You are on page 1of 10

Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 64 (2013) 785 – 794

International Conference on DESIGN AND MANUFACTURING, IConDM 2013

Optimization of Surface Roughness and Delamination Damage of


GFRP Composite Material in End Milling using Taguchi Design
Method and Artificial Neural Network
Reddy Sreenivasulu*
Department of Mechanical Engineering, R.V.R. & J.C.College of Engineering (Autonomous), Guntur, Andhra Pradesh, INDIA

Abstract

The present work is focused on the influence of cutting speed, feed rate and depth of cut on the delamination damage and surface
roughness on Glass Fiber Reinforced Polymeric composite material (GFRP) during end milling. Taguchi design method is
employed to investigate the machining characteristics of GFRP. From the results of ANOVA, it is concluded that cutting speed
and depth of cut are the most significant factors affecting the responses, their contribution in an order of 26.84% and 40.44%
respectively. Confirmatory experiments show that 5.052μm for surface roughness and 1.682 delamination damage to validate the
used approach after conducting with optimal setting of process parameters. Finally, artificial neural network has been applied to
compare the predicted values with the experimental values, the deviations are found in the range of 3.7%, it shows good
agreement between the predictive model results and the experimental measurements.

©©2013
2013The
TheAuthors.
Authors.Published
Publishedby
by Elsevier
ElsevierLtd.
Ltd. Open access under CC BY-NC-ND license.
Selectionand
Selection andpeer-review
peer-reviewunder
underresponsibility
responsibility of
of the
the organizing
organizingand
andreview
reviewcommittee
committee of
of IConDM
IConDM 2013.
2013

Keywords: GFRP; End milling; Taguchi Design Method; S/N ratio; ANOVA; Neural Network.

Nomenclature
Ra Surface roughness (μm)
Fd Delamination damage factor (mm/ mm)
Greek Symbols
Signal to noise ratio

-------------------
* Corresponding author Tel: 09441069440
E-mail address: rslu1431@gmail.com

1877-7058 © 2013 The Authors. Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
Selection and peer-review under responsibility of the organizing and review committee of IConDM 2013
doi:10.1016/j.proeng.2013.09.154
786 Reddy Sreenivasulu / Procedia Engineering 64 (2013) 785 – 794

1. Introducti on
Milling of composite materials is a rather complex task owing to the heterogeneity of the material and a number
of other problems, such as surface roughness and delamination damage, which appear during the machining process
and are associated with the characteristics of the material and the cutting parameters. Surface roughness and
delamination factor are parameters that have a great influence on dimensional precision and performance of
mechanical pieces. For this reason, research and development have been carried ou t through design experiments to
reach a specific surface roughness and a specific delamination factor.

GFRP are increasingly being used for different engineering applications because of their superior qualitative
advantages include high ratio of strength to weight , high fracture strength and toughness , excellent thermal and
corrosion resistance . The capability of composites for specific purposes has been one of their more advantages and
also one of the greater challenges to adopting them as alternative to conventional materials [1-2]. Bhatnagar etal [3]
focused their study on glass fiber reinforced poly mer (GFRP) p ipe made by filament wind technique, but it requires
further machining to facilitate dimensional accuracy for assembly and control of surface quality for functional
aspects. Davim etal [4] studied the influence of cutting parameters on surface roughness in turning glass -fiber
reinforced plastics using statistical analysis. The users of FRP are facing so many d ifficu lties when machining it,
because technical back ground acquired for conventional materials cannot be applied for such new materials, whose
ability to machin ing is different fro m that of conventional materials , thus it is desirable to experimentally
investigate the behaviour of FRPs during the machin ing process [5]. Everstine and Rogers [6] proposed an analytical
approach of machining FRPs. In their study, they developed a theory of plane deformat ion of inco mpressible
composites reinforced by strong parallel fibers. Koplev etal [7] studied how the fiber orientation significance on the
quality of machined surfaces and tool wear rate. Machining of composite materials depends on the type of fiber
inserted in the co mposites, particularly by the mechanical properties. On the other hand, the se lection of parameters
and tool are dependent on the type of fiber inserted in the co mposites and which is very important in the mach ining
process. Ramulu et al [8] carried out a study on machining of polymer co mposites and concluded that increment in
cutting speed gives good surface finish. Tekeyama et al [9] studied the surface roughness on machining of GFRP
composites, reveals that cutting speed is proportional to delamination damage on the machined surface.. They also
studied the machining characteristics of fiber reinfo rced polymer co mposites using the ultrasonic mach ining
technique. Koing et al [10] found experimentally measurement of surface roughness in FRP is less dependable
compared to that in metals, because projecting fiber tips may cause to addit ional errors it may also cause the fibers
to stick on the stylus. Palaniku mar et al [11] studied the effect o f cutting parameters on surface roughness on
mach ining of GFRP co mposites by polycrystalline d iamond tool by developing a second order model for pr ed icting
the surface roughness. Palaniku mar et al [12] developed a procedure to asses and optimize the chosen factors to
attain minimu m surface roughness by incorporating taguchi method and analysis of variance (ANOVA ) technique.
In the present study, a Taguchi design method and Neural Network model based on experimental measurement data
was developed to predict surface roughness and delamination damage factor in end milling of a GFRP co mposite
material. The factors considered in the experiment were the cut
L9 orthogonal array was used to layout of experimental plan as per factors and levels considered.

2. Experi mental Set up:

2.1. Schematic of Machining:


The work material used for present work is glass fiber reinforced poly meric co mposite material fabricated by
hand layup method of 33% fiber and 66% general purpose resin with randomly oriented long fibers supplied by
Saint Gobain Vetrotex India Limited. The d imensions of the work piece are 300mmX50mmX25 mm. In this study,
the experiments were carried out on a CNC vertical mach ining center (KENT and ND Co. Ltd, Taiwan make) to
perform 10mm s lots on GFRP work p iece by K10 carbide, four flute end milling cutter as shown in fig.1.
Furthermore the cutting speed (rpm), the feed rate (mm/ min) and depth of cut (mm) are regulated in this experiment.
Each experiment was conducted three times and the maximu m width of the delamination damage (W max) around
the slot periphery using Travelling M icroscope with an accuracy o f 10μm is measured at five places on each slot
Reddy Sreenivasulu / Procedia Engineering 64 (2013) 785 – 794 787

then average of them in mm after that Delamination Damage Factor calculated. Similarly, the surface roughness is
measured at five p laces on each slot then average of them in μm is considered by a surface analyser of Form
Talysurf 50 (Taylor Hobson Co Ltd).

T able 1: Properties of GFRP provided by the supplier

Mechanical and Thermal Properties Hand layup method

Tensile Modulus , M Pa 169.75

Tensile Strength , M Pa 60
o
Coefficient of linear expansion, m/m c 2x10 -5

Thermal conductivity , W/mK 0.29

Density, Kg/m 3 1260

Fig.1. Machining of GFRP by CNC vertical machining center with K10 carbide End Mill
[Shanghai China (Mainland) Brand Name: OEM or Ostar Model Number: 404 10]
2.2. Delamination and Surface Roughness and its Measurement

Fig. 2 Measurement of (a) surface roughness (according to ISO 4287-1997)


(b) Delamination damage made at five different sample test points
788 Reddy Sreenivasulu / Procedia Engineering 64 (2013) 785 – 794

Failure analysis of laminated composite materials has attracted a great deal of interest in recent years due to the
increased application in a wide range of high perfor mance structures. Experimental and theoretical studies of failure
analysis and prediction are being conducted by various researchers. Delamination, the separation of two adjacent
plies in co mposite laminates, represents one of the most critical failure mo des in co mposite laminates. In fact, it is
an essential issue in the evaluation of composite laminates for durability and damage tolerance. The value of the
delamination damage factor (F d ) can be obtained using the follo wing equation: F d = W max / W, Where W max is the
maximu m width of the damage around the slot periphery and W is width of cut. The surface roughness parameter
used to evaluate surface roughness in this study is the roughness average (Ra). This parameter is also known as the
arith metic mean roughness value, arith metic average or centreline average. Within the presented research
framework, the discussion of surface roughness is focused on the universally recognized Ra. Ra is recognized
universally as the commonest international parameter of rou ghness. The average roughness is the area between the
roughness profile and its centre line, or the integral of the absolute value of the roughness profile height over the
evaluation length, Yang and Chen [13].

Fig3. Measurement of surface roughness using Form Talysurf 50 (Taylor Hobson Co Ltd)

Fig4. Measurement of delamination damage using Travelling Microscope


Reddy Sreenivasulu / Procedia Engineering 64 (2013) 785 – 794 789

3. Experi mentation as per Taguchi Design Method

A plan of experiments based on Taguchi technique has been used to acquire the dat a. An orthogonal array, signal
to noise (S/N) rat io and analysis of variance (ANOVA) are employed to investigate the cutting characteristics of
glass fiber reinforced poly meric co mposite material using K10 carbide end mill. Finally, confirmat ion test have been
carried out to co mpare the predicted values with the experimental values confirm its effect iveness in the analysis of
surface roughness. Also, neural network technique has been applied to compare the predicted values with the
experimental values and compared the deviation between experimental values of surface roughness and
delamination damage. The orthogonal array forms the basis for the experimental analysis in the Taguchi method.
The selection of orthogonal array is concerned with the total degree o f freedo m o f process parameters. Total degree
of freedom (DOF) associated with three parameters is equal to 6 (3X2).The degree of freedo m for the orthogonal
array should be greater than or at least equal to that of the process parameters. There by, a L9 orthogonal array
having degree of freedo m equal to (9-1= 8) 8 has been considered .But in present case each experiment is conducted
three times, therefore total degree of freedom (9X3-1=26) 26 has been considered finally. The mach ining parameters
and their levels, experimental results are depicted in table2 and 3.

T able 2: Parameters and their Levels.

Symbol Factors units Level 1 Level 2 Level 3

A cutting speed rpm 1000 1250 1500


B feed rate mm/min 200 300 400
C depth of cut mm 0.5 1 1.5

Table 3: Responses observed in the Experimentation as per T aguchi Design

Exp.No Machining parameters Responses


Cutting Feed rate(B) Depth of Delamination Average Surface
Speed(A) mm/min Cut(C) damage Roughness(Ra)
rpm mm Factor(Fd) μm
1 1000 200 0.5 1.22 2.630
2 1000 300 1 1.85 5.306
3 1000 400 1.5 1.24 3.810
4 1250 200 1 1.56 5.178
5 1250 300 1.5 1.47 2.066
6 1250 400 0.5 1.68 6.192
7 1500 200 1.5 1.51 3.266
8 1500 300 1 1.18 8.670
9 1500 400 0.5 1.12 3.852

In the Tagu
ignal to Noise
ratio used to measure the quality ch

(1)

Where y i is the value of the average surface roughness and delamination damage factor for i-th test
To obtain optimal cutting perfor mance, the lower-the-better quality characteristic for surface roughness and
delamination damage factor must be taken as per equation (1).
790 Reddy Sreenivasulu / Procedia Engineering 64 (2013) 785 – 794

4. Results and Discussions


Table 4: Signal to Noise Ratios (For Smaller is better)

Level Cutting Speed A (rpm) Feed Rate B (mm/min) Depth of Cut C (mm)

1 -9.114 -8.655 -11.732


2 -9.940 -10.988 -10.981
3 -11.068 -10.479 -7.409
Delta 1.953 2.332 4.322
Rank 3 2 1

Main Effects Plot for SN ratios


Data Means
cutting speed feedrate

-8
-9
-10
Mean of SN ratios

-11
-12
1000 1250 1500 200 300 400
depth of cut

-8
-9
-10
-11
-12
0.5 1.0 1.5
Signal-to-noise: Smaller is better

Fig. 5: Taguchi Response Graph

Interaction Plot for surface rough ness


Data Means
200 300 400
9
cutting
speed
6 1000
cutting speed 1250
1500
3
9
feedrate
200
6 300
feedr ate 400

3
9
depth
of cut
6 0.5
depth of cut 1.0
1.5
3

1000 1250 1500 0.5 1.0 1.5

Fig. 6: Interaction Plot for surface roughness


Reddy Sreenivasulu / Procedia Engineering 64 (2013) 785 – 794 791

Interaction Plot for delamination damage factor


Data Means
200 300 400

1.8 cutting
speed
1000
1.5
cutting speed 1250
1500
1.2

1.8 feedrate
200
300
1.5
feedr ate 400

1.2

1.8 depth
of cut
0.5
1.5
depth of cut 1.0
1.5
1.2

1000 1250 1500 0.5 1.0 1.5

Fig.7: Interaction Plot for delamination damage factor

Fro m main effects plot of S/N rat io for, the optimu m parameters comb ination of surface roughness and
delamination damage factor are A1B1C3 corresponding to the largest values of S/N ratio for all control parameters.
But fro m main effects plots mean o f individual responses A3B2C1 and A2B2C2 are obtained for the optimu m
parameters co mb ination of surface roughness and delamination damage factor resp ectively. Fro m Table 4, it is
observed that depth of cut has more influence out of all parameters, feed rate has moderate influence and cutting
speed has less influence on both surface roughness and delamination damage factor during end milling of GFRP
composite material.

4.3. Confirmation Test:


The purpose of the analysis of variance (ANOVA) is to investigate which design parameter significantly affects
the quality characteristic. The results of ANOVA depicted in table 5and 6.

Table 5: ANOVA for Surface Roughness Average:

Source DF SS MS F % Contribution

Cutting Speed 2 2.68 1.34 2.987 6.26

Feed Rate 2 4.25 2.13 4.738* 9.92

Depth of Cut 2 17.32 8.66 19.308* 40.44

Error 20 8.97 0.4485

Total 26 33.22

*significant, F table at 95%confidence level is F0.05, 2, 20 = 3.49, F e xp


T able 6; ANOVA for Delamination Damage Factor:
792 Reddy Sreenivasulu / Procedia Engineering 64 (2013) 785 – 794

Source DF SS MS F %Contribution

Cutting Speed 2 0.1356 0.0678 6.286* 26.84


Feed Rate 2 0.0354 0.0177 1.641 7.006

Depth of Cut 2 0.1158 0.0579 5.368* 22.92

Error 20 0.2157 0.01078


Total 26 0.5025

*significant, F table at 95%confidence level is F0.05, 2, 20 = 3.49, F e xp


The analysis of variance (A NOVA ) of the experimental data was done to statistically analy ze the relative
significance of the parameters under the investigation on the response variables. Fro m Table 5, it is observe that
depth of cut has statistical and physical mo re significance (40.44%) on the s urface roughness average obtained in
end milling of GFRP with standard K10 carb ide 4 flute end mill. Feed rate factor has moderate significance
(9.92%).Fro m the analysis it shows that cutting speed factor has less significance (6.26%) on s urface roughness
average. Fro m Table 6, it is observe that cutting speed factor has statistical and physical more sig nificance (26.84%)
on the delamination damage factor obtained in end milling of GFRP with standard K10 carbide 4 flute end mill.
Depth of cut factor has moderate significance (22.92%).Fro m the analysis it shows that Feed rate factor has less
significance (7%) on delamination damage factor.

5. Artificial Neural Network


Artificial neural systems are that physical cellular systems which acquire store and utilize experimental
informat ion. Powerful learn ing algorith m and self-organizing rule allow ANN to self adapt as per the requirements
in continually vary ing environ ment (adaptability property). The A NN architecture is a mu ltilayer, feed forward back
propagation architecture. Multilayer perception (M LP) has an input layer, output layer and hidden layer. Input
vector is incident on input layer and then to hidden layer and subsequently to final layer/output layer via weighted
connections. A neural network is a machine that is designed to model the way in which the brain performs a
particular task or function of interest. To achieve good performance, they emp loy a massive interconnection of

mach ine can be defined as a neural network is a massively parallel d istributed processor made up of simple
processing units, which has a natural propensity for storing experimental knowledge and making it available for use .

Table7. Data Summary of Experimental and Predictive results of ANN


Exp Observed Responses Responses obtained through
No From Experiments (a) Query from ANN (b)
Surface Roughness Delamination Damage Factor Surface Roughness Delamination Damage Factor

1 2.630 1.22 3.5465 1.1780


2 5.360 1.85 5.6526 1.7349
3 3.810 1.24 2.7137 1.3060
4 5.158 1.56 4.7198 1.6059
5 2.066 1.47 3.0259 1.3834
6 6.188 1.68 5.6669 1.6871
7 3.266 1.51 3.3747 1.4575
8 8.670 1.18 8.0082 1.1200
9 3.852 1.12 4.3158 1.1268
Reddy Sreenivasulu / Procedia Engineering 64 (2013) 785 – 794 793

Fig 8: Artificial Neural Network

Fig 9 Comparison of Predicted data vs Experimental dat a for surface roughness via ANN

Fig 10 Comparison of predicted data vs Experimental data for Delamination Damage Factor via ANN
794 Reddy Sreenivasulu / Procedia Engineering 64 (2013) 785 – 794

The experimental observations were incorporated into the ANN model. A feed forward neural network was
developed to predict surface roughness and Delamination damage factor. As in future without undergoing the

mach ining process. Confirmatory experiments were conducted for surface roughness and delamination damage
corresponding their optimal setting of p rocess parameters to validate the used approach, obtained the values of
5.122μm for surface roughness and 1.692 delamination damage factor . Pred icted and experimental values of surface
roughness and delamination damage factor are depicted in table 7.

Table 8: Optimal values of individual machining characteristics

machining characteristics Optimal combination Significant parameters Predicted optimum Experimental


of parameters (at 95% confidence level) value value
Surface Roughness (Ra) A3B2C1 B,C 5.033μm 5.052 μm
Delamination Damage Factor (Fd ) A2B2C2 A,C 1.675 mm/mm 1.682 mm/mm

6. Conclusions
The mach ining characteristics of Glass Fiber Reinforced Po ly meric co mposites have been studied. The primary
mach ining characteristics such as surface roughness and delamination damage factor were studied for End milling.
Fro m S/N Ratio response graph, the comb ination of parameters having the values of 1000, 200 and1.5 are obta ined
for cutting speed (rpm), feed rate (mm/ min) and depth of cut(mm) respectively for optimizing surface roughness and
delamination damage .Fro m S/ N Ratio response table, depth of cut, feed rate and cutting speed are the order of
influence of parameters on surface roughness and delamination damage during machining of GFRP.The result of
ANOVA fo r surface roughness, depth of cut has the more significant parameter for affecting the surface roughness ,
for delamination damage, cutting speed has the more significant parameter for affecting the delamination damage.
The results of ANN concluded that experiment No3 obtained relatively more error than remain ing. The deviation
between experimental values and prediction values are found in the range of 3.7%. Finally, c oncluded that
increasing the number of nodes increases the computational cost and decreases the error. Good agreement was
shown between the predictive model results and the experimental measurements .

References
[1] Bannister, M., 2001. Challenges for composites into the next millennium - a reinforcement perspective. Composites Part A: Applied Science
and Manufacturing, 32(7), 901-910.
[2] Kishore, R.A.; Tiwari, R.; and Singh, I., 2009. Investigation of drilling in [(0/90)/0] s glass fibre reinforced p lastics using taguchi method.
Advances in Production Engineering and Manangement , 4(1-2), 37-46.
[3] Bhatnagar, N., Ramakrishnan, N., Naik, N.K., and. Komandurai, R., 1995, On the machining of fiber Reinforced plastics (FRP) c omposite
laminates, Int J.Machine Tool Manuf., 35 (5),pp. 701-716.
[4] Davim, JP, Mata, F, 2004, Influence of cutting parameters on surface roughness using statistical analysis. Indus Lubrication Tribol, 56(5), pp.
270-274.
[5] Santhana krishanan, G., Krishnamoorthy, R., and Malhotra,
-104.
[6] Everstine, G.C., and. Rogers. T.G., 1971.A Theory of machining of fiber reinforced materials J. Comp. Mater, 5; pp 94 -105.
[7] Koplev A, Lystrup A, Vorm T., 1983. The cutting process, chips and cutting forces in machining CFRP. Composites 14(4):371 376
[8] Ramulu, M., Arola, D., and Colligan, K.., 1994. engineering
systems Design and Analysis, ASME, 64(2), pp. 93-101.
[9] Takeyama H, Lijima N., 1988. Machinability of glass fibre reinforced plastics and application of ultrasonic machining. CIRP Annals 37/1:93
96
[10] Koing, W., Wulf, Ch., Grab, P., and Willerscheid, H., 1985. Machining of Fiber Reinforced Plastics Annals of CIRP, 34, pp. 537-548.
[11] Palanikumar, K.., Karunamoothy, L.., and Karthikeyan, R., 2006 . Assessment of Factors Influencing Surface Roughness on the Machining
of Glass Fiber- Reinforced Composites, Journal of Materials and Design, 27 (10), pp. 862-871.
[12] Palanikumar, K., 2008. Application of Taguchi and Response surface Methodology for surface Roughness In Machining Glass Fiber
Reinforced Plastics by PCD Tooling, IJMPT , 36 (1-2), pp. 19-27.
[13] Yang, J.L. and Chen, J.C., 2001. A systematic approach for identifying optimum surface roughness performance in end-milling operations,
Journal of Industrial Technology., Vol.17, pp.1 8.

You might also like