Investigation of Variables Affecting Kerf Width Surface Roughness and Material Removal Rate in Wire Electrical Discharge Machining ____________________________________________________ Ph.D. Dissertation (Session 2006) Submitted By Mr. Aqueel Shah 2006-Ph.D-MNF-05 Supervised By Prof. Dr. Nadeem Ahmad Mufti Department of Industrial and Manufacturing Engineering University of Engineering and Technology Lahore-Pakistan 2012
141
Embed
Investigation of Variables Affecting Kerf Width Surface ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Investigation of Variables Affecting Kerf Width Surface Roughness and Material Removal Rate in Wire Electrical
Department of Industrial and Manufacturing Engineering University of Engineering and Technology
Lahore-Pakistan 2012
Department of Industrial and Manufacturing Engineering
II
Dedication
To my deceased Mother Mrs. Jannat Shah
To my family
Department of Industrial and Manufacturing Engineering
III
Investigation of Variables Affecting Kerf Width Surface Roughness and Material Removal Rate in
Wire Electrical Discharge Machining
Mr. Aqueel Shah 2006-Ph.D-MNF-05
Supervisor
Prof. Dr. Nadeem Ahmad Mufti
A dissertation submitted for the degree of Doctor of Philosophy
in Manufacturing Engineering
Internal Examiner External Examiner
Dr. Nadeem Ahmad Mufti Dr. Syed Amir Iqbal Department of Industrial & Department of Industrial & Manufacturing Engg., Manufacturing Engg., NED University of Engineering & Technology, University of Engineering & Lahore Technology, Karachi
Chairman Dean Department of Industrial & Manufacturing Faculty of Mechanical Engg, University of Engineering &Technology, Engineering, University of Lahore Engg. & Technology, Lahore
Department of Industrial and Manufacturing Engineering
IV
Names of Ph.D. Thesis External Examiners
From within the Country
Dr. Syed Amir Iqbal, Professor Chairman, Department of Industrial and Manufacturing Engineering, NED University of Engineering and Technology, Karachi, Pakistan.
From Abroad 1 Dr. Xun Chen, Professor,
Professor of Manufacturing General Engineering Research Institute Liverpool John Moores University Liverpool L3 3AF, UK
2 Dr. Asif Iqbal, Assistant Professor,
ME122, Department of Mechanical Engineering, Eastern Mediterranean University, Gazimagusa, Turkish Republic of North Cyprus, Via Mersin 10, TURKEY
3 Dr. Ye Li, Assistant Professor
Office: Morgan Hall Room 109B, Department: Industrial & Manufacturing Engineering & Technology, Bradley University, 1501 West Bradley Ave., Peoria, IL 61625, USA.
Department of Industrial and Manufacturing Engineering
V
Declaration
None of the material contained in this thesis has been submitted in support of an
application for another degree or qualification of this or any other university or the
institution of learning.
Department of Industrial and Manufacturing Engineering
VI
Acknowledgements The author is indebted to The Department of Industrial and Manufacturing
Engineering University of Engineering (UET), Lahore, Pakistan, The Higher
Education Commission (HEC) of Pakistan, and Pakistan Navy for having made this
research possible.
Aqueel Shah
Department of Industrial and Manufacturing Engineering
VII
Abstract
Wire-Electrical Discharge Machining (WEDM) is one of the non-conventional
machining processes for machining hard to machine electrically conductive materials.
It has been increasingly used in industry owing to its distinct advantages over the
other cutting technologies. The process can only be employed effectively when all its
properties and complexities are well understood. In addition many aspects of this
technology require to be fully explored in order to increase its capabilities and cutting
performance. This thesis contains an extensive literature review and an experimental
work on the investigations of various variables in Wire-EDM. It is a fact that the
substantial amount of work has been carried out on Wire-EDM, but a very little
research has been reported on the influence of the variables such as the work piece
thickness and hardness on various machining responses such as surface roughness,
kerf width and material removal rate. Accordingly a detailed experimental
investigation is presented in this thesis to study the various cutting performance
measures in Wire-EDM over a wide range of variables or process parameters
including workpiece thickness and hardness. The influence of all these variables/
control factors/ process parameters on the major cutting performance measures in
Wire-EDM have been comprehensively discussed and analyzed under two sets of
experiments.
In the first set of experiments, the influence of eight variables including thickness has
been studied on the machining responses such as kerf width, surface roughness, and
Department of Industrial and Manufacturing Engineering
VIII
material removal rate. The workpiece material used was Tungsten Carbide. Eight
variables including thickness have been taken with three levels each to determine
their influence on the machining responses. In this the Taguchi Orthogonal Array has
been used to reduce the number of runs for meaningful results. Tungsten Carbide
workpieces were machined and the requisite response variables were measured.
Likewise, in the second set of experiments the same material was taken and hardened
to obtain two levels of hardness. The workpiece hardness was taken instead of
thickness with four other variables having two levels each. This was done to validate
the results of first experiment and also to see the influence of hardness. In both the
experiments, ANOVA was carried out after obtaining the responses to determine the
significant factors for each response. The result was consistent with the available
literature however new facts were discovered in the case of workpiece thickness and
hardness. Workpiece thickness appeared to be significant in case of surface roughness
only and hardness was found significant in all the three cases. Finally the
optimization of the machining responses was carried out using S/N ratio as specified
by Taguchi method for the purpose of research papers publications.
Department of Industrial and Manufacturing Engineering
IX
List of Publications
o Aqueel Shah, Nadeem A. Mufti, Dinesh Rakwal, and Eberhard Bamberg,” Material Removal Rate, Kerf, and Surface Roughness of Tungsten Carbide Machined with Wire Electrical Discharge Machining”, Journal of Materials Engineering and Performance, 2011, Volume 20, Number 1, Pages 71-76.
o Aqueel Shah, Nadeem A. Mufti, “Influence of Machine Control Variables and
Material Hardness on machining responses in Wire Electrical Discharge Machining”. The paper is under review in International Journal of Advance Manufacturing Technology for publication.
o Aqueel Shah, Nadeem A. Mufti, “Multiple Response Optimization of Process
Parameters in Wire Electrical Discharge Machining of Tungsten Carbide Using Various Optimization Techniques”. The paper is being reviewed for publication.
Department of Industrial and Manufacturing Engineering
X
Table of contents
Dedication ........................................................................................................................... II
Declaration .......................................................................................................................... V
Acknowledgement ............................................................................................................ VI
Abstract ............................................................................................................................ VII
List of Publications ........................................................................................................... IX
Contents .............................................................................................................................. X
List of Figures ............................................................................................................... XIV
List of Tables ................................................................................................................... XV
CHAPTER 1
INTRODUCTION TO WIRE-EDM ................................................................................... 1
1.1 COMPARISON WITH OTHER NON CONVENTIONAL MANUFACTURING PROCESSES ................................................................................ 1
3.1.2.3 ANALYSIS OF THE EXPERIMENTAL DATA .................................. 36
3.2 ADVANTAGES AND DISADVANTAGES OF TAGUCHI METHOD .......... 38
3.3 RECENT RESEARCH WORK INVOLVING TAGUCHI METHOD .............. 40
CHAPTER 4
EXPERIMENTATION, STATISTICAL ANALYSES AND OPTIMIZATION FOR WORKPIECE THICKNESS ............................................................................................. 43
4.5.3 MATERIAL REMOVAL RATE (MRR) .................................................... 77
4.6 RELATIONS AND MATHEMATICAL MODELS FOR APPROXIMATION ....................................................................................................... 78
machining (PAM), Water jet machining (WJM) and Wire electric discharge machining (W-
EDM) have been considered. For the selection of appropriate process, the techniques such as
QFD-based expert system, analytic network process and data envelopment analysis (DEA) were
used. Some other researchers [7-10] have used hybrid wire EDM process to enhance the
capabilities of the process for a specific application such as (CWEDT) cylindrical wire-electrical
discharge turning.
A specific manufacturing process that proves its suitability under various given conditions may
not necessarily be equally good under other conditions [11]. Therefore, extreme care must be
taken while selecting a process for a given manufacturing task. The analysis can be made from
the following technical point of views:
a. The physical parameters involved
b. Capability in machining various different shapes
c. Applicability of various processes to different types of materials
d. The operational characteristics of the manufacturing
e. Economics involved.
1.1.1 PHYSICAL PARAMETERS
The physical parameters of the non-conventional machining processes directly affect the rate of
the material removal and energy consumed in different processes as given in table 1.1
3
The relation between the rate of Metal removal and power consumed by various non-
conventional machining processes is shown by the figure 1.1
Figure 1.1: Relationship between rate of metal removal and the power consumption [11]
It can be seen that some of the processes like AJM, EBM and ECM are above the mean power
consumption line and consume greater amount of power than the other processes that are below
the mean power consumption line. Hence, the running cost involved in those processes that are
Table 1.1:- Comparison of the Physical Parameters [11] Parameters EDM EBM LBM PAM USM AJM ECM CHM Potential (V)
45 150000 4500 100 220 220 10 -
Current (Amp)
50 (Pulsed
DC)
0.001 (Pulsed
DC)
2.0 500 (DC)
12 (AC)
1.0 10000 (DC)
-
Power (W) 2700 150 - 50000 2400 220 100000 - Gap (m.m.) 0.025 100 150 7.5 0.25 0.75 0.2 - Medium Liquid di-
electric Vacuum Air Argon or
Hydrogen Abrasive In water
Abrasive In gas
Electrolyte
Liquid Chemical
4
lying above the mean is high on the other hand for the processes below that line is comparatively
lesser.
1.1.2 SHAPING CAPABILITY
The shaping capability of the different processes can be judged on the basis of different
machining operations that they can perform. This includes drilling, micro-drilling, cavity
sinking, shallow and deep pocketing, contouring and through cutting etc. Comparison is shown
in table 1.2. Laser beam machining is the only process that has enough capability to micro drill
and for the drilling of shapes having slenderness ratio less than 20, USM, ECM and EDM
processes will be more suitable. EDM and ECM processes have a better capability of pocketing
operation. ECM is suitable process for contouring operations, but no other process has the
capability of contouring operation except for EDM.
Table 1.2:- Shape application of Non-Conventional Processes [11]
Process
Holes Trough cavities
Surfacing Trough cutting
Precision small Standard Precision Standard
Double Contour
Surface of revolution Dia
<0.025 Dia >0.025
Length <20 mm
Length >20 mm
Shallow Deep
EDM - - Better OK Better Better OK - Bad - USM - - Better Bad Better Better Bad - Bad - AJM - - OK Bad Bad OK - - Better - CHM OK OK - - Bad OK - - Better - LBM Better Better OK Bad Bad Bad - - Better OK PAM - - OK - Bad Bad - Bad Better Better
1.1.3 APPLICABILITY TO VARIOUS MATERIALS
Materials applications for various machining processes have been shown at the tables 1.3 and
1.4. The table 1.3 is concerned with metals and alloys while table 1.4 is showing non-metals. It
5
can be noticed that both the ECM and EDM are unsuitable in machining of the electrically non-
conducting materials. In these types of cases the desired results can be obtained by mechanical
methods. USM and AJM are the most suitable processes for the machining of hard material.
Table 1.3:- Metals and Alloys [11] PROCESS Titanium Super alloy Steel Refractory Aluminum EDM Better Better Better Better OK AJM OK Better OK Better OK ECM OK Better Better OK OK CHM OK OK Better OK Better USM OK Bad OK Better Bad EBM OK OK OK OK OK LBM OK OK OK OK OK PAM OK Better Better OK Better
Table 1.4:- Cutting Performance for Non-Metals [11] PROCESS CERAMICS PLASTIC GLASS EDM - - - AJM Better OK Better ECM - - - CHM Bad Bad OK USM Better OK Better EBM Better OK OK LBM Better OK OK PAM - Bad -
1.1.4 THE MACHINING CHARACTERISTICS
Machining characteristics can be analyzed by considering the surface finish, the rate of metal
removal, depth of surface damage, tolerance, and power requirement for machining. The process
capabilities have been compared in table 1.5. The metal removal rate by EDM
6
Table 1.5:- Comparison of Machining Characteristics among Various Processes [11]
is far too low as compared to ECM and PAM. When ECM and PAM are compared to
conventional machining, they are quarter and 1.25 times respectively. Power requirement for
EDM is in the middle zone but it is very high for ECM and PAM in comparison to the rest of the
non-traditional processes of machining. EDM and ECM have very low tool wear rate but ECM
has drawbacks of the contaminating the electrolyte and the corroding the machine parts. The
surface finish for EDM is at par with other non-conventional processes.
1.1.5 THE ECONOMICS OF THE PROCESS
The economics of the various processes have been analyzed and given in table 1.6. They were
analyzed on the basis of capital cost, consumed power cost, tooling cost, the tool wear and the
rate of metal removal. Capital cost of the ECM is too high comparison to the other non-
traditional processes of machining and the capital costs for AJM and PAM are less in
comparison. The tooling cost for EDM is higher than most of the other non-traditional processes
of machining however it consumes less power. The metal removal efficiency for LBM and EBM
is very high than for the other processes and it is high for EDM. It can be concluded
7
that the suitability of application of any process depends on all these factors and they must be
considered before choosing a non-conventional processes.
1.2 WIRE EDM APPLICATIONS
In this section the feasibility of the process in the machining of the various materials is discussed
that are used mainly in tooling applications.
1.2.1 MODERN TOOLING APPLICATIONS
Wire-EDM has obtained wide popularity in the machining of some materials that are utilized in
the modern tooling applications. Various researchers [12,13] have attempted to investigate the
performance of the process by wafering of the silicon and by machining the compacting dies. For
the dressing of a rotating metal bond diamond wheel that is used for ceramics precision form
grinding, the viability of using cylindrical Wire-EDM has been also studied [14]. The analysis
has proven the capability of the process in generating intricate and precise profiles. These
Table 1.6:- The Economics of the process [11]
PROCESS Capital
Cost MRR
Efficiency Tool Wear
Power Consumption Tooling Cost
EDM Medium High High Less High USM Less High Medium Less Less AJM Too Less High Less Less Less ECM Too High Less Too less Too High Medium CHM Medium Medium Too less High Less
EBM High Too High Too less Less Less
LBM Less Too High Too less Too less Less
PAM Too Less Too Less Too less Too less Less
MCG Less Too Less Less Less Less
8
profiles have very small radii at the corners but comparatively the wear rate is high on the
diamond wheel after the first grinding pass. Such a high wear rate of the wheel at initial pass is
attributed to over protruding grains that are not bonded very rigidly to wheel when observed on
completion of the process [15]. A comparative study was carried out on the laser-cutting process
[16] in machining of soft MnZn ferrite magnetic and permanent NdFeB materials. They are used
in the miniature systems. It was ultimately found that the Wire-EDM process yielded better
surface finish quality and dimensional accuracy but the cutting rate was very low. Another study
was carried out to investigate in depth the micro Wire-EDM machining performance in
machining a component of high aspect ratio using variety of various materials that included
nitronic austentic stainless, stainless steel, titanium and beryllium copper [17].
1.2.2 ADVANCED CERAMIC MATERIALS
For the machining operation of the advanced ceramics, WIRE-EDM has emerged as one of the
most feasible substitutes. Sanchez et al. [18] carried out research on the machining of advanced
ceramics that were usually machined by lapping/diamond grinding. Feasibility to machine the
silicon infiltrated silicon carbide and boron carbide with the use of EDM and WIRE-EDM was
also studied. In another study, Cheng et al. [19] explored the feasibility of cutting ZrB2 based
materials with EDM and WIRE-EDM. The response of conductive carbide content, for example
titanium carbide and niobium carbide, upon the surface roughness and cutting rate of zirconia
ceramics after WIRE EDM was examined by Matsuo and Oshima [20]. Lok and Lee [21]
successfully carried out WIRE EDM of Sialon 501 and aluminum oxide titanium carbide. In this,
they found that the material removal rate was very low in comparison to the machining of the
metals like SKD-11 alloy steel and the surface roughness generally remained on the lower side
9
as compared to the EDM process. Dauw et al. [22] also reported that the material removal rate
and surface roughness do not only depend upon the machine variables but they are also
dependant on material. The method dealing with the WIRE-EDM technological limitations has
recently been explored that requires the materials electrical resistivity with specific threshold
values that are about 100 X/cm [23] and 300 X/cm [24]. Engineering ceramics have different
grades. Konig et al. [23] has classified ceramics as non conductor, natural conductor and the
conductor that is obtained by doping nonconductors with conducting materials. In this, Mohri et
al. [25] has been able to bring a new perspective in traditional EDM by using single assisting
electrode to facilitate the spark produced in ceramics which have a high electrical resistivity
value. Both the machining processes of EDM and WIRE EDM have been successfully tested.
Conductive particles from the assisting electrodes were diffused on to the surface of the
particular ceramics in which feeding of the electrode through the insulating material is assisted.
This procedure was also applied on the other various kinds of insulating ceramic materials with
the inclusion of oxide ceramics like Al2O3 and ZrO2. They fall in the category of limited
electrical conductive properties [26].
1.2.3 MODERN COMPOSITE MATERIALS
Among the various material removal processes for machining of the modern composite materials
the WIRE-EDM is considered an effective and very economical source. Many comparative
studies [27,28] have already been carried out between laser cutting and WIRE-EDM especially
in the metal matrix composites processing, carbon fibre and reinforced liquid crystal polymer
composites. These studies revealed that the process provides better quality as far as cutting edge
is concerned and there is a better control over process parameters resulting in lesser damages on
10
the surface of the workpiece. However, it showed lower material removal rate in the case of
every composite material which was tested. Gadalla and Tsai [29] compared the conventional
diamond sawing with WIRE-EDM. They found out that the process produced more surface
roughness as compared to the diamond sawing and the material removal rate was also higher.
Yan et al. [30] studied the different machining processes that are used to machine metal matrix
composites. Their experiment consisted of machining Al2O3/6061Al composite. They used
rotary EDM which was done with the help of electrode which had a shape of a disk. Few more
studies [31,32] have been carried out on WIRE EDM of particulate reinforced composites Al2O3
to investigate the affects of process variables on its performance. As a result of these studies it
was established that the process parameters have a very small affect on the surface roughness but
show a considerable affect on the cutting rate.
1.3 THESIS OVERVIEW
This thesis has been divided into various chapters that will describe the entire working starting
from introduction till the recommendation part. Chapter 1 provides the introduction to the
process. It includes the introduction to the non-conventional machining processes especially
Wire Electrical Discharge Machining. In this chapter, various non conventional machining
processes have been compared to the process on the basis of their inherent advantages. Chapter 2
comprises of the literature survey that has lead to the identification of the problems where a very
little research work has been carried out and on the basis of which the design of experiment has
taken place. The DOE for methodology has been explained in chapter 3. In this thesis basically
two sets of experiments were carried out. One of them was for the workpiece material thickness
and the other for its hardness. The aim was to identify the variables or process parameters that
11
affected the kerf width, surface roughness and the material removal rate in both the sets of
experiments. Besides thickness and hardness machine-specific process parameters were also
investigated. The experimental work for both the workpiece thickness and hardness has been
presented in chapters 4 and 5 respectively. The final chapter contains conclusion and
recommendations.
12
CHAPTER2LITERATUREREVIEW
This chapter provides a review on the various research areas related to the process. The main
section of the chapter focuses on the major research efforts that include the process optimization
along with the process monitoring and control.
2.1 INTRODUCTION
Wire-electrical discharge machining usually known as WIRE EDM, is a widely used non-
conventional process of material removal. With the help of this process manufacturing of parts
with intricate profiles and shapes can be carried out [29]. It is also a form of the EDM process
which uses an electrode for initiation of the spark. However, the process uses a wire electrode
that usually made of tungsten, brass or copper. The wire diameters range between 50 to 300
microns. This enhances the capability of the process in obtaining small corner radii. To avoid the
production of inaccurate parts this wire is kept under tension with the help of a mechanical
device. With the wire travelling ahead through the workpiece material without making any
physical contact, the material ahead of wire is eroded slowly. As there is no contact between the
wire and the material, the chances of production of mechanical stresses are minimized.
Moreover, the process can machine high strength materials. The process has the capability of
machining high temperature resistive materials also. It also eliminates the change in geometry
that is observed while machining steels after heat treatment. The initial introduction of the
process to the manufacturing industries took place around late 1960s. The process evolved as a
result of research being conducted for developing a technique to replace electrode used in
conventional EDM that requires machining before use. Dule-bohn[30], carried out a study in
13
which they were successful to control the shape of part automatically. In this study a system of
optical line follower was used. At this point of time, its popularity rapidly increased, as the
industry had well understood the process and its capabilities [31]. In the late 70s, computer
numerical control (CNC) system was introduced into the process. This was a major breakthrough
in the further development of this machining process. Ultimately, the process capabilities of
wire-EDM were explored extensively for through hole machining that was required because the
wire has to be passed through the workpiece before the hole can be machined. Most common
applications are the manufacturing of fixtures and gauges, the extrusion tools, prototypes, dies,
aircraft components, grinding wheel form tools and medical parts.
2.2 WIRE-EDM
In this section the information regarding the process variations when combined with other
material removal techniques and the basic principles of the process are provided. The material
removal process is same as that in conventional EDM technique where the erosion affect is
evolved with the help of sparks or continuous electrical discharges. In the process, a series of
discrete discharges occur between the wire and the workpiece material and the material erosion
takes place from the workpiece. The wire and the workpiece are segregated by a continuous flow
of dielectric fluid, which is in the machining area [32]. However, now a day, the process is
usually carried out where work material is fully submerged. It is a tank which is filled with
dielectric fluid to support machining. This submerged type of process has few advantages. It
results in temperature stabilization and an efficient flushing when the workpiece thickness is not
uniform. Using electrical energy a channel of plasma is generated by the process between
cathode and anode [33], and then it is transformed it into thermal energy [34] at the temperature
14
range 8000 to 12,000º C [35] or according to some other researchers it is reported as high as
20,000º C [26]. A noticeable amount of heat is generated and melting of material is initialized.
This melting starts on each pole’s surface. The plasma channel breaks down when pulsating
power direct current supply that occurs between 30,000 and 20,000 Hz is stopped [37], the
plasma channel will break down. Therefore a quick reduction in the temperature is caused and
the dielectric hits the plasma channel. This dielectric flushes away the microscopic debris from
the pole surfaces. Although the procedure of material removal in EDM and the process is same
but their functional characteristics are different. The wire is fed through the workpiece
continuously with the help of a microprocessor that provides the ability to machine parts with
exceptionally high accuracy and complex shapes. The achievement of tapper varying degrees
ranging from 30º for 400 mm thick and 15º for 100 mm thick components has become possible.
The gap is maintained constantly between workpiece and the wire with the help of
microprocessor. The gap normally can be controlled between the ranges of 0.025 to 0.05 mm
[31]. Pre-shaped electrode in the process is no more required as in the case of conventional EDM
for the processes of cutting and finishing. For obtaining the required dimensional accuracy and
surface finish the wire is required to make few passes along the profile of the workpiece.
Kunieda and Furudate [38] conducted a study on dry WIRE-EDM in a gaseous atmosphere and
the dielectric was not used. The purpose of the study was to enhance the efficiency of the
finishing process. In case where the workpiece material was D2 tool steel, the typical cutting rate
of the process is 300 mm2/min for cutting thickness of 50 mm and 750 mm2/min for cutting
Aluminum which is 150 mm [39], and the produced surface finish quality was between 0.04 to
0.25µ Ra. Moreover, instead of the hydrocarbon oils the process uses de-ionized water as di-
electric fluid. This de-ionized water is kept in the spark zone. The electrode wears very rapidly if
15
de-ionized water is used in conventional EDM [40], but due to its rapid cooling rate and low
viscosity it is suitable at the highest degree for the Wire-EDM process.
2.3 MAJOR AREAS OF WIRE-EDM RESEARCH
Wire-EDM is a specialized machining process which can machine the parts to the requisite
accuracy [35]. These parts also include the ones with complex shapes and varying hardness.
Generally parts having sharp edges cannot be machined easily, but this process provides the
facility of machining such parts also. The process utilizes basically the same non-contact
sparking phenomenon used in the conventional EDM. Since its introduction, it has become a
simple mean of making tools and dies. It is the best alternative for production of micro scale
parts with better surface finish and dimensional accuracy.
For many past years, this process has proven to be an economical and competitive machining
option. It is capable of fulfilling the most crucial and challenging machining requirements which
have surfaced due to the requirements of the present age that demand shorter product
development cycles and low costs. Research of a tremendous order has been carried out to
successfully explore the ways for obtaining optimized process parameters for different
machining responses analytically. Efforts have also been made by researchers to minimize wire
breakage for improving overall machining reliability. This section gives the review of WIRE-
EDM research that involves the optimization of the process variables along with the survey of
the affect of the various control factors that affect the performance of machining in terms of
response variables. It will also show the controlling of process, and its adaptive monitoring, with
an investigation into viability of different strategies for achieving the optimized conditions of
16
machining. With the development of many hybrid type of machining processes a vast range of
applications has been observed and reported. For easily understanding the process the research
has been classified into two major groups namely “optimization” and the “monitoring/control of
the process”.
2.3.1 OPTIMIZATION OF WIRE-EDM PROCESS
Process optimization is generally a difficult task due to efforts required to regulate control
variables. A change in single variable may affect the machining process in a very complicated
manner [41]. Therefore, variables that can affect the machining process should be studied for
determining the pattern of the process variation. Modeling the machining process is considered
to be an affective tool of addressing the problems associated with co-relation of the control
variables with the machining responses. However, the complex nature of the material erosion
process during machining really needs the application of deterministic and stochastic techniques
[42]. It is therefore, that the optimization of the process has remained a key research area. This
section will provide an over-review of various machining strategies that includes process
parameter design and the appropriate modeling of the processes.
2.3.1.1 PROCESS PARAMETERS DESIGN
The selection of appropriate value for a process parameter is very necessary. It plays an
important role in optimization of the required machining responses. In this section various
statistical and analytical methods are shown that are applied to investigate the affects of these
variables on the machining responses or performance measures like kerf, material removal
rate/cutting rate and the surface finish.
17
DRY WIRE-EDM
Furudate and Kunieda [43] carried out research on dry WIRE EDM. The reaction force of the
process proved to be negligibly small, the wire electrode vibration was minute and gap distance
was narrower to that in the conventional process that uses dielectric liquid. These factors enable
the dry WIRE-EDM to obtain high amount of accuracy in the finish cutting. The workpiece is
not corroded which is an advantage in the manufacturing of molds and dies of high precission.
Wang and Kunieda [44] concluded that the dry WIRE-EDM is suitable for improvement in the
machined surface straightness and for finish cut. Debris around working gap are removed by the
travelling tool and those even in the atmosphere. The straightness in workpiece thickness
direction in this dry process is much better than the process that is carried out in water [45]. In
another study Kunieda and Furudate [46] discovered some flaws of dry WIRE-EDM that
includes lower rate of material removal when compared to the conventional process. In addition
a big flaw during precision finishing cut by the dry EDM is the generation of streaks on the
finished surface. The flaws can be addressed by keeping the depth of cut to a smaller value and
increasing wire speed.
WIRE-EDM IN WATER
Levy [47] carried out experimentation for a high-volume dielectric regeneration and environment
friendly system for the process. In this, a filtration unit that consisted of the membrane
technology was used to obtain a very efficient system of dielectric regeneration. Diane [48]
emphasized on appropriate resin and mix which is utilized in de-ionizing the water used in the
process. Minami et al. [49] have proposed a procedure of applying color to the titanium alloy
while machining to curtail the requirement of any post treatment. Using conventional water
18
WIRE-EDM, the interference phenomenon of light with the anodic oxide film forming due to
electrolysis reaction, forced direct coloring a surface that was re-solidified after being molten.
Average working voltage was responsible for controlling the thickness of the captioned oxide
film. The film thickness was helpful in determining the color tone.
FACTORS AFFECTING THE MACHINING RESPONSES.
It is a complicated machining process that is governed by a vast number of control variables such
as the current intensity, pulse duration and the discharge frequency. Smallest of variations in any
of the control variables will affect the machining responses such as surface roughness and the
cutting rate, which have been declared as the most important aspects of the machining process
[50]. Suziki and Kishi [51] found out that for obtaining a better surface finish the discharge
energy needs to be reduced. Whereas Luo [52] discovered that additional energy was required to
efficiently maintain a higher rate of material removal, but the amount of energy should not also
be that high to damage the wire or cause wire breakage. Few of the authors [53] have carried out
research on the performance of the wire tool that directly affects the accuracy of machining and
various other performance measures. The selection of suitable control variables for the process is
largely based upon the relationship data that plays an important role in relating various control
variables to few machining responses namely the surface finish and material removal rate.
Normally, this was undertaken by consulting the technical data which was rendered by
equipment manufacturer or by relying on the operator’s experience. All of this has lead to the
experiencing of inconsistent performance of machining. Maggi [54] in a study has shown that the
specific control variables settings which were provided by the manufacture were applicable to
common steel grades only and there was a considerable difference when applied to other
19
materials. It proves that the manufacturer data cannot be blindly relied upon. The optimum
control variables settings for cutting of new materials have to be explored for obtaining
optimization of machining responses and this is to be achieved experimentally.
AFFECTS OF THE PROCESS PARAMETERS ON THE CUTTING RATE
Numerous analytical tools have been used for problem-solving and to establish the variables that
are significant. They were also used to study their inter-relationships for obtaining an optimal
machining rate. Konda et al. [55] has classified few of the potential variables that affect the
machining responses in five major categories. They are machine characteristics, the properties of
workpiece materials, component geometry, dielectric fluid and adjustable machining variables.
In addition the technique of the design of experiments was also used for obtaining data and
optimization of the expected possible responses of control variables in design and development
phase of the process. Signal to noise (S/N) ratio technique was applied to validate the results of
the experiment. Neural network system was employed by Tarng et al. [56] coupled with the
simulated annealing algorithm so that a problem of multi response optimization could be solved.
Hence, this was observed that machining variables like open circuit voltage, pulse on/off time,
peak current, table speed, servo voltage, and electrical capacitance were critical factors for
controlling cutting rate and surface finish. Huang et al. [57] after literature survey found out that
much of the published research [41,56,58] was mostly related with the optimization of control
variables for rough machining/cutting operations. They proposed process planning approach for
proceedings to finishing operations from the roughing operations. The results of their
experimentation show that the pulse on time affects the machining rate and surface finish. In
addition, for these responses the distance between the surface of the workpiece and the periphery
20
of the wire is also significant. The affects of discharge energy on the machining rate and surface
finish of the metal matrix composites have also been experimentally investigated by some
researchers [59].
AFFECTS OF THE PROCESS PARAMETERS ON THE SURFACE FINISH
Much of the literature and the published works are available entirely on study of the affects of
control variables settings upon the machined surfaces. Go Kler and Ozano Zgu [62] carried out
research for selecting the most feasible offset and cutting parameter combination for obtaining
the requisite surface quality when both the dielectric flushing pressure and the wire speed were
kept constant. Tosun et al. [63] in an investigation studied the affect of open circuit voltage, wire
speed, the pulse duration and the dielectric pressure upon the surface quality of the machined
workpiece. The results showed that the increase in wire speed, pulse duration and the open
circuit voltage caused an increase in the value of surface roughness, whereas a decrease in the
value of surface roughness was observed with the increase in dielectric flushing pressure. Anand
[64] in another study used fractional factorial design to achieve the most desirable control
variables settings for improving surface roughness and the dimensional accuracy. Spedding and
Wang [65] carried out optimization of control variables settings while applying artificial neural
network modeling technique for characterization of the machined surfaces.
AFFECTS OF THE PROCESS PARAMETERS ON THE KERF WIDTH
Nihat et al. 66 in an experimentation selected wire speed, pulse duration, open circuit voltage,
and flushing pressure to investigate their influence on kerf. They found out that the most
significant parameters were pulse duration and open circuit voltage. The di-electric flushing
21
pressure and wire speed were found to be non significant. Di Shichun et al. 67 carried out
research to analyze the Kerf in micro-WEDM. In this they found out that the kerf was being
affected by two components that is amplitude of wire vibration and breakdown distance. They
built models and relationships among the wire vibration amplitude and control variables were
also analyzed. A. Okada et al. 58 studied the affect of dielectric flushing nozzle stand-off
distance. They found that the flow velocity increased in the kerf by decreasing the stand-off
distance. However, the stagnation area was almost the same for about each stand-off distance,
when the upper and lower distances were same. Stagnation area was formed around the wire with
low velocity of the flow disregarding the flushing pressure values form both the upper and lower
nozzles. In this area the expulsion of debris was not very efficient and therefore flushing from
both the upper and lower nozzles will not be affective always for expulsion of debris from the
kerf. When the flow rate is kept constant, most of the debris is expelled out of the kerf from the
same area. The debris distribution influences the state of the electrical discharges taking place in
the kerf. Secondary discharges occur in the kerf when stagnation of too many debris takes place.
This occurrence of the secondary discharges is concentrated in the same location and leads to
unstable machining performance. W.Y. Peng and Y.S. Liao 69 in their study of kerf found out
that the kerf size depends upon the discharge energy and the cutting wire diameter. On time must
be controlled accordingly to get optimal results. They also proposed to set the servo voltage to a
higher value in order to avoid wire breakage due to the accidental workpiece and wire contact.
AFFECTS OF MACHINING PARAMETERS ON MATERIAL REMOVAL RATE
The influence of control variables upon the material removal rate (volumetric) has been also
considered for measuring the quality of the machining responses. Factorial design was applied by
22
Scott et al. [41] that requires several experimental runs for establishing the most suitable
combination of the control variables. As a result, they realized that dielectric flow rate, wire
tension, and wire speed were not significant, while the pulse frequency, duration and discharge
current were most significant process parameters that were affecting the material removal rate a
great deal. Based on Taguchi method where analysis of variance is also used, Liao et al. [58]
came with a comparatively new approach for determining the control variables optimal setting.
In this research it was found out that surface finish and material removal rate are heavily affected
by the pulse on time and the rate of table feed. The discharging frequency can be controlled by
controlling pulse on time in order to prevent sudden breakage of wire that causes unnecessary
delays in the process and affects its reliability. Similar kinds of results have been produced in
another study in which S/N ratio analyses were used which was carried out by Huang and Liao
[60]. Aiming for the determination of material removal rate and surface finish at different control
variables settings, another study has been carried out by few researchers [61]. The results
obtained have been utilized using a thermal model for analyzing wire breakage phenomena.
2.3.1.2 PROCESS MODELING
Furthermore, the modeling of the process with the help of various mathematical techniques also
has been applied by many researchers in order to co-relate efficiently the various different
machining performance measures of WIRE-EDM to vast number of process parameters.
Modeling techniques have been developed by Spedding and Wang [70] with the use of response
surface methodology and the artificial neural network technology for predicting machining
performance measures such as surface finish, cutting rate and surface waviness. In this the
control variable levels range was kept large in order to achieve better results. A solid modeling
23
method that can accurately represent a geometry produced by the WIRE-EDM process has been
proposed by Liu and Esterling [71]. On the other hand Hsue et al. [72] have succeeded in
developing a model that estimates material removal rate in case of geometrical cutting by taking
into account the deflection of the wire and with the wire centre transformed exponential
trajectory. Spur and Schonbeck [73] carried out WIRE-EDM of a work piece for studying the
workpiece material influence. Anodic polarity was used for the workpiece, and a theoretical
model was proposed by studying influence of the type of pulse and that of the workpiece
material. In order to develop a simulation system for the accurate reproduction of the discharge
phenomenon, Han et al. [74] carried out another study in which they were successful in
achieving the objective. High precision optimal machining conditions are established
automatically as this system also takes in to account the application of adaptive control.
2.3.2 PROCESS MONITORING AND CONTROL OF WIRE-EDM
Affective monitoring and control of the process requires efficient application of adaptive control
system. An overview of the modern monitoring and control systems is provided in this section
which includes the wire breakage, the fuzzy and the self tuning adaptive control systems.
2.3.2.1 FUZZY CONTROL SYSTEM
Generally servo feed control system are comprise of the proportional controllers. This has been
done for evaluating and monitoring the condition of the gap during the process of machining.
However, it was found out that the machining conditions were the main cause for limiting the
performance of these controllers. Machine conditions are expected to have major variations as
the applied control variables settings are varied. A study was carried out by Kinoshita et al. [75]
24
on the gap state for different settings of wire speed, feed rate, tension, and the electrical variables
during the machining process. As a result of this, based on explicit statistical and mathematical
models, a number of conventional control algorithms were developed for the process [76–80].
Various other have carried out research on discrimination system of pulse [81,82]. They have
developed a system in which under different conditions of machining analysis and monitoring of
the pulse trains can be carried out quantitatively. These kinds of control systems will not be able
to cater for the gap condition in case an unexpected disturbance occurs but can be applicable to a
wide range of the conditions of machining [83].
For achieving the optimization of the machining process fuzzy logic has been applied
successfully in the recent past. According to the various authors it has been derived by
implementing a control strategy that basically captures the experience of the operator for the sake
of carrying out the desired machining operation [84]. Furthermore, fuzzy logic controller does
not need very complex mathematical models that adapt to WIRE-EDM operation’s dynamic
behavior [85]. Several authors [83,86] proposed adaptive control systems and sparking frequency
monitoring for better operation of the machining process. Most are based on few strategies of
adjusting that are applied to wide range of the machining conditions and the fuzzy logic. Another
research by Liao and Woo [87] was carried out in which they were able to design a fuzzy
controller to control the different characteristics of the discharge pulse. The discharge noise was
isolated with the use of an online monitoring system for pulse. The ignition delay discrimination
for each pulse was also applied. Fuzzy control systems have been applied successfully by these
researchers to obtain the desired results in order to improve this process of machining.
25
2.3.2.2 WIRE-INACCURACY ADAPTIVE CONTROL SYSTEMS
The most unwanted characteristics in machining is the wire breakage during the operation [88] as
discussed earlier also. The component quality, performance of machining, and accuracy are
affected a great deal by this flaw or phenomenon. Development of adaptive control system for
wire breakage has been attempted by many researchers. Adaptive control system should be such
that it can monitor the online conditions of machining and can identify any abnormalities. The
system should have a solid strategy that can prevent wire breakage without having any adverse
affects on the performance of machining or the requisite machining responses. This section
presents the research collected from the available literature survey that provides information
about the different characteristics of the tool wire like wire vibration, wire breakage and wire lag.
WIRE BREAKAGE
Many control strategies have been designed in order to address the phenomenon of wire
breakage. Few of them will be discussed in this section to have an overview of the available
literature on the problem. Kinoshita et al. [88] in a study have shown that prior to the wire
breakage there is a sudden rise in the frequency of pulse for the gap voltage and continues from 5
to 40 micro seconds. In this they incorporated a monitoring and control system. The function of
this system was to switch off the pulse generator and the servo system before wire breakage.
Thus they were able to increase the efficiency of machining. Several authors [89,90] have shown
that the increase in the localized temperature is the outcome of electrical discharges
concentration at one point and this results in the form of wire breakage. In these studies the main
area of focus was the spark location and the systems were developed for reducing the amount
discharge energy but it is pertinent to mention here that the material removal rate was either
26
ignored or was not given the due consideration. Some researchers have related the breakage of
wire to the short circuits quantity. Prior to the breakage of the wire these short circuits duration
should prolong to 30 micro seconds for the wire breakage to occur [91]. Some of the researchers
have come to a conclusion that a sudden increase of the spark energy causes the wire breakage
[92]. It was also revealed that the material removal rate was affected by their proposed system of
monitoring and control. This system was based upon the online analysis of real time regulation
of the factors like pulse off time and sparking frequency. The remedy to this was provided by
Liao et al. [93] by means of incorporating a newer pulse discrimination system which was
computer aided and relating the material removal rate to machining parameters. For improving
machining rate the discrimination system was based on the pulse train analysis. A self-learning
fuzzy control system which was applied by Yan and Liao [94,95] which not only served as a
means of controlling the sparking frequency but it also maintained a better material removal rate
by constant feed rate and incorporating few adjustments in the online pulse off time while the
feed-rate was kept constant.
Excessive thermal loading is also responsible for the wire breakage. It results in the unwarranted
production of heat on the wire. During machining most of the generated thermal energy is
transferred to the wire while the remaining is being taken away by the flushing fluid or radiated
away [92]. Depending upon the wire’s thermal properties whenever there is an increase in the
instantaneous energy rate beyond some specific limit, the wire shall break. Some of the authors
[96–98] carried out studies on the thermal loading of the wire. In most cases they have analyzed
the relation between the used control variables and the thermal loading of the wire. Thermal
models were developed for simulating the process in real time. Mechanical strength of wire
27
material cannot be neglected as it has a significant effect on frequency of wire breakage in
addition to the sparking characteristics and temperature distribution.
The sparking frequency monitor has been developed by Rajurkar and Wang [99]. This monitor
had a thermal load detector and was meant for on-line monitoring and control for preventing
wire breakage. Thermal model was used for analyzing the phenomena of wire breakage. The
multi objective model was used for establishing relationship between surface finish and material
removal rate under optimal settings. Vibrational behavior of the wire in another study was
considered by Puri and Bhattacharya [100]. Using an analytical approach, they have proposed
the consideration of multiple discharges as a solution to equation of wire vibrational behavior.
They concluded that wire vibration will increase between the wire guides in machining of a
thicker workpiece as compared to a comparatively thinner workpiece.
WIRE LAG AND WIRE VIBRATION
These are the main factors that contribute to the geometrical inaccuracy. There are few forces
that act on the wire during cutting operation. In this case the preplanned programmed path cannot
be followed by the wire and a deviation is experienced. Plasma of erosion mechanism causes the
gas bubbles to be produced frequently and this production of bubbles creates these forces. Some
hydraulic forces are produced due to flushing [101,102]. To cater for almost all these forces, the
application of the axial forces keeps the wire straight. Few more forces act on the wire tool that
is spark inherent electro dynamic forces and electro static forces. Static deflection is obtained as
a consequence of all these forces. It is translated in to the wire lag affect during machining and
has been studied critically for producing accurate path of cutting tool. Many authors [103,104]
28
have made an attempt to present accurate mathematical models of the process and have
performed research on geometrical inaccuracy caused due to wire lag. In this, using a control
algorithm, Beltrami and Dauw [105] attempted to monitor and control the wire position online
with the help of an optical sensor. This enabled machining virtually any contour at a relatively
higher speed. Many geometrical tool motion compensation techniques [106,107] have also been
developed to prevent wire breakage. It includes increasing of the machining gap while machining
small radii or high curvature profiles. Based on fuzzy logic, for improving concentrated sparking
at the corners and the machining accuracy without affecting rate of material removal, Lin et al.
[108] developed a control strategy. Guo et al. [109] carried out another study on ultrasonic wire
vibration. They found out that the hybrid form of the process with ultrasonic vibrations can
facilitate the production of multiple channel discharge. In this way the energy is utilized
efficiently to support more material removal rate and the surface finish is also improved.
Discharge concentration conditions are improved due to vibration of wire at higher frequencies
that helps in minimization of the probability of the wire breakage. Another study was carried out
by Guo et al. [110] on material removal rate and surface finish. They focused on the ultra sonic
aid and concluded that the surface roughness and material removal rate can be improved with use
of this aid. An improvement of 30 % in the material removal rate was reported by the
researchers.
The dynamics of the cutting wire are controlled during machining for avoiding the inaccuracies
that may be experienced during cutting. Some of the discussions [22,111] are available on the
development of control system for providing compensation against the behavior of wire. It has
been found out by Dauw et al. [112] that when the entire wire and the guides are fully submerged
29
in water, the amount of vibrations is reduced. A few authors [113] carried out research by
considering the forces produced in single discharge. They carried out the analysis of the wire
vibration and also produced mathematical models.
2.3.2.3 SELF-TUNING ADAPTIVE CONTROL SYSTEMS
In machining of the workpieces of varying height, some control strategies have been explored in
the recent past for controlling the power supplied to wire tool. In order to control the sparking
frequency while taking into account the height of the workpiece measured online, Rajurkar et al.
[114,115] have produced multiple inputs model with an adaptive control system. Some of the
researchers have found out that during machining if the workpiece height is varying, it will
cause sudden variations in thermal density [88,91] and ultimately will result in wire breakage.
Few researchers carried out to many experiments and statistical analysis to develop explicit
mathematical models [80]. For addressing the problem of wire breakage, Yan et al. in a study
used fuzzy control logic [116] and the neural networks in order to estimate the height of
workpiece being machined.
In order to control the worst machining situation the idea of developing knowledge based control
system has been also undertaken. A system comprising of three modules has been proposed by
Snoeys et al. [117]. These modules included the process control, work preparation and operator
assisted fault diagnosis that enabled the control and monitoring of WIRE-EDM process. In this
the work preparation module determined the optimal control variables settings, whereas the fault
diagnostics and operator assistance databases are meant to diagnose the machining fault and
advise the operators accordingly. Huang and Liao [118] studied and emphasized on the need of
30
the fault diagnostics system in the WIRE-EDM. In their study they recommended artificial
neural network based expert system. This system was developed for the fault diagnosis and
routine maintenance. A thermal model was produced by Dekeyser et al. [119] for precisely
controlling and predicting thermal load on the wire during machining operation.
2.4 GAP ANALYSIS AND PROBLEM STATEMENT
In this literature survey, it has been noticed that much work has been carried out on optimization
of machining responses but there still is a room for improvement especially in case of material
removal rate. Although many researches were available on the machining parameters for
optimizing machining responses but most of the researchers have chosen some specific
parameters to define the machining outcomes. Therefore the need to use a larger number of
machining parameters was felt. Furthermore, very little research work was found on Tungsten
Carbide and it was decided to use the same material in this research work. Apart from selection
of machining parameters some other non-machine factors were also considered in this research
work such as thickness and hardness as very little work was available on these two aspects. So
this research was focused on the areas where the literature contained a very little research. In
short, the research was focused on machine parameters, non-machine parameters like thickness
and hardness, and Tungsten carbide as workpiece material.
31
CHAPTER3DESIGNOFEXPERIMENTS
A technique in applied statistics which includes planning, interpretation, analysis and conduct of
specific experimental runs is known as design of experiments [120]. Its objective is the
evaluation of variables that affect the response variables or the required output of any
experimentation. Manipulation of the input variables can be achieved with this tool for observing
the influence on the required machining response in this case. Interactions of the variables, if not
always, can play a major role some time on the output or response. DOE can also help in
identifying and determining the affects produced by these interactions on the response or the
output. DOE is a flexible tool due to versatility of its use in a vast variety of situations or
experiments. It can be used not only in full factorial designs where all of the experimental runs
are required to be executed but can be used in fractional factorial designs also where a part or
portion of the experimentation is required to be carried out.
DOE is often used if the response variable is suspected to be affected by multiple variables
instead of a single variable. DOE is used not only to establish the relationship between the input
variables and the response but also serves as a means of developing mathematical models for
these relationships. The affects of the control variables on the output or response variable can be
obtained efficiently and effectively if the experimentation is logically planned. In most of the
experimentation the levels of few variables are kept constant while the level of only one variable
is varied at a time to observe its affect.
32
In the present research, for statistical analysis, many researchers are following the technique
introduced by Fisher in 20th century. Fisher emphasized on spending more time on the design of
experiment execution rather than wasting more time in tackling or solving the problems at
analysis phase when the experiment has already been executed.
3.1 TAGUCHI METHOD
The Taguchi method was developed by Dr. Genichi Taguchi of Japan. Applying robust design of
experiments, this technique has been used to reduce process variations [121]. The mean and the
variance of process performance determine the functioning quality of a process. A model was
developed by Taguchi for experimental design in order to investigate the affect of different
control variables on the mean and variance. Taguchi technique involves the identification of the
levels at which the control variables should be operated and the use of the orthogonal array to
observe the affects of control variables on the response variable. In this method execution of all
the experimental runs or the combinations is not mandatory but selected pairs of experimental
runs can be executed to save time and efforts. This technique can always be used in situations
when the contribution of few variables is significant, the variables are from 3 to 50 in number
and incase where interactions of variables is encountered.
The selection of the arrays is governed by the number of levels and the number of control
variables. After the selection of the array the experimental runs are executed as the array
specifies. The collected data is then analyzed for determination of most significant variables by
ANOVA and then selection of the optimal conditions or levels of the control variables by using
S/N ratio analysis.
33
3.1.1 PHILOSOPHY OF THE TAGUCHI METHOD
a. Introduction of quality in a process means the quality of tolerance design, parameter
design, and system design. Parameter design is the identification of the control variables
that have a major affect on the product’s quality. After this they are designed to produce
the requisite quality.
b. Deviation from the target should be minimized for obtaining quality. It means high value
of signal to noise ratio.
c. The cost of quality should be measured as a function of deviation from the standard and
the losses should be measured system wide.
In the next section, the specific steps involved in the application of the Taguchi method will be
described and examples of using the Taguchi method to design experiments will be given.
3.1.2 TAGUCHI METHOD DESIGN OF EXPERIMENTS
Generally following steps are involved in Taguchi Method:
a. Process objective for a response variable must be defined; it can be a target value.
b. Selection of appropriate control variables is required that are expected to affect the
response variable.
c. This follows the selection of orthogonal arrays which will depend upon the number of
control variables and their levels.
d. Execution of experimental runs to measure the response variable.
e. Statistical analysis of the data to select most significant control variables.
34
Figure 3.1 describes the pictorial depiction of these steps and additional possible steps,
depending on the complexity of the analysis.
Figure 3.1: Steps involved in the Taguchi Method 121
3.1.2.1 TAGUCHI LOSS FUNCTION
Minimizing the variance of process performance to reduce the losses to the manufacturer and the
society is the primary objective of Taguchi method. The loss function is defined as the difference
35
between the target value, τ, and the measured value, y, of a performance characteristic and is
shown as below [121]:
l (y) = kc ( y - τ )2 ……………………….(3.1)
The kc is a constant and can be calculated by using acceptable interval, delta.
kc = C / ∆2 ………………………………(3.2)
The kc is difficult to be determined as the τ and C are difficult to define at times. For minimizing
the performance characteristic value, the loss function will be calculated as follows:
l (y) = kc y2 …………………………….(3.3) where τ = 0
For maximization of the performance characteristic value, the loss function will be calculated as
flow rate, cutting speed and the wire offset. Nihat et al. 127 used Taguchi design of
experiments to investigate the affects and optimization of control variables upon material
removal rate and the kerf. The selected control variables were wire speed, pulse duration,
42
dielectric flushing pressure and the open circuit voltage. The appropriate levels for these control
variables were established. Then the influence of these control variables was found out with
ANOVA. The optimal control variable setting was obtained by the analysis of signal to noise
ratio.
43
CHAPTER4EXPERIMENTATION,STATISTICALANALYSESAND
OPTIMIZATIONWITHAFOCUSONWORKPIECETHICKNESS
In this chapter information on DOE and experimental setup has been provided. The collected
data has been presented. It also includes the affects of variables, such as Wire-EDM process
parameters, on the machining responses that are kerf width, surface roughness and the material
removal rate of tungsten carbide samples machined by Wire-EDM. Workpiece thickness was
also taken as a process parameter along with various other machine-specific process parameters.
Rationale behind the selection of workpiece material for this experimentation has also been
discussed in this chapter. So, in total 08 process parameters including thickness were taken with
three levels each. Taguchi technique has been applied for design of experiment. ANOVA was
carried out after obtaining the machining responses to determine the significant factors or
process parameters. Finally optimization of the machining responses was carried out.
4.1 INTRODUCTION
S. H. Lee and X. P. Li 128 carried out a research to study the influence of process parameters
of EDM on the machining responses. They evaluated the effectiveness of the EDM process,
while using tungsten carbide as workpiece material, in terms of the material removal rate, the
relative wear ratio and surface finish of the end product. M.P. Jahan et al. 129 in the study
investigated the influence of major process parameters on the performance of micro-EDM of
tungsten carbide with a focus on obtaining quality micro-holes. However it was strongly felt that
wire EDM of tungsten carbide was also required to be carried as sufficient published material
44
was not available on the relationship of these input and output responses. So, more work was
required to be carried out in order to affectively contribute to the available research literature.
Tool making is the most typical application of the wire EDM 130,132. Wire EDM is applicable
to all kinds of mould development, intricate shapes, special parts, stamping dies, extrusion dies
prototype parts and machining of hard materials such as tungsten-carbide machining136,141.
Many research papers are available on materials such as Tool Steels, Die Steels, Germanium and
Titanium alloys 131-135 etc. However, very little work has been carried out on tungsten-
carbide machining.
A.B. Puri and B. Bhattacharyya 126, in 2003, performed wire EDM on a workpiece of Die
Steel in order to investigate the affects of machining parameters on surface roughness and cutting
speed. In this they were successful in determining some most significant control factors. In their
work they used large number of machine-specific process parameters in order to have a broader
view of the process. However, they did not consider the affect of the workpiece thickness as it
was not in their scope of this study.
Dinesh and Eberhard131, in 2009, carried out kerf width analysis of germanium wafers that
were machined by Wire-EDM. They developed mathematical model for predicting the kerf
width against various process parameters. In these models the workpiece thickness was kept
constant. Hence the affect of workpiece thickness on kerf width was not available.
45
Therefore, in the current experimentation tungsten-carbide has been selected as workpiece
material and the thickness has also been selected as a process parameter in addition to machine-
specific process parameters.
4.2 EXPERIMENTATION
The experimental design, experimental equipment, instrumentation, and the data acquisition
procedures to meet the objectives of this investigation will be discussed in this section.
4.2.1 DESIGN OF EXPERIMENT
In this study large number of process parameters was taken in order to enhance the accuracy of
the results126. There were a total of 08 parameters with 03 levels each. The parameters along
with their levels are listed in table 4.1.
Table 4.1:- Data Summery of Experiments Factor Code Process parameters Units Level 1 Level 2 Level 3
A TH Thickness mm 25.4 50.8 76.2 B OV Open Voltage Volts 75 90 105 C ONT On Time μs 3 4 5 D OFT Off Time μs 20 25 30 E SV Servo Voltage Volts 40 50 60 F WF Wire Feed Velocity mm/sec 30 60 90 G WT Wire Tension Grams 1000 1600 2200 H WL Dielectric Pressure Kg/Sq.cm 10 12 14
Since there were 08 process parameters each with three levels in the experiment, if a full factor
experimental design had been used, there would be 6561 runs which are too many to be
conducted. It would have been time consuming and too expensive to go about. The solution was
46
to use only a fraction of these runs as specified by the full factorial design. There are various
strategies that ensure an appropriate choice of runs, as discussed in chapter 3, for example the
Taguchi's orthogonal scheme 126. Since each input parameter in this experiment was
considered independent, an orthogonal experimental scheme could be used to reduce the total
number of runs. Therefore an orthogonal array L27(3x13) has been used. In this, out of 13
columns, 08 columns were assigned to 08 process parameters and the rest 05 columns were left
alone for interactions. Thus the Taguchi Orthogonal Array in this case appears as table 4.2.
Table 4.2:- Experimental Runs Specified by Taguchi L27(3x13) Orthogonal Array EXP.NO. A B C D E F G H I J K L M
1 25.4 75 3 20 40 30 1000 10 x x x x x
2 25.4 75 3 25 50 60 1600 12 x x x x x
3 25.4 75 3 30 60 90 2200 14 x x x x x
4 25.4 90 4 20 40 30 1600 12 x x x x x
5 25.4 90 4 25 50 60 2200 14 x x x x x
6 25.4 90 4 30 60 90 1000 10 x x x x x
7 25.4 105 5 20 40 30 2200 14 x x x x x
8 25.4 105 5 25 50 60 1000 10 x x x x x
9 25.4 105 5 30 60 90 1600 12 x x x x x
10 50.8 90 5 20 50 90 1000 12 x x x x x
11 50.8 90 5 25 60 30 1600 14 x x x x x
12 50.8 90 5 30 40 60 2200 10 x x x x x
13 50.8 105 3 20 50 90 1600 14 x x x x x
14 50.8 105 3 25 60 30 2200 10 x x x x x
15 50.8 105 3 30 40 60 1000 12 x x x x x
16 50.8 75 4 20 50 90 2200 10 x x x x x
17 50.8 75 4 25 60 30 1000 12 x x x x x
18 50.8 75 4 30 40 60 1600 14 x x x x x
19 76.2 105 4 20 60 60 1000 14 x x x x x
20 76.2 105 4 25 40 90 1600 10 x x x x x
21 76.2 105 4 30 50 30 2200 12 x x x x x
22 76.2 75 5 20 60 60 1600 10 x x x x x
23 76.2 75 5 25 40 90 2200 12 x x x x x
24 76.2 75 5 30 50 30 1000 14 x x x x x
25 76.2 90 3 20 60 60 2200 12 x x x x x
26 76.2 90 3 25 40 90 1000 14 x x x x x
27 76.2 90 3 30 50 30 1600 10 x x x x x
47
4.2.2 EXPERIMENTAL SETUP
A Wire-EDM machine G43S (CHMER EDM CHING HUNG Machinery & Electric Industrial
Co. Ltd. TAIWAN) was used for conducting the experiments as shown in figure 4.1.
Figure 4.1: Wire-EDM machine G43S
This machine uses de-ionized water as dielectric. A brass wire of 250 μm diameter (tensile
strength 800-1000 N/Sq.mm) was used as the tool electrode (cathode). The workpiece material
(anode) used was Tungsten Carbide (0.88% WC, 12% Co, grade DK 500 UF, 92.4 HRA, density
14.15 g/cm3). The tungsten carbide bars were available in square cross section of 12mm x 12 mm
and of about 7 inches length. They were first lengthwise cut to obtain three levels of thickness
that is 1 inch, 2 inches and 3 inches as shown in figure 4.2 and 4.3. It is pertinent to mention here
that this pre-experimentation cutting of samples to desired thickness level helped a lot to
determine the workable levels of machine process parameters. This is because in Wire-EDM the
wire breakage is a biggest hurdle in completing a run and that happens when a too large or small
value for process parameters is selected at individual levels. During this cutting the limitations of
48
Figure 4.2: Samples preparation
Figure 4.3: Samples of 1, 2, and 3 inches thickness
49
the process were well understood and process parameter levels were selected accordingly to
efficiently conduct the runs and save time that is wasted in case of wire breakage. According to
design of experiments, 27 runs were required to be carried out. Each run was repeated three
times to ensure good repeatability. So in total 81 samples were prepared, 27 for each of the three
thickness levels. A special clamp was made for holding the workpieces during cutting as shown
in figure 4.4.
Figure 4.4: Clamping of workpiece
The samples were machined with a total depth of cut of 12 inches in “L” shape, 7 mm in x-axis
and 5 mm in y-axis. This was done intentionally to increase the cutting time and material
removal for obtaining better and reliable data for onward analysis. The photomicrographs of few
of the workpieces are shown in figure 4.5a, b and c.
50
a
b
C
Figure 4.5: Photomicrographs of the cut workpieces
4.2.3 DATA COLLECTION
In this section ways and means that were used for data collection will be discussed for the three
machining responses. After the data were obtained, the statistical analysis of the obtained data
was carried out for validity of the obtained data. In this the standard deviation and the confidence
limits were calculated. The standard deviation of a set of “N” numbers X1, X2, X3, …….., XN is
denoted by “s” and is defined by [138]:
N
s = √ ∑ (Xj – Xm)2 / N ………………………..(4.1) j=1
Here, N is the sample size, Xm is the arithmetic mean, Xj is the jth reading or number. Thus “s” is
the root mean square of the deviations from the mean, or, it is sometimes called the root mean
square deviation. In the case of KF, Ra and MRR, N=3. Standard deviation for each run for the
three responses was calculated. Some fixed relations were used to establish the level of
confidence that are 68.27 % cases are included between Xm + s, 95.45 % of the cases are
included between Xm + 2s and 99.73 % of the cases are included between Xm + 3s.
4.2.4 RESPONSE VARIABLES
There were three response variables that were measured as per the scope of this thesis. This was
the most difficult part of this research work as any single mistake in the measurements could
51
change the results of the study drastically. Utmost care was taken in obtaining the data. The data
acquisition process for the three response variables will be discussed one by one in the following
sections.
4.2.4.1 KERF WIDTH
Kerf width is the width between each side of the cut as shown in the figure 4.6. The kerf width
was measured on a Universal Profile Projector (UK) of the accuracy 0.001 mm. It was measured
at three different points of the kerf width in each run and each run was repeated three times.
Since there were two replicates for each run, therefore total of 09 readings for each run were
obtained. These readings were then divided into top, side and bottom measurements for the
purpose of analysis. The combined data for kerf width along with all the 09 readings for each
experiment is shown in table 4.3a.
Figure 4.6: Schematic presentation of Kerf Width
52
Table 4.3a:- MEASUREMENT OF KERF WIDTH (KF) EXP.NO.