M.Janardhan et al.

/ International Journal of Engineering Science and Technology (IJEST)

DETERMINATION AND
OPTIMIZATION OF CYLINDRICAL
GRINDING PROCESS PARAMETERS
USING TAGUCHI METHOD AND
REGRESSION ANALYSIS
M.Janardhan1 and Dr.A.Gopala Krishna2
1

Professor, AbdulKalam Institute of Tech. Sciences, Vepalagadda, Kothagudem, Khammam Dist.AP., India
2

Associate Professor, Department of Mechanical Engg, JNTU College of Engg, Kakinada.AP., India
E mail:jc_muppala@yahoomail.com

Mobile:9849497834

ABSTRACT
Cylindrical grinding is one of the important metal cutting processes used extensively in the finishing operations.
Metal removal rate and surface finish are the important out put responses in the production with respect to
quantity and quality respectively. The Experiments are conducted on CNC cylindrical grinding machine with
L9 Orthogonal array with input machining variables as work speed, feed rate and depth of cut. Empirical
models are developed using design of experiments and response surface methodology. The adequacy of the
developed model is tested with ANNOVA. The developed model can be used by the different manufacturing firms
to select right combination of machining parameters to achieve an optimal metal removal rate (MRR) and
surface roughness (Ra).The results reveals that feed rate, depth of cut are influences predominantly on
the output responses metal removal rate (MRR) and surface roughness (Ra).The predicted optimal values
for MRR, Ra for Cylindrical grinding process are 62.05 gm/min and 0.816 m respectively. The results are
further confirmed by conducting confirmation experiments.

Key Words:

Cylindrical grinding, Metal removal rate (MRR), Surface roughness (Ra),

Design of experiments (DOE), Taguchi method and regression analysis

I Introduction
Cylindrical grinding is an essential process for final machining of components requiring smooth surfaces and
precise tolerances. As compared with other machining processes, grindings costly operation that should be
utilized under optimal conditions. Although widely used in industry, grinding remains perhaps the least
understood of all machining processes. The major operating input parameters that influence the output
responses, metal removal rate, surface roughness, surface damage, and tool wear, etc. are (i) wheel parameters:
abrasives, grain size, grade, structure, binder, shape and dimension, etc. (ii) Work piece parameters: fracture
mode, mechanical properties and chemical composition, etc. (iii) Process parameters: work speed, depth of cut,
feed rate, dressing condition, etc. (IV) machine parameters: static and dynamic Characteristics, spindle system,
and table system, etc. The present paper takes the following input processes parameters namely Work speed,
feed rate and depth of cut. The main objective of this paper is to show how our knowledge on grinding process
can be utilized to predict the grinding behavior and achieve optimal operating processes parameters. The
knowledge is mainly in the form of physical and empirical models which describe various aspects of grinding
process. A software package has been developed which integrates these various models to simulate what
happens during cylindrical grinding processes. Predictions from this simulation are further analyzed by
calibration with actual data. It involves several variables such as depth of cut, work speed, feed rate, grit size,
type of abrasive, chemical composition of wheel, etc. The main objective in any machining process is to

ISSN : 0975-5462

Vol. 3 No. 7 July 2011

5659

M.Janardhan et al. / International Journal of Engineering Science and Technology (IJEST)

maximize the metal removal rate (MRR) and to minimize the surface roughness (Ra).In order to optimize these
values taguchi method, ANNOVA and regression analysis is used.
2. Taguchi Method and Regression analysis
Taguchi method is the process of engineering optimization in a three step approach namely system design,
parameter design and tolerance design. In the system design, a basic functional prototype design will be
produced by applying scientific and engineering knowledge. In parameter design, independent process
parameter values will be optimized and where as in tolerance design, tolerances will be determined and analyzed
for optimal values set by parameter design. Taguchi method is a powerful design of experiments (DOE) tool for
optimization of engineering processes, in which the concept of S/N ratio is used for the improvement of quality
through variability reduction and improvement of measurement. The transformation of the repetition data to
another value, which is a measure of the variation and at least two data points are required to determine the S/N
ratio. There are several type of S/N ratios are available based on the characteristic namely lower is
better(LB),nominal is the best(NB),higher is better(HB) and mathematical equations of these S/N ratios is
expressed in equations (1),(2)and (3).The optimal level of the process parameters is the level with highest S/N
ratio. A statistical analysis of variance (ANNOVA) is performed to determine the significant process
parameters.
Lower is better S/N =-10 log [1/n (yi2)] ---- Eq. (1).
Higher is better S/N =-10 log [1/n (1/yi2)] ---- Eq. (2).
Nominal is better S/N =10 log [y/Sy2] ---- Eq. (3).
Where n is the number of observations, y is the observed data, Sy2 is the variance of y.
The orthogonal array is selected for the process parameters based on the Taguchi method the total degrees of
freedom for three levels with three process parameters were six. The L9 Orthogonal array with nine
experimental readings and three levels was selected for the Present work. The process parameters and their
levels in this study are presented in the Table 1.
Regression analysis is used to determine the relationships of the variables which influences the output responses
.The multiple regression analysis equations are in the following form
Y=a1+b1x1+b2x2+b3x3+----------+bkxk -------Eq. (4).
Where Y is independent variable and x1, x2, x3----xk are the known variables from which independent variable
will be determined .a1, b1, b2, b3----bk are regression coefficients which can be determined by the method of least
squares. The regression analysis was done by MINITAB15 statistical software.
Parameter

Table 1 Processes Parameters and their levels

Units
Symbol
Low
Center

Work Speed

Rpm

150

200

250

Depth of Cut

20

25

30

Feed rate

mm/min

780

1460

2140

High

3. Experimental Details
A set of experiments were conducted on cylindrical grinding machine on EN8 material to determine effect of
machining parameters namely feed rate (mm/min),Work speed (rpm), depth of cut (m) on metal removal rate
and surface finish (m). Three levels and three factors L9 Orthogonal array used to design the orthogonal array
by using design of experiments (DOE) and relevant ranges of parameters as shown in Table 1. Grinding wheel
used for the present work is the aluminum oxide abrasive with verified bond with constant speed of 1650 RPM
and water miscible coolant was supplied in all grinding experiments. The jobs have undergone turning and
hardening processes before grinding. Hardening is done for better output response and the BHN is maintained at
30-35. Material removal rate (MRR) is calculated by taking the ratio of difference of initial weights and final
weights to machining time. The formula to calculate the MRR is given in Eq. (5)
MRR= [(W1-W2)/machining time] gm/min -------- Eq. (5).
Where W1 is the weight of the job before grinding and W2 is the weight of the job after grinding .The surface
roughness (Ra) of the jobs is evaluated on the Taly surf surface test instrument. The average of ten readings is
taken to determine the reading of surface roughness value for the experiment.

ISSN : 0975-5462

Vol. 3 No. 7 July 2011

5660

M.Janardhan et al. / International Journal of Engineering Science and Technology (IJEST)

4. Results and discussion

Discussion was done by determination of signal to noise ratio(S/N ratio) based on the experimental data. The
S/N ratio is a simply a quality indicator by which the effect of a changing a particular process parameter on the
processes performance. A better signal is obtained when the noise is smaller so that a larger S/N ratio is used for
surface roughness and higher is better characteristic is used for metal removal rate. The surface roughness and
metal removal rate(MRR) for nine trail conditions with two measurement locations along with average response
values and corresponding S/N ratios were determined, shown in Table (2) and Table(3) respectively. The
average values of surface roughness and metal removal rate (MRR) at different levels were computed as shown
in Table (4). As a sample calculation, the average effect of work speed at first level was determined for surface
roughness using the average of surface roughness from the experimental values 1-3 of table (2) viz [
(1.34+0.98+0.879)/3=1.067]. Similarly the average values of surface roughness and metal removal rate (MRR)
for different processes parameters namely work speed, feed rate and depth of cut at different levels were
determined and shown in table (4). In the same procedure, the average values of S/N ratios for surface
roughness and metal removal rate (MRR) for different processes parameters namely work speed, feed rate and
depth of cut at different levels were determined and shown in table (5). Pareto ANNOVA was carried to
determine contribution ratio of the processes parameters on the output responses namely surface roughness and
metal removal rate (MRR) and shown in table (6) .The first Three rows indicate the sum of surface roughness
and metal removal rate (MRR) at respective levels for each process parameter. As a sample calculation, for
process parameter work speed at level 1 is determined as (1.34+0.98+0.879=3.2) from the experimental readings
of surface roughness and metal removal rate (MRR) in table (2) and Table (3) respectively and the same
procedure was extended to other sums at different levels for different process parameters . The Sum of squares
of differences for each process parameter was calculated using the equation Eq. (6).The same procedure was
adopted to determine other sum of squares of differences for each process parameter. The percentage of
contribution ratio was determined for the processes parameters on out put responses .As a sample calculation for
the process parameter of feed rate is [(1.283)/(1.283+0.525+12.7x10-4)]= 70.93 .PARETO diagram for surface
roughness and metal removal rate (MRR)is shown in Fig(1) and Fig(2) respectively.

From the diagram it is evident that the feed rate played vital role on output responses than other process
parameters. The overall optimal condition for the three process parameters were determined based on the least
sum at factor levels for the surface roughness and highest sum at factor Levels for the metal removal rate
(MRR).Hence the overall optimal condition for the surface roughness is A2 B3 C3 and for metal removal rate
(MRR) is A1 B3 C3
A= (A1-A2)2+ (A2-A3)2+ (A3-A1)2 -------- Eq. (6).
Analysis of variance (ANNOVA) using MINITAB15 software was performed to determine the significance of
process parameters on the output responses namely surface roughness and metal removal rate (MRR) listed in
table (7) and table (8) and revels that both models are significant. From table (7) and table (8), it is evident that
feed rate and depth of cut are significant process parameters as their F values are grater than P values for both
the output responses where as other process parameter work speed is not significant as its F value is less than the
P value. The regression equations for output responses in terms of process parameters is given below
Surface Roughness,
Ra =1.062(3.83x10-3 x A)(0.1478xC) (0.098xB) -------- Eq. (7).
Metal Removal Rate,

ISSN : 0975-5462

Vol. 3 No. 7 July 2011

5661

M.Janardhan et al. / International Journal of Engineering Science and Technology (IJEST)

MRR=44.90(1.10x A) + (7.62xB) + (10.763xC) ---------Eq. (8).

Results from Multiple Regression analysis along with experimental values are listed in the table (9) and it is
evident that there is a good agreement between results.

Table 2 Experimental observations and S/N ratio for surface roughness Ra (m)
Trail no

(Raij j=1,2(m)
1

2
1.38

Average response
Value Ra(m)

S/N Ratio (dB)

1.34

-2.545

150

20

780

1.3

150

25

1460

0.98

0.988

0.98

0.14

150

30

2140

0.884

0.874

0.879

0.12

200

25

780

1.132

1.2

1.256

-1.988

200

30

1460

0.918

0.902

0.91

0.818

200

20

2140

1.04

0.98

1.01

-0.09

250

30

780

1.14

1.08

1.11

0.909

250

20

1460

1.12

1.16

1.14

-1.139

250

25

2140

0.92

0.94

0.93

-0.629

Table3 Experimental observations and S/N ratio for Metal Removal Rate MRR (gm/min)
Trail no

MRRij
j=1,2(m)
1
2

Average response
Value MRR(gm/min)

S/N Ratio (dB)

150

20

780

27.01

28.31

27.66

34.86

150

25

1460

47.22

47.46

47.34

39.52

150

30

2140

64.18

62.06

63.12

42.02

200

25

780

32.6

32.8

32.7

36.31

200

30

1460

56.4

56.1

56.25

41.02

200

20

2140

44.96

46.16

45.56

39.19

250

30

780

38.22

38.34

38.28

37.68

250

20

1460

38.28

39.12

38.7

37.77

250

25

2140

54.13

54.95

54.54

40.75

Table4 Average values of MRR and Surface Roughness at different levels.

Parameter
L1

Average values of MRR

L2

L3

46.04

44.83

43.84

1.067

37.3

44.86

52.55

1.163

1.011

0.93

32.88

47.43

54.4

1.23

1.011

0.939

Average values of surface roughness

L1
L2
L3
1.058

1.06

L1 L2 L3 represents levels 1, 2 and 3 respectively.

Table 5 Average values of S/N ratios for MRR and Surface Roughness at different levels.
Parameter

Average S/N ratio of MRR

L1
L2
L3

Average S/N ratio of surface roughness

L1
L2
L3

38.8

38.84

38.73

-0.42

-0.42

-0.286

37.27

40.43

40.24

-0.955

-0.229

0.622

36.28

39.43

40.65

-0.44

-0.06

0.133

Table 6 determination of Pareto ANNOVA for MMR and Surface Roughness

Surface roughness
Process parameters

MRR
C

Sum of Factor Levels

ISSN : 0975-5462

Vol. 3 No. 7 July 2011

5662

M.Janardhan et al. / International Journal of Engineering Science and Technology (IJEST)

3.2

3.49

3.7

138.12

111.92

98.64

3.176

3.17

3.03

134.51

134.58

142.29

3
Total of each column of Parameter

3.18

2.899
9.559

2.819

131.5

157.65
404.15

163.22

Sum of squares of Differences

12.7x10-4

0.525

1.283

65.53

3136.9

6513.9

Contribution ratio %

0.07

29

70.93

0.67

32.28

67.05

Overall optimum condition

A2B3C3

A1B3C3

Table 7 ANOVA analyses for MRR

Source

Sum of Squares

DF

Mean Square

F Value

p-value

Remarks

Model

1050.92

350.31

46.69

0.0004

Significant
Not Significant

7.26

7.26

0.97

0.3705

348.48

348.48

46.44

0.001

Significant

695.18

695.18

92.65

0.0002

Significant

Residual

37.52

7.5

Cor-Total

1088.44

S.D

2.74

R2

0.9655

Mean

44.91

Adj.R2

0.9449

C.V.%

6.1

Pred.R2

0.8873

PRESS

122.72

Adeq Precision

20.135

p-value

Remarks

TabIe-8 ANOVA Analysis for Surface Roughness, Ra

Source

Sum of Squares

DF

Mean Square

F Value

Model

0.19

0.063

18.92

0.0037

Significant

8.82E-05

8.82E-05

0.026

0.8772

Not Significant

0.058

0.058

17.45

0.0087

Significant

0.13

0.13

39.3

0.0015

Significant

3.34E-03

Residual

0.017

Cor-Total

0.21

S.D

0.058

R2

Mean

1.06

Adj.R

0.9191
2

0.8705

0.7425

C.V.%

5.44

Pred.R

PRESS

0.053

Adeq Precision

12.794

TabIe-9 Results from Multiple Regression analysis

Experimental Values

Results from Multiple Regression analysis

Trail No

Surface roughness

MRR

Surface roughness

MRR

1.34

27.66

1.31

27.61

0.98

47.34

1.06

46

0.879

63.12

0.81

64.38

1.256

32.7

1.16

34.13

0.97

56.25

0.91

52.52

1.01

45.56

1.11

48.04

1.11

38.28

40.65

1.14

38.7

1.2

36.18

0.94

54.54

0.95

54.56

Prediction of Optimal Output Responses

At optimal setting conditions, the surface roughness and metal removal rate (MRR) were determined using the
equation (9) and equation (10) respectively

ISSN : 0975-5462

Vol. 3 No. 7 July 2011

5663

M.Janardhan et al. / International Journal of Engineering Science and Technology (IJEST)

Ra= B3R+C3R-Y1 = 0.939+0.939-1.062=0.816---------------- Eq. (9).

MRR= B3M+C3M-Y2 =52.55+54.4-44.9=62.05 ----------- Eq. (10)
Where y1 y2 are mean of average values of surface roughness and Metal Removal Rate respectively.
B3R, C3R are average Surface Roughness for third level of depth of cut and feed rate respectively.
B3M, C3M are average Metal Removal Rate for third level of depth of cut and feed rate respectively.
Response Graphs
Response graphs for the average values and S/N ratios of surface roughness and metal removal rate (MRR) are
shown in fig 3-6.

Fig (3) Response graphs for Ra

Fig (5) Response graphs for MRR

Fig (4) Response graphs for S/N ratio of Ra

Fig (6) Response graphs for S/N ratio of MRR

fig 3 shows the response graphs of surface roughness for two significant parameters namely depth of cut and
feed rate and fig 4 shows the response graphs of S/N ratio for surface roughness with two significant parameters
namely depth of cut and feed rate. It revels that highest s/n ratio values of depth of cut and feed rate are highest
and produced minimum variation in surface roughness at that level. Fig 5 and Fig 6 shows response graphs for
metal removal rate and S/N ratio of metal removal rate and evident that two processes parameters namely depths
of cut and feed rate are significant on output response metal removal rate.
Confidence Intervals
The Confidence intervals at 95% for confirmation tests was determined using the following equation
The Confidence intervals, I= [F (1, fe) x Ve (1/neff)]
Where F is the F ratio at (1-) confidence interval against one degrees of freedom and error degree of freedom
fe.
n eff = ( Number of Experiments)/ (1+ dof of all factors in the estimate of mean)
CI for Surface Roughness
Total no of Trails=18, No of Repetitions=2
neff=3 and F0.05 (1,5)=6.61

ISSN : 0975-5462

Vol. 3 No. 7 July 2011

5664

M.Janardhan et al. / International Journal of Engineering Science and Technology (IJEST)

Error Variance (Ve) =0.0034

Hence, CI = 0.086
Predicted optimal range for surface roughness is given by
0.73 m < Ra < 0.902 m
CI for MRR
Total no of Trails =18, No of Repetitions=2
neff=3 and F0.05(1,5)=6.61
Error Variance (Ve) =7.504
Hence, CI = 4.06
Predicted optimal range for MRR is given by
57.99 gm/min < MRR < 66.11 gm/min
Confirmation of experiments
Two confirmation experiments were conducted at the optimum set of the process parameters. The average
values of surface roughness at the optimum set of the process parameters were 0.826 m and 0.897 m
respectively and they fall in the predicted rage of 0.73 - 0.902 m. The average values of metal removal rate
(MRR) at the optimum set of the process parameters were 60.19 and 62.2 gm/min respectively and they fall in
the predicted rage of 57.99-66.11 gm/min.
5. Conclusions
This paper has presented application of taguchi method to determine the optimal process parameters for
cylindrical grinding process. The concept of ANNOVA and S/N ratio is used to determine the effect and
influence of process parameters namely work speed, feed rate and depth of cut is studied on output responses,
and found that the developed model is significant. MINITAB15 software is used is for analysis of response
graphs of average values and S/N ratios. From the Pareto analysis it is evident that the feed rate played vital role
on output responses surface roughness and metal removal rate (MRR) than other process parameters. The model
predicted in the present work is useful for selecting the right set of process parameters variables for optimal
value of the MRR and Surface roughness.
6. References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]

Taguchi G. introduction to quality engineering (Asian productivity organization Tokyo), 1990.

Roy, RK, Premier on Taguchi Methods (Van Nostrand Reinhold, New York, 1990.
Guptha V and Murthy PN, An introduction to Engineering design methods, Tata McGraw Hill, New Delhi, 1982.
Jae Seob Kwak, Application of Taguchi and response surface methodologies for geometricerror in surface grinding process,
international journal of machine tools &manufacture 45(2005) 327-334.
Jae Seob Kwak, An analysis of grinding power and surface roughness in external cylindrical grinding of hardened SCM 440 Steel
using the response surface method, international journal of machine tools &manufacture 46 (2006) 304-312.
N.Alagumurthi, K. palaniradja and V.Soundararajan, cylindrical grinding a review on Surface integrity. International journal of
precision engineering and manufacturing vol. no 3.
S.S.Mahapatra, Amar Patnaik, Optimization of wire electrical discharge machining (WEDM) Process parameters using Taguchi
method.
Dinesh, Optimization of High speed turning parameters of super alloy inconel Material using Taguchi Technique, Indian journal of
engineering& materials sciences, Feb 2009 pp 44-50.
Naveen Beri, S.Maheswari, Performance Evaluation of Powder Metallurgy Electrode in EDM of AISI D2 Steel using Taguchi Method,
International Journal of Mechanical, Industrial And Aero Space Engineering 2:3 2008.

ISSN : 0975-5462

Vol. 3 No. 7 July 2011

5665