Arch. Metall. Mater.

62 (2017), 3, 1803-1812

DOI: 10.1515/amm-2017-0273

K. SHUNMUGESH*#, K. PANNEERSELVAM**

OPTIMIZATION OF DRILLING PROCESS PARAMETERS VIA TAGUCHI, TOPSIS AND RSA TECHNIQUES

Carbon Fiber Reinforced Polymer (CFRP) is the most preferred composite material due to its high strength, high modulus,
corrosion resistance and rigidity and which has wide applications in aerospace engineering, automobile sector, sports instrumenta-
tion, light trucks, airframes. This paper is an attempt to carry out drilling experiments as per Taguchi’s L27 (313) orthogonal array
on CFRP under dry condition with three different drill bit type (HSS, TiAlN and TiN). In this research work Response Surface
Analysis (RSA) is used to correlate the effect of process parameters (cutting speed and feed rate) on thrust force, torque, vibra-
tion and surface roughness. This paper also focuses on determining the optimum combination of input process parameter and the
drill bit type that produces quality holes in CFRP composite laminate using Multi-objective Taguchi technique and TOPSIS. The
percentage of contribution, influence of process parameters and adequacy of the second order regression model is carried out by
analysis of variance (ANOVA). The results of experimental investigation demonstrates that feed rate is the pre-dominate factor
which affects the response variables.
Keywords: CFRP, Taguchi, TOPSIS, RSA, Vibration, Optimization

1. Introduction ANOVA. Quality holes can be drilled on bi-directional CFRP

composites using TiN-SC tool. Since TiN-SC exhibits lower
Carbon Fiber Reinforced Polymer (CFRP) is well-known thrust force, surface roughness of the drilled holes improves
for their extraordinarily high strength, high modulus, corrosion considerably [3]. Kurt et al. [4] performed an experimental
resistance and rigidity. Carbon fibers reinforced with epoxy investigation with three different drilling tools on hole quality
resin usually find very good application because of their strength and surface roughness in drilling of Al 2024 in dry condition.
and stiffness combination. The CFRPs are superior to steel and Taguchi technique is successfully used to optimize the machin-
glass fiber reinforced polymer (GFRP) with respect to specific ing process parameters. Grilo et al. [5] studied the influence of
tensile strength and specific elastic modulus. CFRPs with their drill geometry and optimize the process parameters in drilling
positive characteristics are mostly used in applications like aero- of CFRP using three different drill bits with the aim of reducing
space engineering, automobile sector, sports instrumentation, the delamination and increasing the production rate. Tsao and
industrial sector, shafts etc. Joining of composite structures is Hocheng [6] conducted an experimental survey of thrust force
an unavoidable process in many applications; hence the quality and surface roughness models in drilling of CFRP. Correlation
of the machined hole plays a vital role on it. During drilling between the machining process parameter and the responses are
of CFRPs several problems like fuzzing, fiber-pullout, fiber obtained by neural network and regression analysis. Sheth and
breakage, delamination, debonding of fiber/matrix etc., occurs George [7] investigated the effect of machining process param-
due to its anisotropic and non-homogeneity nature of structures. eters during drilling of Wrought Cast Steel Grade B (WCB) on
Conventional method of drilling remains at top priority in lieu of cylindricity and perpendicularity. Taguchi L27 (313) orthogonal
its mechanical/thermal properties and also the cost and quality array is used for the experimental design and the significance
of the manufactured goods. of process parameters is evaluated by ANOVA. Nouari et al.
Arul et al. [1] proposed a new technique by inducing [8] analyzed the influence of input process parameters in drill-
vibration in the direction of feed, there by reduces thrust ing aluminium alloy on tool life and hole quality. They have
force and improves the quality of drilled holes. Ramesh et al. recommended using reduced web thickness, increasing the
[2] have worked on multiple performance characteristics of helix and point angle to optimal drill geometry. Feed rate and
drilling process parameters using Grey Relational Analysis drill type are the significant factors which affect thrust force
(GRA), TOPSIS and Response Surface Methodology (RSM). while feed rate and cutting speed contributes more to the total
The adequacy of the developed model is validated by means of variation of torque [9]. Neseli [10] investigated the effect of
* VJCET, VAZHAKULAM, DEPARTMENT OF MECHANICAL ENGINEERING, KERALA, INDIA
** NIT, TRICHY, DEPARTMENT OF PRODUCTION ENGINEERING, TAMILNADU, INDIA
# Corresponding author: shunmugesh@gmail.com

Unauthenticated
Download Date | 10/20/17 2:49 PM
1804
machining process parameters in drilling of AISI 1040 steel with and for better results, average of three replicates was found out.
the aim of reducing thrust force and torque. The responses are The Kistler multicomponent dynamometer 9257B was used to
minimized at higher cutting speed and low feed rate. The major measure the three orthogonal components of cutting forces i.e.,
damages encountered in drilling are at the entry and exit of the Fx, Fy and Fz for all the trial experiments. The dynamometer
holes and surface roughness [11]. The indirect measurement consists of multi-component measuring system and multi-charge
of these damage are through thrust force and torque whereas amplifier channels. These channels convert the charge signals
the direct measurement are through delamination factor and from the dynamometer into output signals. A piezoelectric
surface roughness. Jayabal and Natarajan [12] studied the ef- accelerometer with a sampling rate of 25600 samples/second
fect of machining parameters in drilling of coir fiber reinforced was used as vibration sensor (Dytran make 3055B) and it was
composites on thrust force, torque and tool wear. Their results mounted on the top of the workpiece. The signals from the pickup
indicate that the minimum thrust force is at higher cutting speed were fed to a signal conditioner and a vibration indicator. Tool
and lower feed rate. Dewagan et al. [13] applied Fuzzy-TOPSIS vibration, particularly in the radial direction, is known to have
coupled with sensitivity analysis to optimize the EDM process a deleterious effect on the machined surface texture. During this
parameters with multiple performance characteristics. Gok study, 3 controllable process parameters: cutting speed, feed rate
[14] developed a new approach to optimize the turning process and drill bit type are elite and varied in 3 completely different
parameters using Fuzzy-TOPSIS, Grey Relational Analysis levels as shown in Table 1. The schematic experimental setup
(GRA) and Response Surface Analysis (RSA). Madhavan et is shown in Fig. 1.
al. [15] conducted drilling simulation on CFRP to predict the
force and observed that the fiber orientation angle is the most TABLE 1
influencing process parameter. Zhang et al. [16] optimized Machining parameters and their levels
turbodrill blades design using response surface methodology
Levels
(RSM) and reported that the efficiency of the optimized design Process parameters Symbol
1 2 3
increases the performances by 10%. Bobbili et al. [17] con- Cutting speed, m/min v 30 40 50
ducted experiments to study the effect of WEDM parameters Feed rate, mm/rev f 0.025 0.05 0.1
on multiple performance characteristics, – MRR and Ra and Drill bit type d HSS TiAlN TiN
also optimized the input process parameters.
The objective of the current work is to study the effect of
input process parameters (cutting speed, feed rate and drill bit 3. Result and discussions
type) in drilling of CFRP composite laminate on the experimental
results of thrust force, torque, vibration and surface roughness. 3.1. Response Surface Analysis (RSA)
Finally optimum condition to obtain minimum thrust force,
torque, vibration and surface roughness when drilling 6mm The second order regression model for thrust force, torque,
diameter hole on a 8mm thick laminate are found out using vibration and surface roughness were obtained and are repre-
Multi-objective Taguchi technique, TOPSIS and RSA. In addi- sented in Table 2. These regression models are used to predict
tion significance and contribution of input process parameters the experimental results of thrust force, torque, vibration and
are identified using ANOVA. surface roughness at certain conditions of process parameter.
The predicted values of Fo, To, Vib and Ra were obtained
from Table 2 are compared with experimental results. It is clear
2. Materials and methods and evident that the predicted values are in good agreement with
the experimental results as shown in Fig. 2.
In this paper CFRP manufactured by hand layup and auto
clave was chosen as the work piece material for conducting
drilling experiments. The work piece used for the experiments 3.2. Multi-objective Taguchi Membership Function
is CFRP (T300 Bi-directional Carbon Fibre/Epoxy matrix with (MOTMF)
a volume fraction of 60%). The carbon fiber used in the mate-
rial is of PAN-based and the average thickness of the fabric is In this section, initially the experimental results are analyzed
0.25 mm. The size of the specimen used was 150×15×8 mm. and evaluated individually. Later the MOTMF is used to for the
The experiments were carried out using BFW Ltd BMV 40T20 multiple performance characteristics. The effect of machining
CNC vertical milling machine. The cutting tools used for the dry process parameters on drilling of CFRP was analyzed using
drilling are HSS drill (Miranda Tools India Ltd), Kennametal responses surface analysis. The experimental results of thrust
Solid Carbide Drill (TiAlN Black Coated-KC7325 Grade) and force, torque, vibration and surface roughness are shown in
WIDIA Solid Carbide Drill (TiN Golden coated – WU25PD Table 3. Now using the MOTMF approach, the multi-objective
Grade). The drilling tools of 6 mm diameter and 2 flutes are used function is converted into single objective function. The higher
for all the experimental trials. The surface roughness value Ra value of S/N ratio indicates better performance. The experimental
was measured by using Mitutoyo SJ-210 surface roughness tester responses, normalized data, membership function, MOF and S/N

Unauthenticated
Download Date | 10/20/17 2:49 PM
 1805
TABLE 2
Regression model for thrust force, torque, vibration and surface roughness

Regression model R2 Pred. R2

Fo 100.554  0.0388889v  454.698 f  21.6361d  0.0113889v ^ 2
94.3 87.10
 1345.19 f ^ 2 5.94444d ^ 2  1.69524vf  0.0741667vd  20.1429 fd
To 74.0537  1.83250v  286.286 f  14.4917 d  0.0183333v ^ 2
97.6 93.15
 1828.15 f ^ 2 3.26667 d ^ 2  0.70vf  0.0108333vd  6.52381 fd

Vi 0.0357296  0.000324v  0.497556 f  0.0187278d  7.555e ^  6  v ^ 2

94.2 85.72
 2.8 f ^ 2 0.004194d ^ 2  0.00105714vf  4.58333e ^  5  vd  0.00780952 fd
Ra 8.74222  0.186972v  23.5603 f  0.640778d  0.00191944v ^ 2  123.348 f ^ 2
98.9 97.24
 0.13811d ^ 2  0.0570476vf  0.000841667vd  0.413333 fd

Fig. 1. Experimental setup

Unauthenticated
Download Date | 10/20/17 2:49 PM
1806

(a) (b)

(c) (d)
Fig. 2. Comparison plots of actual and predicted values of a) Thrust force b) Torque c) Vibration and d) Surface roughness

TABLE 3
Experimental design, responses, normalized data, membership function, MOF and S/N ratio

Control factor Responses Normalized data Membership function

Exp S/N
Ra Fo To Vi Ra Fo To Vi Ra Fo To Vi MOF
No v f d Ratio
(μm) (N) (Nm) (μm) (μm) (N) (Nm) (μm) (μm) (N) (Nm) (μm)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1 30 0.025 HSS 4.971 83 30.5 0.035 0.574 0.542 0.559 0.418 0.396 0.429 0.411 0.560 0.585 4.661
2 30 0.025 TiAlN 4.76 78 27.5 0.030 0.454 0.458 0.390 0.177 0.522 0.518 0.590 0.813 0.315 10.032
3 30 0.025 TiN 4.8 78.5 30.4 0.035 0.477 0.466 0.554 0.413 0.498 0.509 0.417 0.565 0.485 6.285
4 30 0.05 HSS 5.32 92 36 0.041 0.773 0.695 0.871 0.691 0.187 0.268 0.083 0.273 1.219 –1.718
5 30 0.05 TiAlN 5 85 31.4 0.037 0.590 0.576 0.610 0.493 0.378 0.393 0.357 0.481 0.682 3.320
6 30 0.05 TiN 5.09 86.2 34.5 0.043 0.642 0.597 0.786 0.790 0.325 0.372 0.173 0.169 1.057 –0.485
7 30 0.1 HSS 5.632 107 37.4 0.044 0.950 0.950 0.950 0.842 0.000 0.000 0.000 0.114 1.798 –5.095
8 30 0.1 TiAlN 5.4 101 34.6 0.040 0.818 0.848 0.792 0.667 0.139 0.107 0.167 0.298 1.295 –2.244
9 30 0.1 TiN 5.47 103 36.3 0.046 0.858 0.882 0.888 0.950 0.097 0.071 0.066 0.000 1.687 –4.540
10 40 0.025 HSS 4.404 75 26.1 0.036 0.251 0.407 0.310 0.446 0.735 0.572 0.673 0.531 0.276 11.195
11 40 0.025 TiAlN 4.197 68 21.5 0.030 0.134 0.288 0.050 0.187 0.859 0.697 0.947 0.804 0.073 22.778
12 40 0.025 TiN 4.245 69 23.4 0.032 0.161 0.305 0.158 0.267 0.831 0.679 0.834 0.719 0.113 18.937
13 40 0.05 HSS 4.83 81 31.6 0.042 0.494 0.508 0.622 0.733 0.480 0.465 0.346 0.228 0.751 2.490
14 40 0.05 TiAlN 4.647 74.5 27.2 0.034 0.390 0.398 0.373 0.394 0.590 0.581 0.608 0.585 0.318 9.949
15 40 0.05 TiN 4.771 93 29.7 0.039 0.460 0.712 0.514 0.620 0.516 0.250 0.459 0.347 0.720 2.853
16 40 0.1 HSS 5.125 96 34 0.043 0.662 0.763 0.758 0.804 0.304 0.197 0.203 0.154 1.179 –1.431
17 40 0.1 TiAlN 4.99 86 29.6 0.037 0.585 0.593 0.508 0.531 0.384 0.375 0.465 0.441 0.650 3.747

Unauthenticated
Download Date | 10/20/17 2:49 PM
 1807

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
18 40 0.1 TiN 4.95 91.5 31.7 0.041 0.562 0.687 0.627 0.719 0.408 0.277 0.340 0.243 0.894 0.975
19 50 0.025 HSS 4.302 64 25.1 0.028 0.193 0.220 0.254 0.092 0.796 0.769 0.733 0.903 0.083 21.566
20 50 0.025 TiAlN 4.05 54 22 0.027 0.050 0.050 0.078 0.050 0.947 0.947 0.918 0.947 0.007 42.884
21 50 0.025 TiN 4.24 58.4 22.8 0.031 0.158 0.125 0.124 0.234 0.834 0.869 0.870 0.754 0.058 24.710
22 50 0.05 HSS 4.74 71 28.8 0.034 0.443 0.339 0.463 0.385 0.534 0.643 0.512 0.595 0.354 9.015
23 50 0.05 TiAlN 4.538 65 24.5 0.034 0.328 0.237 0.220 0.370 0.655 0.751 0.769 0.610 0.184 14.720
24 50 0.05 TiN 4.58 66 28.4 0.038 0.352 0.254 0.441 0.573 0.630 0.733 0.536 0.397 0.374 8.545
25 50 0.1 HSS 5.11 94 31.6 0.037 0.653 0.729 0.622 0.526 0.313 0.232 0.346 0.446 0.853 1.378
26 50 0.1 TiAlN 4.9 80 27.9 0.035 0.534 0.492 0.412 0.432 0.438 0.483 0.566 0.546 0.465 6.660
27 50 0.1 TiN 4.87 81.4 32.9 0.040 0.516 0.515 0.695 0.658 0.456 0.458 0.268 0.308 0.762 2.357

ratio are presented in Table 3. From the Table it is evident that the 3.3. TOPSIS study
experiment number 20 has attained the maximum value of S/N
ratio amoung the 27 number of experiment and the optimum con- Initially all the four experimental results of thrust force,
dition to achieve the multiple performance characteristics (S/N torque, vibration and surface roughness are normalized. In the
ratio = 48.074, cutting speed = 50 m/min, feed rate = 0.025 mm/ current study all the four responses is assigned equal weight-
rev and TiAlN drill bit type). age, hence the weight of thrust force, torque, vibration and
The mean effect Table and graph of S/N ratio are given in surface roughness are taken as 0.25. Once the weights are as-
Table 4 and Fig. 3, respectively. In MOTMF, maximum value of signed, the weighted normalized decision matrix is calculated.
S/N ratio indicates better performance. Thus the optimum level From the weighted normalized decision matrix, the positive
of process parameter is the one with maximum S/N ratio value. and negative ideal solutions are calculated. The positive and
Therefore, to achieve minimum value of thrust force, torque, negative ideal solutions of thrust force, torque, vibration and
+ –
vibration and surface roughness, simultaneously, the optimum surface roughness are said to be u Fo = 0.5727, uFo = 0.2890,
+ – + –
combination of machining parameters obtained are v3 (cutting u To = 0.3311, u To = 0.1903, u Vib = 0.0116, u Vib = 0.0068,
+ –
speed, 50 m/min), f1 (feed rate, 0.025 mm/rev) and d2 (drill bit u Ra = 0.1235 and u Ra = 0.0888. Finally, the comparative
type, TiAlN). closeness of distinct alternative is calculated, and is presented
in Table 5.
TABLE 4 From the Table 5, it is evident that the experiment number
Mean effect TABLE of S/N ratio 20 was the better performer. The order of the experimental run
obtained by TOPSIS was given by 20 > 21 > 19 > 23 > 11 > 12
Level Cutting speed Feed rate Drill bit type > 24 > 22 > 10 > 14 > 2 > 26 > 3 > 13 > 1 > 27 > 17 > 5 > 6 > 15
1 1.135 18.117 4.673 > 18 > 25 > 4 > 16 > 8 > 9 > 7. The higher value of comparative
2 7.944 5.410 12.427 closeness value indicates better performance. From the Table 5,
3 14.648 0.201 6.626 it is evident that the experiment number 20 has attained the maxi-
Max – Min 13.513 17.916 1.953 mum value of comparative closeness amoung the 27 number of
Rank 2 1 3 experiment and the optimum condition to achieve the multiple

Fig. 3. Mean effect plot of S/N ratio

Unauthenticated
Download Date | 10/20/17 2:49 PM
1808
TABLE 5
TOPSIS (normalized & weighted normalized data, separation measures and comparative nearness)

Normalized Weighted Separation Comparative

Exp data normalized data measures nearness
No –
Ra (μm) Fo (N) To (Nm) Vi (μm) Ra (μm) Fo (N) To (Nm) Vi (μm) S +I SI Ci
1 0.436 1.777 1.080 0.035 0.109 0.444 0.270 0.009 0.1756 0.1430 0.44878
2 0.418 1.670 0.974 0.030 0.104 0.418 0.243 0.008 0.1398 0.1793 0.56179
3 0.421 1.681 1.077 0.035 0.105 0.420 0.269 0.009 0.1538 0.1656 0.51847
4 0.467 1.970 1.275 0.041 0.117 0.492 0.319 0.010 0.2421 0.0815 0.25190
5 0.439 1.820 1.112 0.037 0.110 0.455 0.278 0.009 0.1888 0.1299 0.40765
6 0.447 1.846 1.222 0.043 0.112 0.461 0.305 0.011 0.2085 0.1148 0.35520
7 0.494 2.291 1.324 0.044 0.124 0.573 0.331 0.011 0.3186 0.0005 0.00180
8 0.474 2.162 1.225 0.041 0.118 0.541 0.306 0.010 0.2786 0.0409 0.12805
9 0.480 2.205 1.285 0.047 0.120 0.551 0.321 0.012 0.2948 0.0237 0.07465
10 0.386 1.606 0.924 0.036 0.097 0.401 0.231 0.009 0.1198 0.2002 0.62557
11 0.368 1.456 0.761 0.030 0.092 0.364 0.190 0.008 0.0750 0.2537 0.77186
12 0.372 1.477 0.829 0.032 0.093 0.369 0.207 0.008 0.0821 0.2401 0.74511
13 0.424 1.734 1.119 0.042 0.106 0.434 0.280 0.011 0.1708 0.1493 0.46649
14 0.408 1.595 0.963 0.035 0.102 0.399 0.241 0.009 0.1215 0.1972 0.61878
15 0.419 1.991 1.052 0.040 0.105 0.498 0.263 0.010 0.2216 0.1030 0.31745
16 0.450 2.055 1.204 0.044 0.112 0.514 0.301 0.011 0.2517 0.0670 0.21037
17 0.438 1.841 1.048 0.038 0.109 0.460 0.262 0.009 0.18685 0.13269 0.41526
18 0.434 1.959 1.123 0.042 0.109 0.490 0.281 0.010 0.2210 0.0982 0.30776
19 0.377 1.370 0.889 0.028 0.094 0.343 0.222 0.007 0.0625 0.2563 0.80387
20 0.355 1.156 0.779 0.028 0.089 0.289 0.195 0.007 0.0044 0.3166 0.98621
21 0.372 1.250 0.807 0.031 0.093 0.313 0.202 0.008 0.0265 0.2921 0.91665
22 0.416 1.520 1.020 0.035 0.104 0.380 0.255 0.009 0.1126 0.2081 0.64884
23 0.398 1.392 0.868 0.034 0.100 0.348 0.217 0.009 0.0654 0.2533 0.79456
24 0.402 1.413 1.006 0.039 0.100 0.353 0.251 0.010 0.0894 0.2346 0.72399
25 0.448 2.013 1.119 0.038 0.112 0.503 0.280 0.009 0.2332 0.0872 0.27230
26 0.430 1.713 0.988 0.036 0.107 0.428 0.247 0.009 0.1514 0.168 0.52595
27 0.427 1.743 1.165 0.040 0.107 0.436 0.291 0.010 0.1789 0.1436 0.44532

performance characteristics (cutting speed = 50 m/min, feed 3.4. ANOVA

rate = 0.025 mm/rev and TiAlN drill bit type). Thus, the results
of TOPSIS technique are in conformity with the multi-objective Analysis of Variance (ANOVA) is used to validate the
Taguchi technique. regression models and the results are presented in Table 6-9.

TABLE 6
Analysis of Variance for Thrust Force

Source DF Seq. SS Adj. SS Adj. MS F P

Regression 9 4635.65 4635.65 515.07 31.06 0.000
Linear 3 4380.76 4315.26 1438.42 86.73 0.000
v 1 1798.00 1741.58 1741.58 105.01 0.000
f 1 2510.76 2496.89 2496.89 150.55 0.000
d 1 72.00 76.80 76.80 4.63 0.046
Square 3 236.16 236.16 78.72 4.75 0.014
v×v 1 7.78 7.78 7.78 0.47 0.503
f×f 1 16.36 16.36 16.36 0.99 0.335
d×d 1 212.02 212.02 212.02 12.78 0.002
Interaction 3 18.73 18.73 6.24 0.38 0.771
v×f 1 5.03 5.03 5.03 0.30 0.589
v×d 1 6.60 6.60 6.60 0.40 0.537
f×d 1 7.10 7.10 7.10 0.43 0.522
Residual Error 17 281.94 281.94 16.58
Total 26 4917.59

Unauthenticated
Download Date | 10/20/17 2:49 PM
 1809
TABLE 7
Analysis of Variance for Torque

Source DF Seq. SS Adj. SS Adj. MS F P

Regression 9 512.054 512.054 56.895 76.85 0.000
Linear 3 395.904 412.791 137.597 185.86 0.000
v 1 165.620 159.600 159.600 215.58 0.000
f 1 223.562 247.161 247.161 333.85 0.000
d 1 6.722 6.030 6.030 8.14 0.011
Square 3 114.407 114.407 38.136 51.51 0.000
v×v 1 20.167 20.167 20.167 27.24 0.000
f×f 1 30.213 30.213 30.213 40.81 0.000
d×d 1 64.027 64.027 64.027 86.48 0.000
Interaction 3 1.743 1.743 0.581 0.78 0.519
v×f 1 0.857 0.857 0.857 1.16 0.297
v×d 1 0.141 0.141 0.141 0.19 0.668
f×d 1 0.745 0.745 0.745 1.01 0.330
Residual Error 17 12.586 12.586 0.740
Total 26 524.640

TABLE 8
Analysis of Variance for Vibration

Source DF Seq. SS Adj. SS Adj. MS F P

Regression 9 0.000623 0.000623 0.000069 30.51 0.000
Linear 3 0.000437 0.000487 0.000162 71.56 0.000
v 1 0.000113 0.000114 0.000114 50.47 0.000
f 1 0.000323 0.000370 0.000370 163.13 0.000
d 1 0.000002 0.000002 0.000002 1.08 0.314
Square 3 0.000180 0.000180 0.000060 26.44 0.000
v×v 1 0.000003 0.000003 0.000003 1.51 0.236
f×f 1 0.000071 0.000071 0.000071 31.25 0.000
d×d 1 0.000106 0.000106 0.000106 46.55 0.000
Interaction 3 0.000006 0.000006 0.000002 0.81 0.503
v×f 1 0.000002 0.000002 0.000002 0.86 0.366
v×d 1 0.000003 0.000003 0.000003 1.11 0.306
f×d 1 0.000001 0.000001 0.000001 0.47 0.502
Residual Error 17 0.000039 0.000039 0.000002
Total 26 0.000661

TABLE 9
Analysis of Variance for Surface Roughness

Source DF Seq. SS Adj. SS Adj. MS F P

Regression 9 4.24751 4.24751 0.47195 165.61 0.000
Linear 3 3.76493 3.84907 1.28302 450.22 0.000
v 1 1.45238 1.40312 1.40312 492.36 0.000
f 1 2.20084 2.33136 2.33136 818.08 0.000
d 1 0.11171 0.11460 0.11460 40.21 0.000
Square 3 0.47305 0.47305 0.15768 55.33 0.000
v×v 1 0.22106 0.22106 0.22106 77.57 0.000
f×f 1 0.13754 0.13754 0.13754 48.26 0.000
d×d 1 0.11445 0.11445 0.11445 40.16 0.000
Interaction 3 0.00954 0.00954 0.00318 1.12 0.370
v×f 1 0.00570 0.00570 0.00570 2.00 0.176
v×d 1 0.00085 0.00085 0.00085 0.30 0.592
f×d 1 0.00299 0.00299 0.00299 1.05 0.320
Residual Error 17 0.04845 0.04845 0.00285
Total 26 4.29596

Unauthenticated
Download Date | 10/20/17 2:49 PM
1810

(a) (b)

(c) (d)
Fig. 4. Parametric influence of machining parameters on a) Thrust force, b) Torque c) Vibration and d) Surface roughness

The effect of input process parameters on the experimental Mean while, an increase of 66.6 % in cutting speed decreases
results are studied by means of ANOVA. From the ANOVA the torque by 24% of total variation. Minimum value of torque
Table of all the four responses it is evident that the cutting is observed to be 22 Nm at 50 m/min, 0.025 mm/rev and with
speed, feed rate and drill bit type have significant effect on the TiAlN drill bit type.
responses as the “p” value of the models are less than 0.05. According to the experimental results, an increase in 66.6%
Also the adequacy of the model was validated by co-efficient of cutting speed results in decreasing the magnitude of vibration
of determination (R2) and the results are shown in Table 2. The by 10.77% but increasing the feed rate by 75% increases the vi-
response surface methodology (RSM) performed by the statisti- bration by 31.03%. Feed rate is the most significant factor which
cal software (MINITAB 16) was used for mathematical model- affects the vibration as it contributes 59.53% of total variation,
ling of the experimental results. The second order regression followed by cutting speed (17.53%) and drill bit type (16.27%).
model of thrust force, torque, vibration and surface roughness Minimum value of vibration is observed to be 0.0276 μm at
are developed by the correlation of input process parameters 50 m/min, 0.025 mm/rev and with TiAlN drill bit type.
using the RSM technique. Increasing cutting speed by 66.6% results in good surface
The main effect plot and ANOVA results of thrust force, roughness by 12.44% of total variation. From the experimental
torque, vibration and surface roughness are shown in Fig. 4 data, it is observed that 75% increase in feed rate increases
and Table 6-9 respectively. According to the experimental data, vibration by 18.60% of total variation. Feed rate contributes
an increase in 66.6% of cutting speed results in decreasing the 54.50% to the total variation of vibration, followed by cutting
magnitude of thrust force by 31.82% but increasing the feed speed (38.95) and drill bit type (5.30%).
rate by 75% increases the thrust force by 32.83%. Feed rate is
found to be the principal factor which affects the thrust force
as it contributes 54.50% of total variation, followed by cutting 3.5. SEM Images
speed (38.95%) and drill bit type (5.30%). Minimum value of
thrust force is observed to be 54 N at 50 m/min, 0.025 mm/rev SEM images of machined test specimen are given in Fig. 5.
and with TiAlN drill bit type. In Fig. 5a fiber pullout is seen. Fig. 5b shows fiber breakage.
Feed rate contributes 51.39% to the total variation of Fig. 5c shows a sample surface with internal delamination.
torque, followed by cutting speed (36.72%) and drill bit type Fig. 5d shows voids while machining with HSS drill bit at low
(5.78%). From the experimental data, it is observed that 75% cutting speed and high feed rate. Fig. 5e shows a rough surface
increase in feed rate increases torque by 34.8% of total variation. texture that occurred at machining of 30 m/min and feed rate

Unauthenticated
Download Date | 10/20/17 2:49 PM
 1811

{{{{{ [[[[

Fig. 5. SEM Images of the test specimen

of 0.1 mm/rev. Fig. 5f shows an apparently accep Table surface 4. Conclusions

texture that occurred with TiAlN drill bit tool at 50 m/min cutting
speed and lowest feed rate employed, 0.025 mm/rev. The surface In this paper, drilling experiments are carried out as per
quality investigation using SEM suggest optimum surface finish Taguchi L27 (313) orthogonal array experimental design, on CFRP
with highest cutting speed employed, lowest feed rate used and laminates using HSS, TiAlN and TiN drill bit type. The machin-
hardest tool material TiAlN. ing process parameters are optimized using multi-objective

Unauthenticated
Download Date | 10/20/17 2:49 PM
1812
Taguchi technique and TOPSIS. The conclusions drawn from [4] M. Kurt, E. Bagci, Y. Kaynak, The Int. J. Adv. Manu.f Tech. 40
the experimental results are summarized as follows: (5-6), 458-469 (2009).
1. From the multi-objective Taguchi technique, the most [5] T.J. Grilo, R.M.F. Paulo, C.R.M. Silva, J.P. Davim, Compos. Part
predominant process parameter was found to be feed rate, B-Eng. 45 (1), 1344-1350 (2013).
followed by cutting speed and drill bit type. [6] C.C. Tsao, H. Hocheng, J. Mater. Process. Tech. 203 (1), 342-348
2. Minimum surface roughness (Ra = 4.05 μm), minimum (2008).
thrust force (Fo = 54 N), minimum torque (To = 22 Nm) [7] S. Sheth, P.M. George, Precis. Eng. 2016.
and minimum vibration (0.0276 μm) was obtained from [8] M. Nouari, G. List, F. Girot, D. Coupard, Wear 255 (7), 1359-1368
the optimum combination of process parameters (v3 f1 d2). (2003).
3. The results of TOPSIS technique are in good agreement [9] I.S. Shyha, D.K. Aspinwall, S.L. Soo, S. Bradley, Int. J. Mach.
with the multi-objective Taguchi technique. Tool Manu. 49 (12), 1008-1014 (2009).
4. The actual values of thrust force, torque, vibration and [10] S. Neseli, Adv. Mech. Eng. 6, 925382 (2014).
surface roughness matches closely with those predicted by [11] A.M. Abrao, P.E. Faria, J.C. Rubio, P. Reis, J.P. Davim, J. Mater.
the second order mathematical model, shows the goodness Process Tech. 186 (1), 1-7 (2007).
of the developed models. [12] S. Jayabal, U. Natarajan, The Int. J. Adv. Manuf. Tech. 51 (1-4),
5. From the results of ANOVA feed rate is found to be the 371-381 (2010).
most predominant factor which affects the responses. [13] S. Dewangan, S. Gangopadhyay, C.K. Biswas, Measurement
63(1), 364-376 (2015).
[14] A. Gok, Measurement 70 (1), 100-109 (2015).
REFERENCES [15] V. Madhavan, G. Lipczynski, B. Lane, E. Whitenton, J. Manuf.
Process. 20 (2), 431-442 (2015).
[1] S. Arul, L. Vijayaraghavan, S.K. Malhotra, R. Krishnamurthy, Int. [16] X. Zhang, S. Yu, Y. Gong, Y. Li, Adv. Mech. Eng. 8 (2), (2016).
.J Mach. Tool Manu. 46 (3), 252-259 (2006). [17] R. Bobbili, V. Madhu, A.K. Gogia, Defence Tech. 11 (4), 344-349
[2] S. Ramesh, R. Viswanathan, S. Ambika, S, Measurement 78 (1), (2015).
63-72 (2016). [18] R.K. Pandey, S.S. Panda, Measurement 59 (1), 9-13 (2015).
[3] M.A. Herbert, N. Shetty, R. Shetty, D.S. Shetty, G.S. Vijay, Appl.
Soft. Comput. 41 (1), 466-478 (2016).

Unauthenticated
Download Date | 10/20/17 2:49 PM