INTRODUCTION
Nickel based superalloy Inconel 718 finds extensive applications in missile,
aerospace, gas turbines, chemical, petrochemical and nuclear plants due to their
competitive characteristics such as high specific strength, corrosion resistance,
wear resistance and high temperature properties compared with other alloys.
Despite of these good mechanical properties, they have serious issues of poor
machinability and hence regarded as a difficulttocut material (Pawade
et al., 2007). There will be a greater demand to establish process
parameters that produce surfaces with high precision and reliability longevity
in product performance. Surface grinding is extensively used operation in finishing
of flat shaped components. Being a secondary and last in the manufacturing cycle,
a care is vital while selecting the process parameters to induce minimum damage
to the surfaces. It was reported that the higher grinding speed and the temperature
gradients developed during grinding operation could lead to thermal damage on
the work surfaces. A number of researchers have described relationships between
process factors affecting the performance of the grinding fluid. Some of the
relevant findings are as follows.
Osterle and Li (1997) conducted experimental investigation
on grinding of superalloy IN738. They reported that the local melting at contact
spot leading to formation of white layers. It is found that the grinding wheel
characteristics have a significant effect on surface topography of the machined
surfaces. JaeSeob (2004) studied the effect of grinding
parameters on geometric error using Taguchi and response surface methodologies.
They concluded that the depth of cut was dominant parameter influencing geometric
error. HwaSoo (1996) found that the higher surface
speed of grinding wheel results in lower grinding forces when CBN (Christian
Broadcasting Network) wheel was used. Further higher peripheral speed of CBN
wheel results in lower grinding forces. Xu et al.
(2002) conducted experimental investigation in surface grinding of a cast
nickel based superalloy with alumina abrasives wheels to assess the surface
integrity. Shaji and Radhakrishnan (2003) conducted
experimental investigation on effect of coolant on the workpiece quality in
surface grinding process. They reported that the application of graphite reduces
the heat generated in grinding zone and thus found that the tangential force
and normal force is higher as compared to those in conventional grinding. Di
Ilio and Paoletti (1999) found that the decrease in grindability is mainly
caused by clogging of the active surface on the wheel due to chip adhesion rather
than by flattening of grits caused by the abrasion of hard reinforcement. It
is reported that the alumina/SiC grinding wheels produce lower grinding forces
and better surface finish than those with super abrasive wheels. Venugopal
and Rao (2005) reported an improvement in surface finish under the graphite
assisted grinding. Besides, a considerable reduction of tangential force and
specific grinding energy causes reduction in surface damages. Yin
et al. (2005) found that the grinding at conventional as well as
high speed do not have much difference on the grinding characteristics and surface
damages. Both normal and tangential grinding force ratios at high speed were
lower than those at the conventional speed. At higher grinding speed, due to
increased metal removal rate, the normal forces are 46 times higher than the
pure normal grinding forces. Kwak et al. (2007)
optimized the grinding parameters in surface grinding of MMC using surface roughness
and grinding force with Taguchi and response surface methods. It is observed
during grinding of MMC that the normal grinding force changes very rapidly with
the increase in the number of passes due to the wheel clogging. The tangential
and normal forces found to be increased moderately with an increase in the depth
of cut. It is reported that the surface roughness increases with an increase
in the depth of cut and work speed during surface grinding Di
Ilio et al. (1996).
As far as the superalloy Inconel 718 is concerned, very few studies have been
reported till date. These include investigation of grinding mechanism and surface
integrity. Manufacturers of nickelbased and cobaltbased turbine engine components
often use plated CBN wheels in various grinding steps due to their capability
of supporting high removal rates and their ability to hold consistent form without
the need for inline dressing steps (Harpster, 1991).
However, they are costly and thus not commercialized in the shop floors. PeiLum
(1994) studied the grinding of Inconel 718 in which he examined the surface
finish, grinding force, wheel wear using different grinding wheels; He found
that the CBN grinding wheel gives better performance as compared to WA and GC
grinding wheels.
In the view of the above, the study reports the analysis of the grinding parameters
on the surface roughness and the grinding force and determining the optimal
sets of grinding parameters to be able to improve the grinding performance.
MATERIALS AND METHODS
Taguchi design of experiments: The Taguchi method is a traditional approach
for robust experimental design that seeks to obtain a best parametric combination
and the levels with the lowest cost and higher quality to achieve customer satisfaction.
In the Taguchi design, the factors that can be controlled by designers called
as control parameters and noise factors (factors that cannot be controlled by
designers, such as environmental factors) are considered to be influential on
process output. Therefore, the Taguchi design selects the levels of process
parameters and reduces the effects of noise factors. That is parameter setting
should be determined with the intention that the product response (quality characteristic)
has minimum variation, while its mean is close to the desired target (Box
GEP, 1985). Taguchi recommends the following stepbystep procedure for
designing the experiments and analysis:
• 
Determining the quality characteristics (response variables)
to be optimized 
• 
Identifying the noise factors and test conditions 
• 
Identifying the control factors and their alternative levels 
• 
Selecting orthogonal matrix 
• 
Selecting performance characteristics, S/N ratio 
• 
Conducting the matrix experiment 
• 
Analysing the data and determine optimum levels for control factors 
• 
Predicting the performance at these levels 
Determining the quality characteristics to be optimized: In this study,
two types of response variables were used. These are: (i) inprocess variables
and (ii) postprocess variables.
Grinding forces: Cutting forces in grinding influence surface and subsurface
quality. Therefore, magnitudes of the cutting force components during grinding,
viz. TangentialF_{t} and normalF_{n} were selected as in process
response variables.
Surface roughness: It is a geometrical feature of the machined surface
that significantly influences the machined surface integrity. Therefore, the
arithmetic average (R_{a}) (centerline average of peak to valley) surface
roughness was selected as the post process response variables.
Identifying the control factors and their levels: Taguchi classified
the factors in an experiment as control factors and noise factors. The factors
which can be controlled by the designer or operator, are called as control factors.
On the other hand, the noise factors cannot be controlled in the actual process
(Phadke, 1989). The noise factors can be humidity, dust,
temperature of the environment, material property variation, machine vibration,
etc. In surface grinding, process, wheel and work material related parameters
influence the quality of the ground surfaces. As far as the scope of this investigation
is concerned, a few of them have been selected with consideration to constraints
on availability of machine, grinding wheel and the work material.
Table 1: 
Selection of control factors and their levels 

Table 2: 
Standard L_{27 }(3^{13}) orthogonal matrix 

The machining parameters were selected on the basis of the knowledge available
in the literature and the past experience. It is believed that the table speed,
infeed, grit size and the type of lubricant employed during grinding influences
the forces and the mechanism of surface generation. As a result, it alters the
nature of machining deformation and consequently nature of the surface generated.
Therefore, the three parameters that have been chosen as control factors are:
table speed, infeed, grit size and the type of lubricant (Table
1). Further, three twofactor interactions such aslubricant and grit size,
lubricant and table speed and grit size and table speed were chosen. The higher
order interactions are generally not significant in engineering applications
and hence were neglected (Ross, 1996).
Selecting orthogonal matrix: Taguchi’s orthogonal matrix provides
an alternative to standard factorial designs (Phadke, 1989).
In the present study, four grinding parameters were considered as independent
variables: table speed, infeed, grit size and the type of lubricant.
Table 3: 
Experimental matrix with assigned parameters and response
values 

A: Lubricant, B: Grit size, C: Table speed, D: Infeed, F_{t}:
Tangential force, F_{n}: Normal force, R_{a}: Surface roughness 
Each of them has three different levels. Further, some interactions between
them were selected. Thus total 20° of freedom are needed to calculate main
effects of the factors and their interactions. Therefore, an L_{27}
orthogonal matrix was selected (Table 2) because it has 26°
of freedom which is greater than the required 20° of freedom. The experimental
design matrix with assigned parameters is shown in Table 3.
Selecting performance characteristics: In Taguchi’s methodology
it is the “signaltonoise” ratio which is used as a measure of performance.
Here, the term ‘signal’ represents the desirable value (mean) and
the ‘noise’ represents the undesirable value standard deviation).
Thus, the S/N ratio represents the amount of variation present in the performance
characteristic. As every process performance characteristic would have a target
or nominal value. The robust design reduces the variability around this target
or nominal value and models the departures from the target value as a loss function.
Depending upon the objective of the performance characteristic, there are three
types of S/N ratios.

Fig. 1: 
Factors affecting grinding performance in Inconel 718 
Table 4: 
Chemical composition of received Inconel 718 

Here, the desirable objectives are to minimize the values of tangential and
normal forces. Hence, the LowertheBetter (LB) type S/N ratio, as defined below
was applied for transforming the raw data (Ross, 1996).
where, y_{i} is the value of the characteristic in an observation i
and n is the number of observations or number of repetitions in a trial.
CONDUCTING THE EXPERIMENTS
Work material, grinding wheel and equipment: The work material used
for the investigation was superalloy Inconel 718. The chemical composition of
work material is shown in Table 4. These are prepared as a
square shaped specimen with dimensions 20x20x5 mm. Annealing of the work specimen
was carried out at temperature 9951065°C over 3060 min duration. It was
performed in a salt bath furnace. After annealing the work specimen were cooled
in the air. The purpose of annealing was to relieve induced stresses from previous
machining operation used for sample preparation to the required size. The Al_{2}O_{3}
grinding wheel was used for the grinding tests. Factors affecting grinding performance
in Inconel 718 are shown in Fig. 1.

Fig. 2: 
Schematic of experimental set up and photograph of surface
grinding 
Experimental procedure: A photograph of detailed set up and close view
of grinding operation is shown in Fig. 2. A PRAGA make 451
AP model surface grinder was used for the experimental work. The specimens are
clamped to the fixture on grinding chuck. A kistler dynamometer model 9257A
was fitted on grinding table for measurement of grinding force. A suitable arrangement
was made to supply the coconut oil and SAE+graphite to the grinding zone. Liquid
nitrogen was supplied to the grinding wheel and work piece interface from the
cryocan. The grinding wheel was mounted on machine spindle and properly balanced.
Machine is adjusted to required stroke and desired table speed is set for the
experiment. Dial probe was touched on work piece and desired depth of cut is
set. As per the design matrix 27 experiments were performed. The grinding wheels
were dressed with a commercial single point diamond dresser after every two
tests to keep the cutting edges sharp during grinding.
The grinding force magnitude and pattern were acquired through cutting force
dynamometer. Surface roughness of the ground specimen along crossfeed direction
was measured at three locations on the specimen’s surface using a portable
surface roughness tester (MakeMitutoyo, ModelSurftest SJ301). The cutoff
and sampling length for each measurement were kept as 0.8 and 4.8 mm, respectively.
Surface topography of the ground specimens was examined using scanning electron
microscopy (MakeJEOLJSM) at various magnifications.
EXPERIMENTAL RESULTS AND DISCUSSION
Statistical Analysis of Surface Roughness R_{a}: It is observed
from Fig. 3 that the lubricant used and the grit size of grinding
wheel have shown linear trend, however infeed and table speed show nonlinear
effect on the surface roughness. Further, it was seen that the grit size of
grinding wheel used for experiments in this study has larger effect on the surface
roughness compared to the other parameters. It is seen that the use of coconut
oil leads to efficient cooling effects and thus suppressed the effect of grinding
temperature on the work surface.
Therefore, the surfaces show lower values of the surface roughness. However,
the highest surface roughness is observed when liquid nitrogen coolant was used
while grinding. In the case of effect of grit size, the higher grit size results
into lower surface roughness of the ground specimen. On the contrary, the smaller
grit size wheels produced the higher surface roughness during grinding. It is
found that the smaller chip thickness produced on account of smaller grain size
of the grinding wheel abrasives. This will lead to generation of smaller peak
to valley height during grinding. Hence higher grit wheels produce lower surface
roughness.
The percentage contribution of various input factors on the selected performance
characteristic can be estimated by performing ANOVA. The total variation in
the result is the sum of variation due to various controlled factors and their
interactions and variation due to experimental error. The ANOVA for S/N ratio
have been performed to identify the significant parameters and to quantify their
effect on the performance characteristic.
It is observed from analysis of variance (Table 5) that the
grit size, followed by the infeed and the interaction between lubricant type
and grit size show statistical significance at 95% confidence level on the surface
roughness. The higher value of the coefficient of correlation, R^{2}
= 99.3 shows that the model developed is adequate to fit actual values with
the predicted values of the surface roughness.
Figure 4 and 5 shows the normal probability
plots of residual. A residual is the difference between the actual and predicted
values of response variable.

Fig. 4: 
Normal probability plot for R_{a }across the table
feed 

Fig. 5: 
Main effects plot of tangential grinding force 
It is observed that almost all the residuals lies along the straight line.
It indicates the best fit of the model and is adequate.
Statistical analysis of tangential grinding force: It is seen that the
parameter grit size of grinding wheel has linear trend, however table speed,
type of lubricant and infeed have nonlinear trend on the tangential force. It
is seen that the use of coconut oil leads to efficient cooling effects due higher
heat removal capacity and thus suppressed the effect of grinding temperature
on the work surface.
Table 5: 
Analysis of variance for means (R_{a} Across) 

S = 0.07157, R^{2} = 99.3%, R^{2} (adj) =
96.9% , *Significant at 95% confidence level, SS: Sum of squares, SS’:
Adjusted sum of squares, DF: Degrees of freedom, V: Variance, A: Lubricant,
B: Grit size, C: Table speed, D: Infeed 
Table 6: 
Analysis of variance for tangential force (F_{x}) 

S = 12.10, R^{2} = 79.0% R^{2} (adj) = 8.8%,
*Significant at 95% confidence level, SS: Sum of squares, SS’: Adjusted
sum of squares, DF: Degrees of freedom, V: Variance, A: Lubricant, B: Grit
size, C: Table speed, D: Infeed 
This in turn facilitates the deformation and thus the lower magnitude of forces
is required for removal of material during grinding. However, the highest forces
are observed when SAE+graphite were used as a coolant while grinding. In the
case of grit size effect, the higher grit size results into lower forces. On
the contrary, the smaller grit size wheels produced the higher magnitude of
forces during grinding. It is found that the area of metal removed per unit
chip thickness would be higher, thus higher forces are required for removal
of material. It is seen that the forces generated during grinding were of lower
magnitude when the table speed was at its highest level of 12 m min^{1}.
However, the grinding forces are of larger magnitude when the table speed was
10 m min^{1}. A change of table speed from 10 to 12 m min^{1}
resulted into drastic reduction in the grinding forces. It is observed that
the increase in infeed from 0.05 to 0.10 mm reduces the magnitude of grinding
force by a significant amount. Further increase in the infeed to 0.15 mm causes
an increase in the magnitude of grinding force. This increasing trend of force
is due to increase in the deformation area involved during the grinding process.
It is observed from analysis of variance (Table 6) that the
interaction between grit size and table speed (BC) shows statistical significance
at 95% confidence level on the grinding force. The value of the coefficient
of correlation, R^{2} = 79% shows that the model developed is adequate
to fit actual values with the predicted values of the grinding force.

Fig. 6: 
Main effects plot of maximum normal grinding force 
It is noted that the interaction BC has the largest share of 68.51% on the
variability of grinding force magnitude. Among the independent factors, only
infeed has shown maximum share on the grinding force variability determined
using ANOVA.
Statistical analysis of normal grinding force : Effects of grinding
process parameters on the normal grinding force are discussed using main effects
plots as shown in Fig. 6. It is observed that all the selected
grinding parameters show linear trend on the force produced during grinding.
It is seen that the use of coconut oil is more effective in reducing the heat
generated during grinding. Therefore, the lower magnitude of forces is required
for removal of material during grinding when coconut oil is used. However, significantly
higher forces are observed when liquid nitrogen was used as a coolant while
grinding. It is observed that the forces produce at 60 grit size wheel is smaller.
But increase in the grit size to either 120 or 220 causes increase in the magnitude
of forces. It is found that the area of metal removed per unit chip thickness
would be less, thus lower forces are required for removal of material for lower
grit grinding wheel. Table speed has significant effect on the generation of
grinding forces. It is seen that the forces generated during grinding were of
lower magnitude when the table speed was at its highest level of 12 m min^{1}.
higher grinding forces are seen when the table speed was at its lowest level
of 8 m min^{1}. It is observed that the increase in infeed from 0.10
to 0.15 mm decreases the magnitude of grinding force by a significant amount.
However, a negligible change in the force magnitude is observed when the infeed
increases from 0.05 to 0.10 mm.
It is observed from analysis of variance (Table 7) that none
of the main factors and their interaction show statistical significance on grinding
force at 95% confidence level.
Table 7: 
Analysis of variance for normal grinding force (F_{z}) 

S = 0.1964, R^{2} = 82.1%, R^{2} (adj) = 22.5%,
*Significant at 95% confidence level, SS: Sum of squares, SS’: Adjusted
sum of squares, DF: Degrees of freedom, V: Variance, A: Lubricant, B: Grit
size, C: Table speed, D: Infeed 
Table 8: 
Optimal sets of grinding parameters to improve grinding performance 

The value of the coefficient of correlation, R^{2} = 82.1% shows that
the model developed is adequate to fit actual values with the predicted values
of the grinding force.
OPTIMIZED GRINDING PARAMETERS
The grinding performance of the superalloy Inconel 718 was evaluated during
surface grinding process. The optimal sets to improve the surface roughness
and to reduce the grinding force were determined and are presented below. It
may be noted that except table speed, all grinding parameters having the same
value were applicable to achieve good surface finish and to reduce the tangential
grinding force (Table 8). It was very difficult for the same
grit size to satisfy both the good surface finish and the lower value of tangential
grinding force. The medium value of the table speed is better for the surface
finish but worst for the tangential grinding force. Further, the parameter levels
that produce the smaller normal grinding force are different than the parameters
satisfying the lower surface roughness. Hence, it is necessary to select the
grinding parameters sacrificing one of them. From the above results, it follows
that, if the criterion for selection of coolant for the smaller magnitude of
forces and good surface finish, the best choice will be the use of coconut oil
as a coolant during grinding.
SURFACE TOPOGRAPHY OBSERVATIONS
Surface topography reveals the surface features of an object or "how it looks",
its texture; detectable features limited to a few micrometers. Surface topography
of few ground samples was examined using Scanning Electron Microscope (SEM).
Scanning electron micrographs of these samples are presented in Fig.
7.

Fig. 7: 
SEM micrographs of ground surfaces 
The surface presents discrete streaks of grinding lay (grinding marks) The
surface texture also presents different flaws such as abrasion marks, smeared
layers and microparticle deposits. Smeared layers seen on the ground surfaces
are the indication of higher grinding temperature and thus show the intensity
of plastic deformation. Microparticle deposits are shown spread over the surfaces
are due to localized pull out of the surface. The number of microparticles seen
on the surface is more due to increase in the infeed. Abrasion marks are the
result of interaction of abrasive cutting point with the work material. In the
case of surfaces produced when SAE+graphite coolant was used, the microparticles
and the material flow was less due to presence of graphite.
CONCLUSIONS
Surface grinding of Inconel 718 reported in this article is an important finishing
process for machining of precision components. Within the range of test parameters
used, the following conclusions can be drawn:
• 
The grit size of grinding wheel shows dominant effect on surface
roughness Ra measured across the grinding direction 
• 
In the case of lubricant, the coconut oil is more effective in reducing
the surface roughness/ 
• 
The grit size, infeed and interaction between lubricant type and grit
size shows statistical significant on surface roughness at 95°/CI 
• 
In the case of grinding forces use of coconut oil leads to efficient cooling
effects and lower tangential and normal forces are produced 
• 
Table speed shows dominant effect in reduction of grinding forces 
• 
The ground surfaces show the alterations such as abrasion marks, micro
particle deposits and smeared layers 
ACKNOWLEDGMENT
The authors would like to express sincere thanks to Department of MHRD, Government
of India for the funds provided under TEQIP scheme for carrying out experimental
work on the precision machining facility set up at the university.