 855 spindle speed is significantly different
means.

53
 Table IV.12 Grouping Information Using Tukey Method and 95% Confidence
(Spindle Speed)

Spindle
N Mean Grouping
Speed
2000 3 -0.95 A
1350 3 -3.261 A
855 3 -8.301 B

In Table IV.13, the confidence interval for the difference between the means of
1350 - 855 spindle speed from 1.35 to 8.82 and 2000 – 855 spindle speed from 3.62
to 11.09 which indicates that the difference is statistically significant because it does
not touch zero points. The individual confidence level is 97.80%.

Table IV.13 Tukey Simultaneous Test for Differences Means (Spindle Speed)

Difference Difference SE of T- Adjusted

95% Cl
of Levels of Means Difference Value P-Value
1350 - 855 5.08 1.22 (1.35, 8.82) 4.18 0.014
2000 - 855 7.36 1.22 (3.62, 11.09) 6.04 0.002
2000 - 1350 2.27 1.22 (-1.47, 6.01) 1.86 0.229

In figure IV.24, there are 3 lines, each of which represents a group of factors such
as 1350 - 855, 2000 - 855, and 2000 - 1350. Obtained on lines that not touch zero
pints are lines 1350 - 855 and 2000 - 855 which the difference between means is
significant.

Figure IV.24 Differences of Means for SNRA1 (Spindle Speed)

54
IV.2.6.2 Tukey Comparison Test Feed Rate
Table IV.14 shows that group A contains 0.13, 0.17 and 0.21 feed rate with number
of observations in each group (N) is 3. The higher results of differences between
means is 0.13 feed rate with (-3.29) mean and the lowest is 0.21 feed rate with (-
4.91) mean.

Table IV.14 Grouping Information Using Tukey Method and 95% Confidence
(Feed Rate)

Feed
N Mean Grouping
Rate
0.13 3 -3.29 A
0.17 3 -4.27 A
0.21 3 -4.91 A

In Table IV.15, the confidence interval for the difference between the means of 0.17
- 0.13, 0.21 – 0.13, and 0.21 – 0.17 feed rate indicates that the difference is
statistically not significant because of its touch zero points. The individual
confidence level is 97.80%.

Table IV.15 Tukey Simultaneous Test for Differences Means (Feed Rate)

Difference Difference SE of T- Adjusted

95% Cl
of Levels of Means Difference Value P-Value
0.17 - 0.13 -0.99 3.24 (-10.93, 8.96) -0.3 0.951
0.21 - 0.13 -1.62 3.24 (-11.56, 8.32) -0.5 0.874
0.21 - 0.17 -0.63 3.24 (-10.58, 9.31) -0.2 0.979

55
In figure IV.25, there are 3 lines, each of which represents a group of factors such
as 0.17 - 0.13, 0.21 - 0.13, and 0.21 - 0.17. Obtained on lines that not touch zero
points are none of them because the 3 lines touching the zero point, which the
difference between means is not significant.

Figure IV.25 Differences of Means for SNRA1 (Feed Rate)

IV.2.6.3 Tukey Comparison Test Depth of Cut

Table IV.16 shows that group A contains 0.5, 0.4 and 0.3 feed rate with number of
observations in each group is 3. The higher results of differences between means
are 0.5 feed rate with (-2.95) mean and the lowest is 0.3 feed rate with (-5.34) mean.

Table IV.16 Grouping Information Using Tukey Method and 95% Confidence
(Depth of Cut)

Depth
N Mean Grouping
of Cut
0.5 3 -2.95 A
0.4 3 -4.17 A
0.3 3 -5.34 A

56
In Table IV.17, the confidence interval for the difference between the means of 0.4
– 0.3, 0.5 – 0.3, and 0.5 – 0.4 depth of cut indicates that the difference is statistically
not significant because of its touch zero points. The individual confidence level is
97.80%.

Table IV.17 Tukey Simultaneous Test for Differences Means (Depth of Cut)

Difference Difference SE of T- Adjusted

95% Cl
of Levels of Means Difference Value P-Value
0.4 - 0.3 1.17 3.16 (-8.53, 10.87) 0.37 0.928
0.5 - 0.3 2.39 3.16 (-7.31, 12.09) 0.75 0.742
0.5 - 0.4 1.21 3.16 (-8.49, 10.91) 0.38 0.923

In figure IV.26, there are 3 lines, each of which represents a group of factors such
as 0.4 - 0.3, 0.5 - 0.3, and 0.5 - 0.4. Obtained on lines that not touch zero points are
none of them because the 3 lines touching the zero point, which the difference
between means is not significant.

Figure IV.26 Differences of Means for SNRA1 (Depth of Cut)

57
IV.2.7 Optimum Condition Confirmation
In the final step, confirm the optimal conditions of cutting parameters and their
levels with the result of the surface with minimal roughness. In Table IV.18, it
shown that the optimum conditions are obtained there are spindle speed 2000 rpm
with level 3, feed rate 0.13 mm/rev with level 1, and depth of cut 0.5 mm with level
3.

Table IV.18 Optimum Condition of Cutting Parameters

Parameters
Parameters Spindle Feed Depth
and Levels Speed Rate of Cut
2000 0.13 0.5
Levels 3 1 3

Table IV.19 showed experiment and prediction experiment which the table shows
experiments and predictions of optimal conditions on all three cutting parameters.
In experimental experiments, the turning process is carried out with parameters:
2000 rpm spindle speed, feed rate 0.13 mm/rev, and depth of cut 0.5 mm. The
results obtained in the experiment are the mean with numbers 1.310 and the signal
to noise ratio of 0.860. Then in prediction, the results obtained in the prediction is
the mean with the number 0.806333 and the signal to noise ratio of 1.12288.

Table IV.19 Experiment and Prediction Condition

Experiment and
Prediction Condition
Measurement
Experiment Prediction
S/N Ratio 0.860 1.12288
Mean 1.310 0.806333

58
 CHAPTER V ANALYSIS

V.1 Analysis of DOE Result

In this chapter, we will analyze the experiments that have been carried out on C45
steel parts which will produce an output variant by testing several different
parameters. After turning the process by testing different parameters (spindle speed,
feed rate, and depth of cut) 9 times, then measuring the surface roughness which
results in a variety of numbers. Based on measurements of surface roughness that
have been done before, the largest 𝑅𝑎 number is 2.957 𝜇𝑚 with parameters are:
spindle speed 855 rpm, feed rate 0.13 mm/rev, and depth of cut 0.4 mm. The
lowest 𝑅𝑎 number is 0.860 𝜇𝑚 with its parameters spindle speed 2000 rpm, feed
rate 0.13 mm/rev, and depth of cut 0.5 mm.

V.1.1 Analysis of Signal to Noise Ratio

Based on the results of the response table for signal to noise ratios, the results of the
delta show that the largest delta number in the spindle speed is 7.3551 of rank 1
while the feed rate is 1.6207 and the depth of cut is 2.3857. In the spindle speed
parameter, obtained the smallest number with level 3 with a number of -0.9462.
Then in feed rate parameter, obtained the smallest number with level 1 with a
number of (-3.2856). In the depth of cut parameter, the smallest number level 3 with
a number of (-2.9546) is obtained.

V.1.2 Analysis of Normality Test

Based on the results of normality tests on the three cutting parameters (spindle
speed, feed rate, and depth of cut). The results show that the spindle speed
parameter with a P-Value of 0.049 is smaller than the significance value 0.05 then
the variable is not normally distributed. Then the feed rate and depth of cut
parameters obtained a P-Value of 0.057 which is greater than the significance value
0.05 so it can be said that the feed rate and depth of cut variables are normally
distributed.

59
V.1.3 Analysis of Kruskal-Wallis Test
Based on the Kruskal-Wallis test that has been done on the spindle speed parameter,
shows that the P-Value generated is 0.051 which is greater than the significance
value 0.05 so that it can be concluded that the variable is different from the others
in other words reject null hypothesis (H0) or accept alternative Hypothesis (H1).

V.1.4 Analysis of ANOVA

The results of the ANOVA One-Way test show that cutting parameters P-value in
feed rate is 0.883 and the depth of cut is 0.762. The contribution to minimizing
surface roughness by depth of cut 8.67% and feed rate 4.06%. Seen from the P-
value on feed rate parameter has a P-Value of 0.883 which is greater than the
significance value so that it can be said that rejecting the alternative hypothesis (H1)
or the factor can be said to have no significant effect. In depth of cut parameter has
a P-Value of 0.762 which is greater than the significance value so that it can be said
that rejecting the alternative hypothesis (H1) or the factor can be said to have no
significant effect.

V.1.5 Analysis of Tukey Pairwise Comparison

Based on the results of the Tukey Pairwise Comparison test, the spindle speed
variable results in that the variables 2000 and 1350 rpm do not have a significant
difference in means of surface roughness. Whereas the 855 rpm variable when
compared with 2000 and 1350 rpm has a significant difference in means because of
the lines in the groups 1350 - 855 and 2000 - 855 touch the zero point.

V.1.6 Analysis of Optimal Condition

From the One-Way Anova test that has been done, the results obtained in the table
above which generated contributions to each parameter. Table V.1 showed the
results obtained from the ANOVA test showed optimal cutting parameters, such as
spindle speed with 2000 rpm of level 3, feed rate with 0.13 mm / rev of level 1, and
depth of cut with 0.5 mm of level 3. The results obtained in the experiment are the
mean with numbers 1.310 and the signal to noise ratio of 0.860. Then in prediction,

60
the results obtained in the prediction is the mean with the number 0.806333 and the
signal to noise ratio of 1.12288.

Table V.1 Optimal Condition

Optimal Condition
Cutting
Level Description Experiment Prediction
Parameters S/N
Spindle Ratio
3 2000 0.860 1.12288
Speed
Feed Rate 1 0.13
Depth of Mean 1.310 0.806333
3 0.5
Cut

61
 CHAPTER VI CONCLUSION

VI.1 Conclusion
Research to optimize the cutting parameters has the aim to minimize surface
roughness on C45 steel objects in the turning process on S530 x 1000 lathe. The
initial step of the research is to cut long cylindric raw material into 9 parts first, then
the turning process is carried out on each part with the predetermined cutting
parameters. After that, the process of measuring surface roughness is carried out on
the nine parts with the aim of getting the average surface roughness. Average
roughness measurement results have been obtained. After the cutting parameters
have been determined, then the normality test is then performed to determine
whether the variables are normally distributed or not. Note that the spindle speed
parameter is not normally distributed then the parameter is continued to the
Kruskal-Wallis test (nonparametric). Then using the Taguchi method, the study
uses a signal to noise ratio and One-Way Anova Test. After using signal to noise
ratio and the One-Way Anova test, the results are obtained which shows the optimal
conditions of the cutting parameter and its level. The next step is to carry out the
Tukey test to find out if there are significant differences between the variables. It is
known that the spindle speed parameter, the speed of 2000 rpm and 1350 rpm does
not have a significant difference, but between 2000-1350 rpm and 850 rpm have a
significant difference. The feed rate and depth of cut parameters at levels 1-3 do not
have significant differences. Selected cutting parameters for optimal conditions are
spindle speed of 2000 rpm of level 3, the feed rate of 0.13 mm/rev of level 1, and
depth of cut of 0.5 mm of level 3. The results showed optimal conditions with a
signal to noise ratio of experiment is 0.860 and prediction is 1.12288 with the
selected optimal condition parameters, the object becomes smooth with less surface
roughness.

62
VI.2 Suggestion
The suggestion for this research is to first determine the object to be examined, then
design the size of the part you want to measure first to make it easier to measure.
After that, determine the cutting parameters and measure the factors and levels.
Next, take measurements to find the optimal conditions to minimize surface
roughness.

63
 REFERENCES

Acar, Elif F., and Sun, Lei. 2012. A Generalized Kruskal–Wallis Test Incorporating
Group Uncertainty with Application to Genetic Association Studies.
arXiv:1205.0534.
Anonym. 2019. Methods and Formulas for The Analysis of Variance in One-Way
Anova. https://support.minitab.com/en-us/minitab/19/help-and-how-
to/modeling-statistics/anova/how-to/one-way-anova/methods-and-
formulas/analysis-of-variance/. Accessed on 15 December 2019.

Asfar, Mukhammad, et al. 2018. Pengendalian Kualitas Produk Bata Ringan AAC
dengan Metode Taguchi di PT AFU 28. Vol.8 No.2 Edisi Nopember.

Balajia, M., et al. 2016. Optimization of Cutting Parameters in Drilling of AISI 304
Stainless Steel Using Taguchi and ANOVA. Procedia Technology 25:
1106 – 1113.

Benhabib, Beno. 2003. Manufacturing Design, Production, Automation, and

Integration. Marcel Dekker, Inc. New York.

Black, J T., Kohser Ronald A. 2008. DeGarmo's Materials and Processes in

Manufacturing Eleventh Edition. John Wiley & Sons, Inc. United
States.

Burakowski, Tadeusz., et al. 1999. Surface Engineering of Metals Principles,

Equipment, Technologies. CRC Press LLC. Boca Raton.

Camposeco-Negrete, Carmita, et al. 2015. Optimization of cutting parameters to

minimize energy consumption during turning of AISI 1018 steel at
constant material removal rate using robust design. Int J Adv Manuf
Technol. 83:1341–1347.

Dubitzky, Werner., et al. 2013. Encyclopedia of Systems Biology. Springer. New

York.

64
Ghasemi, Asghar and Zahediasl, Saleh. 2012. Normality Tests for Statistical
Analysis: A Guide for Non-Statisticians. Int J Endocrinol Metab.
2012;10(2):486-489.

Groover, Mikell P. 2010. Fundamentals of Modern Manufacturing 4th Edition.

John Wiley & Sons, Inc. United States.

Ibrahim, Gusri Akhyar, et al. 2017. Analysis of Surface Roughness Value When
Drilling Magnesiumaz31 Using Taguchi Method. INSIST Vol. 2 No. 2
(71–74).

Iriawan, N., Astuti S.P. 2006. Mengolah Data Statistik dengan Mudah
Menggunakan Minitab 14. CV ANDI. Jakarta.

Kaur, Amandeep and Kumar, Robin. 2015. Comparative Analysis of Parametric

and Non-Parametric Tests. Journal of Computer and Mathematical
Sciences, Vol.6(6),336-342.

Khandey, Umesh. 2009. Optimization of Surface Roughness, Material Removal

Rate and cutting Tool Flank Wear in Turning Using Extended Taguchi
Approach. National Institute of Technology Rourkela 769008, India.

Mason, Robert L., et al. 2003. Statistical Design and Analysis of Experiments with
Applications to Engineering and Science Second Edition. John Wiley
& Sons, Inc. New Jersey.

Mazumder, Prasun., et al. 2016. Optimization of Process Parameters in Drilling of

Bamboo Fiber Reinforced Polymeric Composites. ISBN: 978-81-
931039-0-6.

Mircioiu, Constantin and Atkinson, Jeffrey. 2017. A Comparison of Parametric and

Non-Parametric Methods Applied to a Likert Scale. Pharmacy 2017, 5,
26.

Montgomery, Douglas C. 1985. Introduction to Statistical Quality Control, Seventh

Edition. Wiley. New Jersey.

65
Montgomery, Douglas C, and George C. Runger. 2003. Applied Statistics and
Probability for Engineers Third Edition. John Wiley & Sons, Inc. New
Jersey.

Murray, James M. 2017. One-Way Analysis of Variance (ANOVA). University of

Wisconsin - La Crosse.

Olleveant, N. A., 1999. Tukey multiple comparison test. Journal of Clinical

Nursing, 8, pp.299-304.
Osman, M.H., et al. 2017. Effect of Cutting Parameters on Surface Roughness in
Dry Drilling of AISI D2 Tool Steel by Using Taguchi Method. eISSN:
2289-8107.

Ostertagová, Eva., et al. 2014. Methodology and Application of the Kruskal-Wallis

Test. Applied Mechanics and Materials Vol. 611 (2014) pp 115-120.

P.G. Benardos, G.-C. Vosniakos. 2003. Predicting Surface Roughness in

Machining: A Review. International Journal of Machine Tools &
Manufacture 43, 833–844.

Phillip J., Ross. 1996. Taguchi techniques for quality engineering. McGraw-Hill.
New York.

Ribeiro, João, et al. 2017. Optimization of Cutting Parameters to Minimize the

Surface Roughness in the End Milling Process Using the Taguchi
Method. Periodica Polytechnica Mechanical Engineering pp: 30-35.

Roy, Ranjit K. 2001. Design of Experiments Using Taguchi Approach. Wiley. New
York.

Rutherford, Andrew. 2011. ANOVA and ANCOVA A GLM Approach Second

Edition. John Wiley & Sons, Inc. New Jersey.

S. Basavarajappa, et al. 2006. Application of Taguchi techniques to study dry sliding

wear behavior of metal matrix composites. Materials and Design 28:
1393–1398.

66
Saravanakumar, A., et al. 2016. Optimization of Machining Parameters using
Taguchi Method for Surface Roughness. J. Mater. Environ. Sci. 7 (5),
1556-1561.

Singh, Gurmeet., et al. 2016. Optimization of process parameters for drilled hole
quality characteristics during cortical bone drilling using Taguchi
method. Journal of the mechanical behavior of biomedical materials 62:
355–365.
Singh, Rajender. 2006. Introduction to Basic Manufacturing Processes and
Workshop Technology. New Age International (P) Ltd., Publishers.
New Delhi.

Sutoni, Akhmad., et al. 2018. Penerapan Metode Taguchi Dalam Interaksi

Komposisi Batako Pres Terhadap Kualitas Daya Serap Dan Daya
Tekan Batako. Jurnal Telematika edisi Industrial Engineering Seminar
and Call for Paper (IESC) 2018 p-ISSN: 1858-2516 e-ISSN: 2579-
3772.

Taguchi, Genichi. Chowdhury, Subir. Wu, Tuin. 2005. Taguchi's Quality

Engineering Handbook. Wiley. New Jersey.

Trehan, R., et al. 2013. Optimization of mechanical properties of polyester hybrid

composite laminate using Taguchi methodology – Part 1. Journal of
Materials: Design and Applications.

Walsh, Ronald A and Cormier, Denis R. 2006. Mcgraw-Hill Machining and

Metalworking Handbook. The McGraw-Hill Companies, Inc. United
States of America.

Z. Zhang, Julie., et al. 2007. Surface roughness optimization in an end-milling

operation using the Taguchi design method. Journal of Materials
Processing Technology 184: 233–239.

Zerti, Oussama., et al. 2017. Taguchi Design of Experiments for Optimization and
Modeling of Surface Roughness When Dry Turning X210Cr12 Steel.
Applied Mechanics, Behavior of Materials, and Engineering Systems.

67
 APPENDIX A
Surface Roughness Test Results

68
Experiment 1 Experiment 2

Experiment 3 Experiment 4

69
Experiment 5 Experiment 6

Experiment 7 Experiment 8

Experiment 9

70
APPENDIX B
Signal to Noise Ratio Results

71
72
APPENDIX C
Normality Test Results

73
74
 APPENDIX D
Analysis of Variance (ANOVA) Test Results

75
76
 Interval Plot of SNRA1 vs Spindle Speed
95% CI for the Mean

-2

-4
SNRA1

-6

-8

-10

-12
855 1350 2000
Spindle Speed
The pooled standard deviation is used to calculate the intervals.

Interval Plot of SNRA1 vs Feed Rate

95% CI for the Mean
4

-2
SNRA1

-4

-6

-8

-10

-12
0.13 0.17 0.21
Feed Rate
The pooled standard deviation is used to calculate the intervals.

Interval Plot of SNRA1 vs Depth of Cut

95% CI for the Mean
4

-2
SNRA1

-4

-6

-8

-10

-12
0.3 0.4 0.5
Depth of Cut
The pooled standard deviation is used to calculate the intervals.

77
 APPENDIX E
Tukey Pairwise Comparisons Results

78
Spindle Speed

79
Feed Rate

80
Depth of Cut

81
APPENDIX F
Kruskal-Wallis Test Results

82
83
APPENDIX G
Insert Tools Specimen Results

84
Experiment 1 Experiment 2

Experiment 3 Experiment 4

Experiment 5 Experiment 6

85
Experiment 7 Experiment 8

Experiment 9

86