Bharatiya Jyotish Sastra-1

Journal of Composite Materials
ANALYSIS AND PREDICTION OF SLIDING WEAR RESPONSE

OF EPOXY-PALMYRA FIBER COMPOSITES USING TAGUCHI
Fo
DESIGN AND NEURAL COMPUTATION
r
Journal: Journal of Composite Materials
Manuscript ID JCM-15-1073
Pe
Manuscript Type: Original Manuscript
Date Submitted by the Author: 27-Nov-2015

er
Complete List of Authors: BISWAL, SOMEN; National Institute of Technology, Rourkela, Mechanical
Engineering
Satpathy, Alok ; NIT Rourkela.India , Mechanical Engineering
Re
Epoxy composites, Palmyra fiber, Sliding wear, Taguchi method, Neural

Keywords:
computation
This paper presents a parametric analysis of dry sliding wear process

undergone by a new class of composite material consisting of epoxy and
vi
short palmyra fibers (SPF). It is an attempt to explore the possibility of

improving the dry sliding wear resistance of neat epoxy by reinforcing it
SPF. Palmyra fiber is an inexpensive, strong and naturally available fiber
ew
derived from the stalks of palmyra leaf. In the present investigation, epoxy
based composites with different weight proportions of SPF (0, 4, 8 and 12
wt%) are fabricated though the simple hand-lay-up route. The composite
samples are subjected to wear trials using a pin-on-disc test rig as per
ASTM G 99-05 test standards under different operating conditions. The
Abstract: effects of various operational variables on the dry sliding wear behavior of
these epoxy composites with and without short palmyra fiber
reinforcement are studied. The design of experiments approach using
Taguchi’s L16 orthogonal array is used for parametric analysis of the wear
process. Significant parameters affecting the wear rate of epoxy are
identified. This study revealed that while the fiber content and the sliding
velocities have significant influence on the wear arte of the composites, the
effects of sliding distance and normal load are ignorable. An optimal
parameter setting for minimum specific wear rate is obtained. The
morphologies of the worn surfaces are examined by scanning electron
microscopy and possible wear mechanisms are identified.
http://mc.manuscriptcentral.com/jcm
Page 1 of 19 Journal of Composite Materials
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Journal of Composite Materials Page 2 of 19
1
2
3
4
5 Analysis and Prediction of Sliding Wear Response of
6
7 Epoxy-Palmyra Fiber Composites Using Taguchi Design
8
9
and Neural Computation
10
11
12
13
14 Somen Biswal and Alok Satapathy
15 Department of Mechanical Engineering, National Institute of Technology, Rourkela-
16
17
769008, India
18 Email: somenbiswal7@gmail.com, alok@nitrkl.ac.in
Fo
19
20
21 Abstract
22
r
23
24 This paper presents a parametric analysis of dry sliding wear process undergone by a new
Pe
25 class of composite material consisting of epoxy and short palmyra fibers (SPF). It is an
26
27 attempt to explore the possibility of improving the dry sliding wear resistance of neat
28
er
29 epoxy by reinforcing it SPF. Palmyra fiber is an inexpensive, strong and naturally

30
31 available fiber derived from the stalks of palmyra leaf. In the present investigation, epoxy
32
based composites with different weight proportions of SPF (0, 4, 8 and 12 wt%) are
Re
33
34 fabricated though the simple hand-lay-up route. The composite samples are subjected to
35
36 wear trials using a pin-on-disc test rig as per ASTM G 99-05 test standards under
vi
37
38 different operating conditions. The effects of various operational variables on the dry
39
ew
40 sliding wear behavior of these epoxy composites with and without short palmyra fiber
41 reinforcement are studied. The design of experiments approach using Taguchi’s L16
42
43 orthogonal array is used for parametric analysis of the wear process. Significant
44
45 parameters affecting the wear rate of epoxy are identified. This study revealed that while
46
47 the fiber content and the sliding velocities have significant influence on the wear arte of
48
the composites, the effects of sliding distance and normal load are ignorable. An optimal
49
50 parameter setting for minimum specific wear rate is obtained. The morphologies of the
51
52 worn surfaces are examined by scanning electron microscopy and possible wear
53
54 mechanisms are identified.
55
56 Keywords: epoxy composites, palmyra fiber, sliding wear, taguchi method, neural
57 network
58
59
60
1
2
3
4
5
6 Introduction
7
8
9 Natural fiber reinforced composites now-a-days are being used extensively because of
10
11
12 many advantages offered by them. Unlike synthetic fibers, natural fibers are low cost,
13
14 bio-degradable, eco-friendly, light weight and easily available. Being completely
15
16
renewable, natural fibers provide many advantages to the environment. Also the waste
17
18
Fo
19 generated by these natural fibers is organic and hence can be used for other purposes such
20
21 as electricity generation, house building material, etc. These advantageous properties of
22
r
23
24
natural fibers have attracted researchers from all over the world to develop new materials
Pe
25
26 based on them.
27
28
er
29
Considerable work by researchers have been reported in the past on natural fibers such as
30
31
32 flax [1], oil palm [2], bamboo [3], kenaf [4], sisal [5], etc. and it has been found that these
Re
33
34 natural fibers can be potential substitutes for synthetic fibers. Few works are also
35
36
available in the literature on polymer composites reinforced with palmyra fibers.
vi
37
38
39 Srinivasababuet. al. [6] studied and characterized treated and untreated palmyra fiber
ew
40
41 reinforced polyester composites based on mechanical properties such as tensile, impact
42
43
44 and flexural strength. Wu et. al. [7] characterized palmyra palm fiber based poly butylene
45
46 succinate composites on the basis of their mechanical and thermal behavior.
47
48 Velmuruganet. al. [8] investigated the mechanical properties of randomly oriented short
49
50
51 palmyra/glass fiber hybrid composites. They also studied the water absorption capacity of
52
53 these hybrid composites. Mahesh et. al. [9] studied the mechanical and thermal properties
54
55
56
57
58
59
60
1
2
3
of palmyra fiber composites with polyester as the matrix. Velmuruganet. al. [10]
4
5
6 investigated the tensile properties of epoxy composites reinforced with palmyra fiber.
7
8 A number of studies on friction and wear characteristics of natural fiber composites are
9
10
11
also reported in recent past. Chand et. al. [11] studied the influence of fiber orientation on
12
13 the wear characteristics of sisal fiber reinforced composites with epoxy as the matrix
14
15 material. It was found that the wear rate decreased considerably by addition of sisal fiber
16
17
18 to the epoxy composites. It was also observed that the wear rate was minimum when the
Fo
19
20 fibers were normally oriented and maximum wear was recorded when the fibers were
21
22 oriented in longitudinal direction. Yousifet. al. [2] studied and compared the mechanical
r
23
24
and tribological properties of oil palm fiber reinforced polyester composites (OPRP) and
Pe
25
26
27 chopped strand mat glass fiber reinforced composites (CGRP). It was found that OPRP
28
er
29
showed much better wear characteristics than CGRP. Similarly, Tong et. al. [3] studied
30
31
32 the sliding wear behavior of bamboo fiber against grey iron (HT200). It was seen that
Re
33
34 with increase in sliding velocity, the wear volume increased. It was observed that normal-
35
36
oriented fiber specimens showed better wear characteristics than parallel-oriented fiber
vi
37
38
39 specimens. Chin et. al. [4] studied the potentiality of kenaf fiber as reinforcement for
ew
40
41 application against wear. It was found that normal-oriented kenaf fiber specimens showed
42
43
44 better wear characteristics than parallel-oriented specimens. It was also observed that
45
46 presence of fiber served as a dominant factor over applied load and sliding velocity on
47
48 the specific wear rate of the specimen. The wear mechanisms of the composites were
49
50
51 predominated by micro-cracks and debonding in the fibrous regions and deformation in
52
53 the resinous regions. Tayeb [12] studied sugarcane fiber reinforced polyester composites
54
55
56
for tribological characterizations. He compared and found that sugarcane fiber reinforced
57
58
59
60
1
2
3
polyester composites are more resistant towards wear than glass fiber reinforced
4
5
6 polyester composites. The friction coefficient was also found to decrease with increase in
7
8 load for chopped and unidirectional sugarcane fibers. It was also found that the wear rate
9
10
11
decreased on increasing the fiber length up to a maximum of 5mm and then again
12
13 increased indicating the existence of a critical fiber length for minimum wear rate.
14
15 Nirmalet. al. [13] studied the effect of betelnut fiber on the adhesive wear of polyester
16
17
18 composites. A superior wear performance was observed in anti-parallel oriented betelnut
Fo
19
20 fiber followed by parallel orientation and normal orientation of the fiber. Siva et. al. [14]
21
22 studied the mechanical and tribological characteristics of woven coconut fibers and found
r
23
24
the wear resistance to be maximum when a combination of silane treated glass fiber and
Pe
25
26
27 coconut fiber was used. Chand et. al. [15] studied the effect of silane coupling agent and
28
er
29
applied load on the frictional and sliding wear behavior of sisal fiber reinforced polyester
30
31
32 composites.
Re
33
34
35 Most of the published works on palmyra fibers have reported on the strength properties of
36
vi
37
38 the composites and not on the effects of fibers on friction and wear behavior. Moreover,
39
ew
40 although a lot of work has been done on the wear characteristics of composites with other
41
42
43
natural fibers, no work has been reported so far on epoxy based composites reinforced
44
45 with short palmyra fiber. In view of this, the present work includes the investigation of
46
47 dry sliding wear behavior using a pin-on-disc set up for epoxy filled with short palmyra
48
49
50 fibers with an objective to find ways to use these locally available inexpensive fibers as a
51
52 substitute for highly expensive synthetic fibers for wear resistance applications. Taguchi
53
54 technique and Artificial Neural Network (ANN) have been used in an integrated manner
55
56
57
58
59
60
1
2
3
in this work for parametric analysis of the wear process and to predict the sliding wear
4
5
6 response of epoxy-palmyra fiber composites within and beyond the experimental limit.
7
8
9 Experimental Details
10
11
12 Composite Fabrication
13
14 In the present work Epoxy LY556 which belongs to the epoxide family has been used as
15
16
the matrix material. Epoxy LY556 is a low-temperature curing resin and it is mixed with
17
18
Fo
19 hardener HY951 in the ratio of 10:1. The short palmyra fiber (SPF) used in the present
20
21 work is extracted from the palmyra tree leaf stalk. The stalk after being cut from the tree
22
r
23
24
is soaked in water for 10 days and is beaten to separate the cellulose and lignin present in
Pe
25
26 the fiber. The fiber is then dried in the sunlight for 5 days until it dries out completely.
27
28 The fiber is then cut into 2-3mm in length with scissor. This fiber is thoroughly mixed
er
29
30
31 with the epoxy resin (LY556) in different weight proportions (0, 4, 8 and 12 wt%). The
32
Re
33 dough (epoxy and SPF) is then slowly poured into the cylindrical glass moulds (9mm dia
34
35 and 100mm length), coated beforehand with silicone-releasing agent. These are left to
36
vi
37
38 cure for 24 hours after which the moulds are broken and specimens are released.
39
ew
40
41 Sliding Wear Test
42
43
44 The samples are subjected to sliding wear test using a pin-on-disc type friction and wear
45
46 monitoring test rig as per ASTM G 99-05. The counter body is a disc made of hardened
47
48 ground steel (EN-32, hardness 72 HRC, surface roughness 0.6 µmRa). The specimen is
49
50
51 held stationary while the disc rotates. Normal load is applied through a lever mechanism.
52
53 A series of tests are conducted with four sliding velocities of 63, 125,190 and 250 cm.s−1
54
55
56
each under four different normal loads of 5, 10, 15 and 20 N. The material loss from the
57
58
59
60
1
2
3
composite surface was measured using a precision electronic balance with accuracy ±0.1
4
5
6 mg and the specific wear rate (mm3/N-m) was expressed on a volume loss basis as:
7
8 Ws= ∆m/ (ρ. t. Vs. Fn), [1]
9
10
11
where∆mis the mass loss of the composite in the test duration (g), ρ is the density of the
12
13 composite (g/mm3), t is the test duration (s), Vs is the sliding velocity (cm.s−1), and Fnis
14
15 the average normal load (N).
16
17
18
Fo
19 Taguchi Experimental Design
20
21 Taguchi experimental design is a very powerful tool to determine the degree of effect of
22
r
23
24
various parameters on the output result. Selecting the control factors affecting the
Pe
25
26 performance output in design of experiments is the most vital stage. In the initial stage, a
27
28 number of factors are considered and subsequently the less important factors are
er
29
30
31 eliminated leaving behind only the more important factors. In the present sliding wear
32
Re
33 test, four major factors have been considered such velocity, load, sliding distance and
34
35 fiber content each at four levels according to Taguchi’s L16 orthogonal array as shown in
36
vi
37
38 Table 1. The experiments are conducted at room temperature. Use of Taguchi L16
39
ew
40 reduced the number of experiments from 44= 256 conventional runs to just 16 runs
41
42
43
thereby saving a lot of experimental time and cost. These experimental observations are
44
45 again converted into signal-to-noise (S/N) ratios. The S/N ratio for minimum wear rate
46
47 can be expressed as “smaller is better” which is calculated as logarithmic transformation
48
49
50 of loss function as shown below:
51
52 S 1
53
N
= −10 log
n
(∑ y )
2
54 , [2]
55
56 wheren is the number of observations and y is the observed data.
57
58
59
60
1
2
3
4
5 Prediction Using Artificial Neural Network
6
7 Artificial Neural Network (ANN) is a very powerful tool for solving a huge
8
9 variety of problems from various fields. It uses the basic principle of a
10
11
12 biological neural system to predict the results. It was developed to simulate
13
14 the strong learning, clustering and reasoning capacity of biological neurons.
15
16
With a strong learning capability and use of parallel computation and
17
18
Fo
19 nonlinear mapping, neural networks can be successfully applied for
20
21 identifying several nonlinear systems and control problems [16]. ANN is
22
r
23
24
widely used in the field of classification and control of dynamic system,
Pe
25
26 medical science, image and speech recognition, etc. In the present study
27
28 ANN serves as a very helpful tool for predicting the specific wear rate as it
er
29
30
31 is very difficult to get a mathematical formulation for the wear behavior.
32
Re
33 This proposed approach not only yields a sufficient understanding of the

34
35 effects of process parameters but also produces an optimal parameter setting
36
vi
37
38 to ensure that the composites exhibit the best wear performance
39
ew
40 characteristics. The details of this methodology are described by Kadi[17].

41
42
43
44 RESULTS AND DISCUSSION
45
46 Sliding Wear Test Results
47
48 The specific wear rates and their corresponding S/N ratios obtained by
49
50
51 Taguchi L16 orthogonal array for all the 16 experiments are presented in
52
53 Table 2. From the Table 2 it is found that the overall mean of the S/N ratio is
54
55
56
-12.4429 dB for epoxy based composites reinforced with short palmyra
57
58
59
60
1
2
3
fibers. A commercial software Minitab 14 was used for the application of
4
5
6 design of experiment. From the S/N ratio output, it is observed that among
7
8 all the four factors considered for the analysis, the specific wear rate is
9
10
11
mostly affected by the SPF content in the epoxy composite followed by the
12
13 sliding velocity, sliding distance and normal load. The analysis showed that
14
15 at sliding velocity (A) of 63 cm/s, normal load (B) of 20 N, sliding distance
16
17
18 (C) of 1000 m and SPF content (D) of 12 wt%, the specific wear rate is
Fo
19
20 minimum as shown in Figure 1. The response table for S/N ratios is shown
21
22 in Table 3.
r
23
24
Pe
25
26
27
28
er
29
30
31
32 Artificial Neural Network for Prediction
Re
33
34
35 ANN is a very useful tool to predict the input and output pattern of nonlinear
36
vi
37 problems such as sliding wear. ANN technology can be effectively used for
38
39
solving complex, nonlinear and multidimensional problems because it is
ew
40
41
42 able to imitate the learning capability of human beings. As a result of this,
43
44 ANN could predict the results effectively from the examples without using
45
46
47 any formula and could even predict new results based on the trends
48
49 obtainedfrom the examples. In the present analysis sliding velocity, normal
50
51 load, sliding distance and SPF content are taken as the four input
52
53
54 parameters. The database is then divided into three categories, namely: (1) a
55
56 validation category, which is required to define the ANN architecture and
57
58
59
60
1
2
3
adjust the number of neurons for each layer; (2) a training category, which is
4
5
6 exclusively used to adjust the network weights; and (3) a test category,
7
8 which corresponds to the set that validates the results of the training
9
10
11
protocol. The input variables are normalized so that it lies in the range 0-1.
12
13 The output layer has one neuron which represents the specific wear rate.
14
15 Different structures of ANN were tested by varying the number of neurons
16
17
18 in the hidden layer and gradient value. Finally one structure with minimum
Fo
19
20 error was selected for training. The optimized 3-layer neural network used in
21
22 this simulation is shown in Figure 2. The selected structure was trained
r
23
24
rigorously by selecting maximum number of cycles. The Table 4 below
Pe
25
26
27 shows the parameters during the training. The predicted values are
28
er
29
compared with the experimental values and the results are shown in Table 5
30
31
32 with the associated error percentages.
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48 The Figure 3-4 below shows the effect of two most dominant factors i.e. SPF content and
49
50
51 sliding velocity on the specific wear rate of the specimens and a simulated specific wear
52
53 rate has been predicted using ANN
54
55
56
57
58
59
60
1
2
3
4
5
From the figure 3-4 it can be concluded that while the specific wear rate reduced
6
7
8 invariably with the increase in the SPF content, it increased with increase in the sliding
9
10 velocity. The figure clearly indicates an improvement in the wear resistance with increase
11
12
13
in SPF content. The increase in the wear rate with increase in the sliding wear velocity
14
15 can be attributed to the fact that with increase in the sliding velocity, the interface
16
17 temperature between the specimen and the counter surface of the pin-on-disc set-up
18
Fo
19
20 increases. Since both the fiber and the epoxy matrix has different thermal expansion
21
22 coefficient, it creates an interfacial stress. When this interfacial stress exceeds the
r
23
24 bonding strength of the fiber and epoxy, cracks starts forming and more wear takes place.
Pe
25
26
27
28 Morphology of the Specimen
er
29
30
31 Figure 5-a presents the surface morphology of a typical palmyra fiber used in this
32
Re
33 investigation. It is clearly seen that the surface is uneven and exhibits a peculiar
34
35
36
arrangement of hemispherical dots and bony ridges of different size and thickness. The
vi
37
38 surface features of epoxy SPF composite with a fiber content of 8 wt. % is shown in
39
ew
40 figure 5-b.
41
42
43
44
The worn surface morphology (sliding velocity of 63 cm/sec and normal load of 5 N) of
45
46 the same composite is presented in Figure 5-c in which local removal of the matrix layer
47
48 as a result of the dry sliding wear is visible. Figure 5-d presents the SEM image of a worn
49
50
51 composite surface at relatively higher sliding speed of 190 cm/sec. Removal of the upper
52
53 matrix layer and protrusion of fibers beneath the epoxy resin are revealed. The direction
54
55 of sliding is also clear in this micrograph. Figure 5-e presents another typical worn
56
57
58 surface of a composite with low fiber content (4 wt.%). Formation of wear debris, surface
59
60
1
2
3
cracks and signs of plastic are evident. But such features are not adequately seen in
4
5
6 Figure 5-f which presents the worn surface morphology of an epoxy-SPF composite with
7
8 fiber content as high as 12 wt.%. Signs of plastic deformation are however clearly seen in
9
10
11
the micrograph and the wear tracks, indicated by arrows.
12
13
14 CONCLUSIONS
15
16
17 1. Successful fabrication of short palmyra fiber reinforced epoxy composites are
18
Fo
19 possible and the results show that with the increase in SPF content, the sliding wear
20
21
22 rate decreases considerably.
r
23
24
Pe
25 2. The effect of various factors such as sliding velocity, normal load, sliding distance
26
27 and SPF content can be successfully determined using Taguchi L16 experimental
28
er
29
30 design. It was observed from the analysis that SPF content is the most dominant
31
32 factor followed by sliding velocity. The analysis showed normal load as the least
Re
33
34 affecting factor for specific wear rate.
35
36
vi
37
38 3. The work also illustrates the use of artificial neural network for predicting the
39
ew
40 specific wear rate of SPF composites with increase in SPF content and sliding
41
42 velocity beyond the experimental domain. The predicted and experimental values of
43
44
45 wear rate exhibited good agreement, validating the remarkable capability of a well-
46
47 trained neural network as far as the present work is concerned.
48
49
50 REFERENCES
51
52
53
54
1. Baiardo M, Zini E and Scandola M. Flax fibre–polyester composites. Composites Part
55
56 A: applied science and manufacturing 2004; 35: 701-710.
57
58
59
60
1
2
3
2. Yousif BF and TayebNSM. Mechanical and wear properties of oil palm and glass fiber
4
5
6 reinforced polyester composites.International Journal of Precision Technology 2009; 1:
7
8 213-222.
9
10
11 3. Tong J, Arnell RD and Ren LQ. Dry sliding wear behaviour of bamboo. Wear 1998;
12
13
14 221: 37–46.
15
16
17 4. Chin CW and Yousif BF. Potential of kenaffibres as reinforcement for tribological
18
Fo
19 applications.Wear 2009; 267: 1550–1557.
20
21
22
5. Bajpai PK, Singh Iand Madaan J. Tribological behavior of natural fiber reinforced
r
23
24
Pe
25 PLA composites. Wear 2013; 297: 829–840.

26
27
28 6. Srinivasababu N, Kumar JS and Reddy KVK. Manufacturing and Characterization of
er
29
30
31
Long Palmyra Palm/BorassusFlabellifer Petiole Fibre Reinforced Polyester Composites.
32
Re
33 Procedia Technology 2014; 14: 252 – 259.

34
35
36 7. Wu CS, Liao HT and Jhang JJ. Palm fibre-reinforced hybrid composites of
vi
37
38
polybutylene succinate: characterization and assessment of mechanical and thermal
39
ew
40
41 properties. Polymer Bulletin 2013; 70: 3443–3462.
42
43
44 8. Velmurugan R and Manikandan V. Mechanical properties of palmyra/glass fiber
45
46
hybrid composites. Composites: Part A 2007; 38: 2216–2226.
47
48
49
50 9. Mahesh C, Kondapanaidu B, Govindarajulu K, et al. Experimental investigation of
51
52 thermal and mechanical properties of palmyra fiber reinforced polyster composites with
53
54 and without chemical treatment and also addition of chalk powder. International Journal
55
56
57 of Engineering Trends and Technology 2013; 5: 259-271.
58
59
60
1
2
3
10. Velmurugan G, Vadivel D, Arravind R, et al. Tensile test analysis of natural fiber
4
5
6 reinforced composite. International Journal of Mechanical and Industrial Engineering
7
8 2012; 2: 2231 –6477.
9
10
11 11. Chand N and Dwivedi UK. Influence of fiber orientation on high stress wear behavior
12
13
14 of sisal fiber-reinforced epoxy composites. Polymer Composites 2007; 28: 437-441.
15
16
17 12. Tayeb NSM. A study on the potential of sugarcane fibers/polyester composite for
18
Fo
19 tribological applications. Wear 2008; 265: 223–235.
20
21
22
13. Nirmal U, Yousif BF, Rilling D, et al. Effect of betelnutfibres treatment and contact
r
23
24
Pe
25 conditions on adhesive wear and frictional performance of polyester composites. Wear

26
27 2010; 268: 1354–1370.
28
er
29
30
31
14. Siva I, Jappes JTW and Suresha B. Investigation on mechanical and tribological
32
Re
33 behavior of naturally woven coconut sheath-reinforced polymer composites. Polymer

34
35 Composites 2012; 33: 723-732.
36
vi
37
38
15. Chand N and Dwivedi UK. Sliding wear and friction characteristics of sisal fibre
39
ew
40
41 reinforced polyester composites: effect of silane coupling agent and applied load.
42
43 Polymer Composites 2008; 29: 280-284.
44
45
46
16. Padhi PK and Satapathy A. Analysis of sliding wear characteristics of BFS filled
47
48
49 composites using an experimental design approach integrated with ANN. Tribology
50
51 Transactions 2013; 56: 789-796.
52
53
54 17. Kadi HE. Modeling the mechanical behavior of fiber-reinforced polymeric composite
55
56
57 materials using artificial neural networks—a review. Composite Structures 2006; 73: 1–2.
58
59
60
1
2
3 Main Effects Plot (data means) for SN ratios
4 A B
-5
5
-10
6
Meanof SNratios
7 -15
8 -20
63 125 190 250 5 10 15 20
9 -5
C D
10
-10
11
-15
12
13 -20
250 500 750 1000 0 4 8 12
14 Signal-to-noise: Smaller is better
15
16
Figure 1.Main effects plot for S/N ratios (with short palmyra fiber)
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29 Sliding Velocity
30
31 Normal Load Specific Wear
32
Re
33 Sliding Distance Rate

34
35 SPF Content
36
vi
37
38
39
ew
40
41
42
Input Layer Hidden Layer Output Layer
43
44
Figure 2.The Three Layer Neural Network
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21 Figure 3.ANN prediction of variation in specific wear rate with SPF content for SPF-Epoxy composites.
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48 Figure 4. ANN prediction of variation in specific wear rate with sliding velocity for SPF-Epoxy
49
50 composites.
51
52
53
54
55
56
57
58
59
60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16 (a) (b)
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
(c) (d)
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49 (e) (f)
50
51 Figure 5. Typical SEM images of SPF-Epoxy composites.
52
53
54
55
56
57
58
59
60
1
2
3
4
5 Table 1. Control factors and their selected levels of dry sliding wear
6
7
8
9 Levels
10 Control factors
11
12 1 2 3 4 Units
13
14 Sliding velocity (A) 63 125 190 250 Cm/sec
15
16 Normal load (B) 5 10 15 20 N
17
18
Sliding distance (C) 250 500 750 1000 m
Fo
19
20
21 Fiber content (D) 0 4 8 12 Wt %
22
r
23
24
Table 2. Specific Wear Rates Obtained for Different Test Conditions with S/N Ratios
Pe
25
26
27
Sliding Normal Sliding SPF
28 Test Velocity Load Distance Content Sp. Wear Rate S/N Ratio
er
29 Run A B C D (10-4mm3/N-m) (dB)

30 (cm/sec) (N) (m) (wt %)
31
32 1 63 5 250 0 6.3600 -16.0691
Re
33 2 63 10 500 4 5.4690 -14.7582

34
35 3 63 15 750 8 3.6250 -11.1862
36 4 63 20 1000 12 0.9165 0.7574
vi
37
5 125 5 500 8 3.8560 -11.7227
38
39 6 125 10 250 12 2.3320 -7.3546
ew
40 7 125 15 1000 0 7.8180 -17.8619

41
8 125 20 750 4 5.8230 -15.3029
42
43 9 190 5 750 12 1.4440 -3.1913
44 10 190 10 1000 8 2.8280 -9.0296
45
46 11 190 15 250 4 6.4300 -16.1642
47 12 190 20 500 0 8.8180 -18.9074
48
13 250 5 1000 4 7.4690 -17.4652
49
50 14 250 10 750 0 9.4840 -19.5398
51 15 250 15 500 12 2.6660 -8.5172
52
16 250 20 250 8 3.6560 -11.2601
53
54
55
56
57
58
59
60
1
2
3 Table 3. Response Table for Minimum Sliding Wear for Composite with Short Palmyra Fiber
4
5 Level A B C D
6
1 -10.314 -12.112 -12.712 -18.095
7
8 2 -13.061 -12.671 -13.476 -15.924
9 3 -11.823 -13.432 -12.305 -10.800
10
11 4 -14.196 -11.178 -10.900 -4.576
12 Delta 3.882 2.254 2.577 13.518
13 Rank 2 4 3 1
14
15
16
17 Table 4. Input Parameters Selected During Training
18
Fo
19
Input Parameters Values
20
21 Number of input layer neuron (I) 4
22
r
23 Number of hidden layer neuron (H) 10
24 Number of output layer neuron (O) 1
Pe
25
No of epochs 100000
26
27 Gradient value 1.00 e-14
28 Validation check 100000
er
29
30
31
32 Table 5. Comparison of Experimental Results with ANN Predicted Values
Re
33
34 Specific wear rate (10-4 mm3/N-m)
35
36 Test Run Experimental ANN Predicted Error %
vi
37 1 6.360 6.359136 0.013

38 2 5.469 5.468302 0.012
39
ew
40 3 3.625 3.601238 0.655

41 4 0.917 0.979166 6.837
42
5 3.856 4.343180 12.634
43
44 6 2.332 2.328292 0.159
45 7 7.818 7.773308 0.571
46
8 5.823 5.818631 0.075
47
48 9 1.444 1.443930 0.004
49 10 2.828 2.805736 0.787
50
51 11 6.430 6.425846 0.064
52 12 8.818 8.119834 7.917
53 13 7.469 7.382000 1.520
54
55 14 9.484 9.433750 0.529
56 15 2.666 2.654097 0.446
57
16 3.656 3.660072 0.111
58
59
60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Bharatiya Jyotish Sastra-1

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Bharatiya Jyotish Sastra-1

Загружено:

Авторское право:

Доступные форматы

Journal of Composite Materials

ANALYSIS AND PREDICTION OF SLIDING WEAR RESPONSE

Manuscript Type: Original Manuscript

Date Submitted by the Author: 27-Nov-2015

Epoxy composites, Palmyra fiber, Sliding wear, Taguchi method, Neural

This paper presents a parametric analysis of dry sliding wear process

short palmyra fibers (SPF). It is an attempt to explore the possibility of

29 epoxy by reinforcing it SPF. Palmyra fiber is an inexpensive, strong and naturally

33 This proposed approach not only yields a sufficient understanding of the

40 characteristics. The details of this methodology are described by Kadi[17].

25 PLA composites. Wear 2013; 297: 829–840.

33 Procedia Technology 2014; 14: 252 – 259.

25 conditions on adhesive wear and frictional performance of polyester composites. Wear

33 behavior of naturally woven coconut sheath-reinforced polymer composites. Polymer

14 Signal-to-noise: Smaller is better

33 Sliding Distance Rate

29 Run A B C D (10-4mm3/N-m) (dB)

33 2 63 10 500 4 5.4690 -14.7582

40 7 125 15 1000 0 7.8180 -17.8619

37 1 6.360 6.359136 0.013

40 3 3.625 3.601238 0.655

Вам также может понравиться