This article was downloaded by: [Syracuse University Library]

On: 28 April 2013, At: 08:26

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

Machining Science and Technology: An International

Journal
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/lmst20

Improving Electrical Discharge Machined Surfaces

Using Magnetic Abrasive Finishing
a a a b
Biing-Hwa Yan , Geeng-Wei Chang , Jung-Hsien Chang & Rong-Tzong Hsu
a
Department of Mechanical Engineering, National Central University, Chung-Li, Taiwan
b
Department of Mechanical Engineering, Lee-ming Institute of Technology, Tai-Shan,
Taipei, Taiwan
Version of record first published: 07 Feb 2007.

To cite this article: Biing-Hwa Yan , Geeng-Wei Chang , Jung-Hsien Chang & Rong-Tzong Hsu (2004): Improving Electrical
Discharge Machined Surfaces Using Magnetic Abrasive Finishing, Machining Science and Technology: An International
Journal, 8:1, 103-118

To link to this article: http://dx.doi.org/10.1081/MST-120034246

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss, actions,
claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
 MACHINING SCIENCE AND TECHNOLOGY
Vol. 8, No. 1, pp. 103–118, 2004

Improving Electrical Discharge Machined Surfaces

Using Magnetic Abrasive Finishing#
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

Biing-Hwa Yan,1,* Geeng-Wei Chang,1 Jung-Hsien Chang,1

and Rong-Tzong Hsu2
1
Department of Mechanical Engineering, National Central University,
Chung-Li, Taiwan
2
Department of Mechanical Engineering, Lee-ming Institute of Technology,
Tai-Shan, Taipei, Taiwan

ABSTRACT

A recast layer is invariably present on surfaces produced by electrical discharge

machining (EDM). For some metals with high hardness, the recast layer may
contain micro-cracks. This damaged layer can affect the service life of the parts
produced by this method. This investigation demonstrates that magnetic abrasive
finishing (MAF) process using unbonded magnetic abrasives (UMA), can
improve the quality of EDM machined surfaces effectively. The UMA used
herein is a mechanical mixture of steel grit and SiC abrasive. SKD11 tool steel
was used as the workpiece. Experimental results show that the recast layer and
micro-cracks on EDM machined surfaces can be completely removed and a new
surface of roughness on the order of 0.04 mm Ra can be produced. Additionally,
experiments using the Taguchi method and L18 orthogonal array enable the
determination of the optimum process conditions for improving the surface

#
This article has not been published elsewhere, nor has it been submitted for publication
elsewhere.
*Correspondence: Biing-Hwa Yan, Department of Mechanical Engineering, National Central
University, Chung-Li 32054, Taiwan, R.O.C.; Fax: þ886-3-425-4501; E-mail: bhyen@cc.ncu.
edu.tw.

103

DOI: 10.1081/MST-120034246 1091-0344 (Print); 1532-2483 (Online)

Copyright & 2004 by Marcel Dekker, Inc. www.dekker.com
 ORDER REPRINTS

104 Yan et al.

finish. Further, the significance of the control factors was identified with the
assistance of analysis of variance (ANOVA), and the optimum combination of
the process parameters was verified by conducting several confirmatory
experiments.

Key Words: Magnetic abrasive finishing; Recast layer; Micro-cracks; Taguchi

method; ANOVA; EDM.

INTRODUCTION
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

Electrical discharge machining (EDM) is an important technology in the field of

mechanical machining, especially in the die and mould industry. EDM is preferred
for making parts from metals with high hardness or making parts with complex
shapes. However, an EDM machined surface is covered with a recast layer, which
generally features craters and micro-cracks, especially for some metals with high
hardness. The surface quality is poor and the life of a die is significantly reduced due
to the existence of the recast layer. Eliminating the recast layer by manual polishing
is often used to finish the die cavity surfaces. However, such a method clearly lacks
both efficiency and precision. Thus, an alternate finishing process for EDM
machined surfaces is required to improve the surface quality and to extend the life of
the die in service. This study demonstrates that the MAF is a precise and efficient
process to improve the EDM machined surfaces in comparison with manual
polishing.
Magnetic abrasive finishing is a precise micro-polishing process that depends on
the magnetic field in the working gap to retain the ferromagnetic particles and to
apply abrasion pressure to the work surface by a magnetic force (Krymsky, 1993).
Since Korgalov introduced this process in 1938, fundamental research has been
performed in several countries (Anzai et al., 1993; Kim and Choi, 1995; Kremen
et al., 1996). Shinmura et al. have been involved in extensive research in Japan since
1980, and have reported many outstanding achievements (Shinmura et al., 1990;
Shinmura and Yamaguchi, 1995; Yamaguchi and Shinmura, 1999). Chinese
researchers have also been actively involved in this field over the past decade. A
novel machine with CNC system designed for MAF applying on the surface with
complex shapes has been completed by Zhao and Jiang in 2000 (Zhao and Jiang,
2000).
While producing neither a deteriorated layer nor micro-cracks on the finished
surface, MAF offers the advantages of self-sharpening and self-adaptability, and the
finishing tool requires neither compensation nor dressing. In the past, most
researchers have used a sintered magnetic abrasive that was held in a ferromagnetic
matrix. The UMA used in this study is simply a mechanical mixture of steel grit (SG)
and SiC abrasive (SA) with SAE30 lubricant. Thus, the SA is not physically bonded
to the ferromagnetic particle and can move freely within the constraints of the
adjacent SG. A previous study by the authors (Chang et al., 2002) demonstrated that
the finishing characteristics of UMA, with a mixture of larger SG and smaller
SA, are as good as those of sintered magnetic abrasive.
 ORDER REPRINTS

Improving EDM Surfaces Using MAF 105

Magnetic abrasive finishing is currently applied to industrial manufacturing, and

some products have been produced using this technique. Although the finishing
characteristics of MAF on both magnetic and nonmagnetic workpieces, with either a
circular or a curved surface, have been studied, MAF has not been used to improve
EDM machined surfaces. This article will show that the MAF using UMA can be
used to eliminate the recast layer effectively on EDM machined surfaces and a
mirrorlike finished surface can be obtained. Although cylindrical workpieces are
used in this study, this method can be applied to complex shapes to remove the recast
layer (Zhao and Jiang, 2000). Within (Zhao and Jiang, 2000), the apparatus was
designed as a vertical milling machine, and the traveling path of cutting tool was
controlled by a CNC system. The cutting tool, which was vertically mounted, was a
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

magnetic brush of the magnetic abrasives (a schematic is shown in Fig. 5). SAE 1045
carbon steel was used as the workpiece, and a multi-curved surface was produced by
Wire EDM to form the complex shape. The roughness of the work surface was
initially about 2.4 mm Ra. After having been processed for six passes with a feed
speed of 20 mm/min on the work surface, it was improved to 0.2 mm Ra.
Advanced EDM systems are capable of generating good surface finish under
0.1 mm Ra. However, this process takes long duration of time; furthermore, the
recast layer still remains on the work surface even though it is extremely thin. In
contrast, using the traditional EDM process followed by MAF will not only produce
a superior refined surface but also involve a short time; moreover, no recast layer
remains on the work surface.

2. PROCESSING PRINCIPLES

In MAF operation, the gap between the workpiece and the magnetic poles is
filled with a mixture of SG and SA. Figure 1 is a schematic showing magnetic field
distribution and magnetic force on a SG within cylindrical MAF (Chang et al.,
2002). The magnetic force, F, is proportional to both the susceptibility and the
volume of SG, the magnetic field strength and its gradients (Shinmura et al., 1990).
The SG forms a flexible magnetic brush along the line of magnetic force within the
working gap, and this will cause a pressure, P, to the work surface. The pressure, P,
acts on the SA beneath the SG, and generates abrasion. The abrasive pressure of the
floating SA must come from the contiguous SG, or else the SA does not cut, since
it is not ferromagnetic.
During finishing, the forces that act on a SG near the work surface are in
position ‘‘B’’ (Fig. 1). A cutting resistance, Rt, acts on the SG in the tangential
direction of the rotational motion, due to the rotation of the workpiece.
Furthermore, a normal force, Rn, exerted by the SG, pushes the work surface,
while simultaneously, a magnetic force, Fm, acts on the SG in a direction opposite to
that of Rt, because of the presence of magnetic field strength gradients in the
working gap. The resultant force of the SG determines both its motion and its
stability. Both the motion and the stability of SG significantly affect the finish. When
Fm exceeds Rt, the SG transmits abrasion pressure effectively to the SA beneath it
within the working gap. This finishing process has been proven to yield a superior
surface (Chang et al., 2002). When Fm is smaller than Rt, the SG rolls on the work
 ORDER REPRINTS

106 Yan et al.

Downloaded by [Syracuse University Library] at 08:26 28 April 2013

Figure 1. Schematic showing magnetic field distribution and magnetic force on a

ferromagnetic particle.

surface. However, when Fm is very small, the SG splashes out of the working gap.
The finished quality is reduced since the rolling SG stops transmitting an abrasion
pressure to the SA beneath it. The SG must be sufficiently stable not to splash out of
the working gap, but also to roll therein, and thus ensure a successful finishing
process. Independent of the cutting resistance, the stability of the SG during
finishing is greatly influenced by the magnetic flux density, which can be controlled
by the input current to the electromagnets, within the working gap.

3. EXPERIMENTAL SETUP AND TEST CONDITIONS

Figure 2 shows the experimental apparatus used. A brushless DC motor, not

shown in the figure, was used to rotate the workpiece. The chuck together with the
workpiece was vibrated to enhance the finishing. The axial vibration was generated
by an eccentric cam mechanism, which was driven by an induction motor and a
frequency converter. Two solenoid coils were connected in series as electromagnets,
and soft iron with a high relative magnetic permeability was used for both the
magnetic cores and the poles. Copper wire of f1.0 diameter was wound 2150 times
around each core, to form a solenoid.
 ORDER REPRINTS

Improving EDM Surfaces Using MAF 107

Downloaded by [Syracuse University Library] at 08:26 28 April 2013

Figure 2. Three dimensional view of experimental apparatus.

Based on past experience, eight process parameters, each with three levels,
except for the working clearance with two levels, were selected for this study. The
eight process parameters have been proven to affect the improvement of the surface
finish. Table 1 specifies these eight control factors and their levels. The experiments
were planned with the help of the Taguchi method and the L18 (21
37) orthogonal
array was adopted to minimize the number of experimental trials and to obtain a
superior estimation. The L18 orthogonal array was often used to study the main
effects of all the control factors. All the main effects of the control factors, and the
implicit interaction of the factors that occupied the first and the second column
within the L18 array, were analyzed.
The SKD11 workpiece was quenched to a hardness of HRC55, while the
hardness of SG was HRC63–68. The workpiece was a cylindrical bar with a diameter
of f15 mm and a length of 80 mm. Firstly, an 8 mm wide peripheral EDM machined
 ORDER REPRINTS

108 Yan et al.

Table 1. Control factors and levels for experimentation.

Level
Control factor 1 2 3

A. Working clearance (mm) 1 2

B. Average particle size of SA (mm) 5.5 3.0 1.2
C. Average particle size of SG (mm) 180 250 320
D. Weight of SA within UMA (g) 0.5 0.75 1.0
(sum of weight of SA and SG: 5 g)
E. Magnetic flux density (T) 0.25 0.5 0.75
F. Circumferential speed of workpiece (m/s) 0.2 0.4 0.6
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

G. Axial vibration frequency of workpiece (Hz), 2 5 8

(vibration amplitude: 5 mm)
H. Finishing time (min) 5 10 15

surface was produced on a cylindrical workpiece using a Sodick A30R EDM

machine. The cylindrical workpiece was vertically clamped on the rotating mandrel.
Feeding the copper electrode horizontally to the rotating workpiece, the peripheral
EDM machined surface was produced on the workpiece. The cross section of the
copper electrode was 8
2 mm2. The operation conditions of EDM process were
negative polarity, open voltage: 90 V, pulse on time: 100 ms, pulse off time: 100 ms,
gap voltage: 45 V, and peak current: 2–3 A. The rotating speed of the workpiece was
150 rpm during EDM. By applying the above conditions, the roughness of the EDM
machined surface was initially within 1.6–3.3 mm Ra, while the thickness of the recast
layer was about 10 mm. The finishing experiments were performed on the EDM
machined surface and the experimental conditions were extracted from combinations
of the L18 orthogonal array, within which 18 process parameter combinations are
available. Three experiments were conducted for each combination to reduce the
experimental errors. The reduction ratio, R, of the finished surface roughness (SR),
defined in Eq. (1), was calculated and used as an observation for each experiment
since the initial roughness of the EDM machined surfaces was not the same for all
the workpieces.

R ¼ ðInitial SR  Final SRÞ=Initial SR ð1Þ

The UMA used herein was composed of SG, SA, and SAE30 lubricant. Both the
particle sizes and the contents of SG and SA applied in UMA were determined from
Table 1 according to the experimental conditions of the L18 orthogonal array. The
sum of weight of SG and SA within UMA was 5 g, while the SAE30 lubricant was
exclusively 0.6 g. As well as lubricating, the SAE30 held the SA to the SG to prevent
SA from scattering into the air during finishing. After the mixture of SG and SA
with SAE30 lubricant was uniformly stirred, it was poured into the working gap.
Then, the finishing experiment was conducted. The abrasive slurry (working fluid),
consisting of 5% SA by weight, mixed with distilled water, was introduced during
finishing to improve the quality of the finished surface. The slurry was supplied at
 ORDER REPRINTS

Improving EDM Surfaces Using MAF 109

a flow rate of 2.4 mL min1 through an electric stirrer and a micro tube pump. In
addition to cooling and lubricating the workpiece, the slurry would also supplement
the SA.

4. RESULTS AND DISCUSSION

4.1. Taguchi Method

Table 2 presents experimental results concerning the EDM machined surfaces.

The signal-to-noise (S/N) ratio,
, is determined from the reduction ratios. In the
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

finishing process, a larger surface roughness reduction ratio corresponds to a higher

surface quality. Thus, the surface finish improvement is a higher-the-better (HB)
quality characteristic. The equation for the S/N ratio, with the HB quality
characteristic, is defined as:
" #
1X n
1

¼ 10 log ðdBÞ ð2Þ
n i¼1 y2i

where yi represents the observed reduction ratio of ith experiment and n is the
number of replications under each set of experimental conditions. n ¼ 3 since each
experiment is performed thrice.
Table 3 shows the average
for each level of the eight control factors, obtained
from the numerical values of
in Table 2. The average
are commonly called the
main effects of the factors at each level. A larger average
indicates that the factor at
that level contributes more to the improvement in the finish of the surface. A larger
average
variation implies that a factor is more significant related to the surface
finish improvement.
Figure 3 plots the average
listed in Table 3. The figure more clearly shows the
contribution of the main effects. The most significant factor is magnetic flux density,
E, followed by steel grit particle size, C, and finishing time, H. The optimum
combination of process parameters to achieve the best surface finish can be obtained
by selecting the level with the highest
for each control factor. The optimum
combination is clearly A1B2C3D1E3F3G2H3. The best levels of the leading three
significant factors are as follows:
E. Magnetic flux density: 0.75 T.
C. Average particle size of SG: 320 mm.
H. Finishing time: 15 min.

4.2. Analysis of Variance (ANOVA)

Analysis of Variance can be used to support the Taguchi method to identify the
significance of the control factors and their interactions by the decomposition of
variance. Table 4 is the ANOVA table for
. It reveals that the mean squares of
factors D, F, G, and the interaction, A
B, are so small that the variation in their
 ORDER REPRINTS

110 Yan et al.

Table 2. Experimental results and L18(21
37) orthogonal array.

Control factor SR, Ra (mm)

Expt. SR Red. ratio S/N ratio
no. A B C D E F G H Rep. Initial Final difference R
(dB)

1 1 1 1 1 1 1 1 1 a 2.436 2.024 0.412 0.1691

b 1.88 1.35 0.530 0.2819
c 2.056 1.662 0.394 0.1916 13.966
2 1 1 2 2 2 2 2 2 a 2.5 0.56 1.940 0.7760
b 1.53 0.38 1.150 0.7516
c 1.702 0.594 1.108 0.6510 2.856
3 1 1 3 3 3 3 3 3 a 2.59 0.1 2.490 0.9614
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

b 1.514 0.094 1.420 0.9379

c 1.782 0.116 1.666 0.9349 0.496
4 1 2 1 1 2 2 3 3 a 2.48 0.86 1.620 0.6532
b 1.57 0.53 1.040 0.6624
c 2.102 0.682 1.420 0.6755 3.563
5 1 2 2 2 3 3 1 1 a 1.868 0.34 1.528 0.8180
b 1.494 0.302 1.192 0.7979
c 1.808 0.248 1.560 0.8628 1.672
6 1 2 3 3 1 1 2 2 a 2.57 1.23 1.340 0.5214
b 1.42 0.67 0.750 0.5282
c 1.955 1.148 0.807 0.4128 6.412
7 1 3 1 2 1 3 2 3 a 2.19 1.702 0.488 0.2228
b 1.702 1.214 0.488 0.2867
c 1.958 1.37 0.588 0.3003 11.603
8 1 3 2 3 2 1 3 1 a 2.47 1.66 0.810 0.3279
b 1.958 1.14 0.818 0.4178
c 1.896 1.258 0.638 0.3365 9.005
9 1 3 3 1 3 2 1 2 a 2.54 0.58 1.960 0.7717
b 1.626 0.22 1.406 0.8647
c 1.826 0.262 1.564 0.8565 1.643
10 2 1 1 3 3 2 2 1 a 2.37 1.23 1.140 0.4810
b 1.938 0.83 1.108 0.5717
c 2.112 0.88 1.232 0.5833 5.366
11 2 1 2 1 1 3 3 2 a 2.07 1.356 0.714 0.3449
b 1.988 1.32 0.668 0.336
c 1.632 1.168 0.464 0.2843 9.947
12 2 1 3 2 2 1 1 3 a 2.44 0.72 1.720 0.7049
b 1.806 0.658 1.148 0.6357
c 1.944 0.426 1.518 0.7809 3.101
13 2 2 1 2 3 1 3 2 a 2.35 1.23 1.120 0.4766
b 1.442 0.718 0.724 0.5021
c 1.848 0.794 1.054 0.5703 5.814
14 2 2 2 3 1 2 1 3 a 1.942 1.462 0.480 0.2472
b 1.71 1.138 0.572 0.3345
c 1.896 1.424 0.472 0.2489 11.403

(continued )
 ORDER REPRINTS

Improving EDM Surfaces Using MAF 111

Table 2. Continued.

Control factor SR, Ra (mm)

Expt. SR Red. ratio S/N ratio
no. A B C D E F G H Rep. Initial Final difference R
(dB)

15 2 2 3 1 2 3 2 1 a 2.174 0.95 1.224 0.563

b 1.652 0.588 1.064 0.6441
c 1.692 0.566 1.126 0.6655 4.163
16 2 3 1 3 2 3 1 2 a 2.62 1.5 1.120 0.4275
b 1.96 1.044 0.916 0.4673
c 1.88 1.002 0.878 0.4670 6.883
17 2 3 2 1 3 1 2 3 a 3.33 0.91 2.420 0.7267
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

b 1.856 0.41 1.446 0.7791

c 1.826 0.66 1.166 0.6386 3.005
18 2 3 3 2 1 2 3 1 a 2.14 1.74 0.400 0.1869
b 2.074 1.65 0.424 0.2044
c 1.848 1.478 0.370 0.2002 14.122

Table 3. Average responses of factors.a

Average
by level (dB)

Factor 1 2 3

A 5.691 7.089
B 5.955 5.504 7.710
C 7.866 6.315 4.990
D 6.048 6.528 6.594
E 11.24 4.928 2.999
F 6.884 6.492 5.794
G 6.445 5.567 7.158
H 8.049 5.593 5.529
a
Overall mean
¼ 6.39 dB.

effects caused by changing their levels can be ignored. All 17 degrees of freedom were
used to evaluate the factorial effects since all eight columns in the L18 orthogonal
array were occupied by the eight factors. Therefore, the error term has zero degrees
of freedom, and the sum of squares due to the error is also zero, indicating that the
error mean square or error variance cannot be directly estimated. However, an
approximate estimate of the error variance can be obtained by pooling the values of
the aforementioned four small terms, which provide small variance. The sums of
squares and the degrees of freedom of the four terms are separately added as an
estimate of the pooled sum of squares and the pooled degrees of freedom,
respectively, which are shown in parentheses in Table 4. Consequently, the error
variance, which equals the error mean square, can then be estimated, and is called
the pooled error mean square (Phadke, 1989).
 ORDER REPRINTS

112 Yan et al.

−2
−3
−4
−5
S/N Ratio (dB)

−6
−7
−8
−9
−10
−11
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

−12
A1 A2 B1 B2 B3 C1 C2 C3 D1 D2 D3 E1 E2 E3 F1 F2 F3 G1 G2 G3 H1 H2 H3

Figure 3. Response plots of factor effects. Control factors and levels, defined in Table 1, are
on the abscissa. (View this art in color at www.dekker.com.)

Table 4. ANOVA table for S/N ratio.

Degrees of Sum of Mean square

Factor freedom squares (variance) F-value

A 1 8.803 8.803 3.514

B 2 16.298 8.149 3.253
C 2 24.870 12.435 4.964b
D 2 1.067a 0.533
E 2 223.05 111.52 44.52b
F 2 3.661a 1.830
G 2 7.614a 3.807
H 2 24.78 12.39 4.946b
A
B 2 7.70a 3.85
Error 0 0 —
Total 17 317.841
(Error) (8) (20.042) (2.505)
a
The four terms added together to form the pooled error sum of squares is shown in
parentheses.
b
Indicates the significant factors. F0.05 (1, 8) ¼ 5.318, F0.05 (2, 8) ¼ 4.459.

The F test, a significance test, is important in classical statistical experimental

design, and is commonly used to determine whether a control factor is significant.
The F value in Table 4 is the ratio of the mean square due to a factor and the pooled
error mean square, and they are calculated to represent the relative importance of the
various factors in relation to the error variance. The F ratio is an index for
identifying the significance of that factor. A higher F ratio means that the factor is
more important in influencing the process response,
. The significance of that factor
 ORDER REPRINTS

Improving EDM Surfaces Using MAF 113

can be determined by comparing the F ratio with the value in a statistical F

distribution table at a specific significance level. At a significance level of  ¼ 0.05,
Table 4 reveals that the significant factors are C, E, and H, of which the magnetic
flux density, E, is the most significant. This result agrees with that obtained using
the Taguchi method.

4.3. Predicting g under Optimum Conditions

After both the optimum combination of process parameters has been

determined and the significant factors identified,
can be predicted for optimum
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

conditions. The additive model, expressed in Eq. (3), can be used to predict the value
of
for surface finish improvement under optimum conditions, represented by
opt
(Phadke, 1989).
X

opt ¼ m þ ðmi  mÞ
¼ m þ ðmC3  mÞ þ ðmE3  mÞ þ ðmH3  mÞ ð3Þ
where m is the overall mean value of
, and mi is the optimum
of the significant
factor, i. The contributions of all but the significant factors are ignored to avoid
an over prediction of
opt. Substituting the values of Table 3 into Eq. (3), yields the
predicted value
opt ¼ 0.738 dB, which is equivalent to a surface roughness reduction
ratio of 0.9185.

4.4. Confirmatory Experiments

The confirmatory experiments are to verify the validity of the optimum process
combination. After the response under optimum conditions are predicted, some
experiments with the optimum combination of process parameters are conducted
and the observed value is compared with the prediction. If the observation differs
markedly from the prediction then the control factors strongly interact, and another
appropriate orthogonal array must be selected, or suitably constructed, to estimate
those interactions and the main effects.
In the present study, two confirmatory experiments under optimum conditions
were conducted. Table 5 shows the results. The observed surface roughness

Table 5. Results of confirmatory experiments.a

Surface roughness, Ra (mm)

SR Red.
Trial no. Initial Final difference ratio R

1 2.038 0.176 1.862 0.9136

2 1.850 0.164 1.686 0.9114
a
The predicted surface roughness reduction ratio, R, is 0.9185.
 ORDER REPRINTS

114 Yan et al.

reduction ratio, R, matches the predicted value to well within 1%. Accordingly, the
optimum process combination is credible and the optimum process parameters
obtained using the parametric design are verified.

4.5. Verifying Significance

Several experiments, based on optimum conditions, were conducted to verify

the significance of the factors. The SiC particle size and significant factors such
as magnetic flux density and steel grit particle size were investigated. One of the
experiments was performed under optimum conditions and the result was used as a
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

reference for comparison. Other experiments employed optimum combinations,

except that the optimum level of one selected factor was replaced by the level of the
lowest
. The level of the lowest
for magnetic flux density, steel grit particle size
and SiC abrasive particle size were 0.25 T, 180 and 1.2 mm, respectively.
Figure 4 shows the results of the four experiments. The effect of magnetic flux
density caused the greatest deviation of surface roughness from optimum conditions,
while the SiC particle size caused the least deviation. A larger deviation implied that
the factor was more significant. Hence, these experimental results reconfirm the
significance of the selected factors.
Concerning the workpiece with similar material to SKD11, it is no doubt that
the optimum combination of process conditions obtained in this study can be
applied to the cylindrical workpiece. Even though the workpieces are different in
diameter, they will have the same good results on surface finish. The optimum

2.0
1.8
Optimum
Surface Roughness Ra (µm)

1.6 SiC_1.2 ***

1.4 St. grit_180 **
1.2 0.25 Tesla *

1.0
0.8
0.6
0.4
0.2
0.0
0 3 6 9 12 15
Finishing Time (min)

Figure 4. Variation in the surface roughness with finishing time for each significant factor at
the level of the lowest
, in comparison with the optimum conditions. Note: *, **, and ***
represent the level of the lowest
for magnetic flux density, steel grit particle size and SiC
abrasive particle size, respectively.
 ORDER REPRINTS

Improving EDM Surfaces Using MAF 115

Downloaded by [Syracuse University Library] at 08:26 28 April 2013

Figure 5. Schematic showing the finishing of complex surface using a rotating magnetic pole
with slotted spherical head.

combination can also be employed to a multi-stepped cylindrical bar or a circular

part with barrel shape if the workpiece does not have large variation in diameter. A
satisfactory surface should be obtained after finishing for these types of workpiece
since the circumferential speed of workpiece is not a significant factor. However, the
optimum combination cannot be used to an internal finishing of circular pipe or a
complex shapes finishing such that presented in (Yugang and Shicheng, 2000),
which is shown in Fig. 5. Since their tooling configurations are different from that of
the external cylindrical finishing, there may be factors that have not been considered
in present study. Thus, a new determination of the optimum process conditions for
the specific configuration should be made. However, since the abrasion mechanisms
are all the same to all kinds of magnetic abrasive finishing during regular finishing,
the qualitative finishing characteristics of the process parameters obtained in
this study are still valid that can be used as a reference during the experimental
planning of the new determination. If the material of the workpiece is far different
from SKD11, the optimum combination does not necessarily yield the optimum
surface since the finishing results are varied with both the material structure
and the mechanical properties of the workpiece, even though it is a uniform
cylindrical bar.

4.6. Improving Surface Finish

Figure 6 shows SEM micrographs and surface roughness profiles of the EDM
machined surface before and after finishing. Figure 6(a) shows the craters and
 ORDER REPRINTS

116 Yan et al.

10 µm 10 µm 10 µm

µm µm µm
5 1 1
0 0 0
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

0 0.1 mm (a) 0 0.1 mm (b) 0 0.1 mm (c)

Figure 6. SEM micrographs and surface roughness profiles of the EDM machined surface
before and after finishing. (a) As received EDM surface (2.0 mm Ra), (b) after processing for
15 min under optimum conditions (0.176 mm Ra), (c) after processing for 30 min under
optimum conditions (0.04 mm Ra).

micro-cracks within the recast layer on the EDM machined surface before MAF.
The initial surface roughness is 2.0 mm Ra. However, after having been processed for
15 min using MAF under optimum conditions, Fig. 6(b) demonstrates clearly that
most of the work surface has been finished well, but a small crater remains, on which
micro-cracks still exist. Consequently, the surface roughness is only improved to
0.176 mm Ra. Figure 6(c) shows that both the craters and the micro-cracks are
completely removed after being processed for 30 min under optimum conditions,
yielding a refined surface with 0.04 mm Ra.

4.7. Eliminating the Recast Layer

Figure 7(a) shows a SEM micrograph of the recast layer, observed in cross-
section, on the EDM machined surface before MAF. The micro-cracks are clearly
observed within the recast layer, the variation in whose thickness is within 10 mm, by
measuring from the figure. This is the poor layer that always exists on a work surface
after EDM. However, MAF can be employed to eliminate such layers effectively,
especially on the curved work surface, due to the flexible finishing tool. Figure 7(b)
presents a SEM micrograph of the refined surface. The recast layer, together with
the micro-cracks, is removed completely after being processed for 30 min under
optimum conditions.
The elimination of the recast layer was alternatively demonstrated by measuring
the Micro-Vickers hardness on work surface. Before finishing, the Micro-Vickers
hardness of the recast layer was HV660–740 (HRC58.3–61.8), while after processing
for 30 min under optimum conditions, the hardness of the refined surface was
HV583–623 (HRC54.3–56.4), almost the same hardness as that of the original
surface before EDM. This result also demonstrates that the recast layer was
removed.
 ORDER REPRINTS

Improving EDM Surfaces Using MAF 117

Recast layer

10 µm 10 µm

(a) (b)
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

Figure 7. SEM micrographs of recast layer on the EDM machined surface before and after
finishing. (a) As received EDM surface (2.0 mm Ra), (b) after processing for 30 min under
optimum conditions (0.04 mm Ra).

5. CONCLUSIONS

1. The significance of control factors was identified using ANOVA and the
F test. The results were reconfirmed by experiments based on optimum
conditions.
2. Optimum combination of process conditions for improving the finish of
EDM machined surfaces was determined. The optimum combination was
working clearance: 1 mm, average particle size of SiC abrasive: 3.0 mm,
average particle size of steel grit: 320 mm, weight of SiC abrasive within
UMA: 0.5 g, magnetic flux density: 0.75 T, circumferential speed of
workpiece: 0.6 m/s, axial vibration frequency of workpiece: 5 Hz, and
finishing time: 15 min. Magnetic flux density was found to be the most
significant, followed by steel grit particle size and finishing time.
3. Two confirmatory experiments established that the observed surface rough-
ness reduction ratio matched the prediction under optimum conditions to
within 1%, indicating that the optimum process combination is reasonable.
4. Improved surface finish will not be obtained on EDM machined surfaces
unless the craters within the recast layer are removed first.
5. The recast layer including micro-cracks on EDM machined surfaces can be
completely removed by applying the MAF process for 30 min, and a finished
surface of 0.04 mm Ra can be obtained.

ACKNOWLEDGMENTS

The authors would like to thank the National Science Council of the Republic
of China for financially supporting this research under Contract No. NSC 89-2212-
E-008-050.
 ORDER REPRINTS

118 Yan et al.

REFERENCES

Anzai, M., Kawashima, E., Otaki, H., Nakagawa, T. (1993). Magnetic abrasive
finishing of WC-CO curved surfaces. In: Proceedings of International
Conference on Machining of Advanced Materials, Gaithersburg, MD, USA,
NIST Special Publication 1993, 847, 415–422.
Chang Geeng-Wei, Yan Biing-Hua, Hsu Rong-Tzong. (2002). Study on cylindrical
magnetic abrasive finishing using unbonded magnetic abrasives. International
Journal of Machine Tools & Manufacture 42:575–583.
Kim Jeong-Du, Choi, Min-Seog. (1995). Simulation for the prediction of surface-
accuracy in magnetic abrasive machining. Journal of Materials Processing
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

Technology 53(3–4):630–642.
Kremen, G. Z., Elsayed, E. A., Rafalovich, V. I. (1996). Mechanism of material
removal in the magnetic abrasive process and the accuracy of machining.
International Journal of Production Research 34(9):2629–2638.
Krymsky, M. D. (1993). Magnetic abrasive finishing. Metal Finishing 91(7):21–25.
Phadke Madhav, S. (1989). Quality Engineering Using Robust Design. Englewood
Cliffs, NJ 0763: Prentice-Hall.
Shinmura, T., Takazawa, K., Hatano, E., Matsunaga, M. (1990). Study on magnetic
abrasive finishing. Annals of the CIRP 39(1):325–328.
Shinmura, T., Yamaguchi, H. (1995). Study on a new internal finishing process by
the application of magnetic abrasive machining. JSME International Journal
Series C 38(4):798–804.
Yamaguchi, H., Shinmura, T. (1999). Study of the surface modification resulting
from an internal magnetic abrasive process. Wear 225–229, 246–255.
Yugang Zhao, Shicheng Jiang. (2000). New kind of machine tool for magnetic
abrasive finishing complex surface. Chinese Journal of Mechanical Engineering
36 (3):100–103. In Chinese.
 Request Permission or Order Reprints Instantly!

Interested in copying and sharing this article? In most cases, U.S. Copyright
Law requires that you get permission from the article’s rightsholder before
using copyrighted content.

All information and materials found in this article, including but not limited
to text, trademarks, patents, logos, graphics and images (the "Materials"), are
the copyrighted works and other forms of intellectual property of Marcel
Dekker, Inc., or its licensors. All rights not expressly granted are reserved.
Downloaded by [Syracuse University Library] at 08:26 28 April 2013

Get permission to lawfully reproduce and distribute the Materials or order

reprints quickly and painlessly. Simply click on the "Request Permission/
Order Reprints" link below and follow the instructions. Visit the
U.S. Copyright Office for information on Fair Use limitations of U.S.
copyright law. Please refer to The Association of American Publishers’
(AAP) website for guidelines on Fair Use in the Classroom.

The Materials are for your personal use only and cannot be reformatted,
reposted, resold or distributed by electronic means or otherwise without
permission from Marcel Dekker, Inc. Marcel Dekker, Inc. grants you the
limited right to display the Materials only on your personal computer or
personal wireless device, and to copy and download single copies of such
Materials provided that any copyright, trademark or other notice appearing
on such Materials is also retained by, displayed, copied or downloaded as
part of the Materials and is not removed or obscured, and provided you do
not edit, modify, alter or enhance the Materials. Please refer to our Website
User Agreement for more details.

Request Permission/Order Reprints

Reprints of this article can also be ordered at

http://www.dekker.com/servlet/product/DOI/101081MST120034246