You are on page 1of 10

Multi response optimization of sheet forming of Kenaf-Polypropylene composites using

grey based fuzzy algorithm


Erfina Oktariani, Rita Istikowati, Hendro Sat Setijo Tomo, Rafliansyah Rizal, and Yosea Pratama

Citation: AIP Conference Proceedings 1931, 030001 (2018); doi: 10.1063/1.5024060


View online: https://doi.org/10.1063/1.5024060
View Table of Contents: http://aip.scitation.org/toc/apc/1931/1
Published by the American Institute of Physics

Articles you may be interested in


Preface: The 3rd International Conference on Industrial Mechanical, Electrical, and Chemical Engineering
AIP Conference Proceedings 1931, 010001 (2018); 10.1063/1.5024054

Potential ability of zeolite to generate high-temperature vapor using waste heat


AIP Conference Proceedings 1931, 020001 (2018); 10.1063/1.5024055

Extraction of pectin from passion fruit rind (Passiflora edulis var. flavicarpa Degener) for edible coating
AIP Conference Proceedings 1931, 030002 (2018); 10.1063/1.5024061

Green infrastructure and urban sustainability


AIP Conference Proceedings 1931, 020002 (2018); 10.1063/1.5024056

Study on the carboxymethylation of glucomannan from porang


AIP Conference Proceedings 1931, 030005 (2018); 10.1063/1.5024064

Non-slag co-gasification of biomass and coal in entrained-bed furnace


AIP Conference Proceedings 1931, 020003 (2018); 10.1063/1.5024057
Multi Response Optimization of Sheet Forming of Kenaf-
Polypropylene Composites Using Grey Based Fuzzy
Algorithm
Erfina Oktariani1, a), Rita Istikowati2, b), Hendro Sat Setijo Tomo3, c), Rafliansyah
Rizal1, d) and Yosea Pratama1, e)

1
Department of Polymeric Chemical Engineering, Polytechnic of STMI Jakarta, Jakarta 10510, Indonesia
2
Department of Garment Production Engineering, Community Academy of Textile and Textile Product Industry of Surakarta,
Surakarta 57126, Indonesia
3
Laboratory of Technology of Polymer, Agency for the Assessment and Application of Technology, Tangerang 15134, Indonesia

a)
Corresponding author: erfina@kemenperin.go.id
b)
istikowati@yahoo.com
c)
satsetijo@yahoo.com
d)
rizaltk12@gmail.com
e)
osea.tama@yahoo.com

Abstract. Composites from natural fiber reinforcement are developed as the alternative sheet materials of plastic composite
for small-size bodywork parts in automotive industries. Kenaf fiber is selected as the reinforcement of plastic composite.
Press forming of Kenaf-Polypropylene is experimentally produced in this study. The aim of this study is to obtain the
optimal factor of the process of sheet forming of Kenaf-Polypropylene. The Kenaf delignified is cut into 5 cm lengths and
distributed on the surface of Polypropylene sheet for 3 and 5 ply layers. The layers of Kenaf-Polypropylene then pressed
by hot press at 190 and 210C, 40 and 50 bar, for 3 and 5 minutes. However, there are limitations in handling multi
responses in design of experiments. The application of the fuzzy logic theory to the grey relational analysis may further
develop its performance in solving multi-response problems for process parameter optimization. The layer of Kenaf and
Polypropylene, temperature, the duration of hot press and pressure are factors that affect the process. The result of
experimental investigation and as well as analysis, shows that the best combination values were 3 ply layer, 210ºC, 5
minutes of hot press and 50 bar.

INTRODUCTION

Environmental and economic concerns are stimulating the development of new materials for furniture,
construction, packaging and automotive industries.1 The materials development for automotive industries has been
started from the very earlier one (wood) to the latest and most sophisticated modern materials.2 Numerous studies
confirm that composite materials are lightweight, fatigue resistant and easily molded to any shape.2 In other words,
composite materials can be considered as substitute for metal materials. A composite material from polymer matrix
imbedded with high-strength fibers (glass, aramid and carbon) is known as fiber reinforced polymer (FRP).3 Natural
fibers as alternative reinforcement in polymer composite have been more developed rapidly due to their advantages
over synthetic fibers.3
Natural fiber reinforced polymer composites have been developed and employed in some automotive industries.
Honda as one of large automotive manufacturers in the world has utilized natural fiber (wood) reinforced composites

The 3rd International Conference on Industrial, Mechanical, Electrical, and Chemical Engineering
AIP Conf. Proc. 1931, 030001-1–030001-9; https://doi.org/10.1063/1.5024060
Published by AIP Publishing. 978-0-7354-1623-9/$30.00

030001-1
on the floor of Sport Utility Vehicle (SUV). BMW Group is considering to utilize mixed natural fibers from Kenaf
and linen in the door panel for Saturn L 300S.4 Other automotive industries in United States have developed natural
fiber reinforced composites for big size car components such as door, floor, sliding door filler, etc. Toyota is
developing to employ Kenaf as reinforced composite for body cases of cars.5
This study develops natural fiber reinforced composite as the alternative sheet materials of plastic composite for
small-size bodywork parts in automotive industries. Kenaf fiber is selected as the reinforcement of plastic composite.
Polypropylene is utilized as the polymer matrix. Furthermore, composite has some mechanical properties such as
tensile and flexural strength for as quality characteristics. The objective of this study is to obtain the optimal factor of
the process of Kenaf-Polypropylene sheet forming. Engineering judgment, regression analysis, Principal Component
Analysis (PCA), and Grey Relational Analysis (GRA) can be used as method to solve multi response problems.6
Engineering judgment uses expert opinion to solve multi responses problem. Regression analysis uses process
parameter as input variables and multi characteristics responses as output variables. PCA transforms original multi
responses into key components that are not correlated. GRA converts multi responses to the value of Grey Relational
Grade (GRG) and optimizes the parameters by analyzing the GRG. GRA, based on the theory of Grey System, can be
applied to solve problems of multi quality characteristics.7 Design of Experiment (DOE) based the Grey Relational
Analysis (GRA) offers a different approach from the traditional single response optimization for multi response
optimization processes.
The analysis based on fuzzy-logic is commonly used for vague and uncertain environment such as in optimization
manufacturing process. Fuzzy logic based is used in multi-criteria decision making approaches.8 Fuzzy logic is
integrated with other optimization methodologies to improve fuzzy logic capability. GRA has strong potential to
further enhance the capability of fuzzy-logic in multi-objective optimization problems. Here, the optimization of
multiple response characteristics can be effectively transformed into optimization of single grey fuzzy reasoning grade
(GFRG).

RESEARCH METHODS

Experimental Method
Raw Kenaf fibers were supplied by Balai Penelitian Tanaman Pemanis dan Serat (BALITTAS), Bojonegoro, East
Jawa. At first, Kenaf fibers were cut and soaked in solution of NaOH 5% (wt) for 24 hours. The fibers were then
neutralized with water and dried in an oven at 60-80C for 12 hours. Dried fibers were chopped into  5 cm and then
stored in a drying box. Polypropylene for extrusion applications was selected in this experiment (HE2.0TF). Collin E
30 M extruder, a single type of screw, was selected to make polypropylene sheet. Kenaf fiber has different character
with polypropylene. Therefore, a kind of coupling agent that is polypropylene graft maleic anhydride (PP-g-MA) was
employed. Polypropylene and PP-g-MA 5% (wt) were fed into the extruder for polypropylene sheet forming.
Kenaf fibers and polypropylene sheets were compiled into layers and placed in the hot press machine. Hot press
machine (Collin P 300 P) was selected to form the sheet of Kenaf-polypropylene composite. The layer of Kenaf and
polypropylene, temperature, pressing time, and pressure influence the forming process. Each parameter was designed
to have two different level, which are shown in Table 1.
TABLE 1. Experiment level factor
Control Factor Level
Coded Uncoded 1 2
A Layer 3 ply 5 ply
B Temperature 190C 210C
C Pressing time 3 min 5 min
D Pressure 40 bar 50 bar

According to a full factorial design, experimental design for two factors with two levels requires 16 experiments.
Different experiment showed different level combination of control factor as shown by Table 2. Every experiment
was conducted with five repetitions (M1, M2, M3, M4, M5) to measure the different of variance that arises. Therefore,
matrix of experiments for design of experiment as shown on Table 2.

030001-2
TABLE 2. Matrix of Experiments
Control factor
Experiment
A B C D
1 1 1 1 1
2 1 1 1 2
3 1 1 2 1
4 1 1 2 2
5 1 2 1 1
6 1 2 1 2
7 1 2 2 1
8 1 2 2 2
9 2 1 1 1
10 2 1 1 2
11 2 1 2 1
12 2 1 2 2
13 2 2 1 1
14 2 2 1 2
15 2 2 2 1
16 2 2 2 2

Optimization Method
The ambiguity of a system model and deficiency of information can be analized by GRA. It can create discrete
sequences for the correlation analysis of such sequences with processing uncertainty, multi-factors and discrete data.
In the GRA, the first step was to perform the normalization of experimental data to make the range within 0 to 1. This
step was called grey relational generating. According to the importance of quality characteristics, this can be divided
into three criteria for optimization in GRA, namely “larger-the-better,” “smaller-the-better,” and “nominal-the-best”.8
Quality characteristic, tensile and flexural strength, are “larger- the-better”. The expectancy is larger-the-better, then
it can be normalized with value of grey relation with the following Eq. (1).

(1)

Where is the normalized decimal value, is mean value of each response, is target value and
and are the largest and smallest values , i = 1,2,3,…,12.
The larger normalized results should be expected, therefore; ideal (best) normalized result should be equal to one.
In the next step, grey relational coefficient (GRC) is calculated to express the relationship between the ideal (best) and
the actual normalized experimental results the grey relational coefficient is calculated by using the following Eq. (2).
∆min ∆max
ξ k (2)
∆ ∆max
Where, ∆ k is the absolute difference between and , Δmin and Δmax are minimum and maximum values
of the absolute differences of all comparing sequences, respectively. ξ is the distinguishing or identification coefficient
whose value lies from 0 to 1. Most of the distinguishing coefficient is assumed as 0.5 to fit the practical requirements.
Each response in GRA was characterized its quality characteristics. The results of the characterization indicated
some level of uncertainty. This kind of uncertainty level could be efficiently investigated by fuzzy-logic technique.
Thus, complex multi-responses optimization problem can be solved by combining fuzzy-logic and GRA techniques.
Fuzzy-logic technique consists of a fuzzifier, membership functions, fuzzy rule base, inference engine and defuzzifer.
The steps of grey-fuzzy-logic method were illustrated in Figure 1, which can be described as follows:
1. The experimental values of coefficient of tensile strength and flexural strength were normalized in between 0 and
1.
2. Grey relational coefficient (GRC) was calculated for each mean response.
3. Then fuzzy-logic technique was applied. In this technique, the fuzzifier used the membership functions to fuzzify
the Grey relational coefficient (GRC) of each response.
4. Fuzzy rules (if-then control rules) can be generated and finally fuzzy predicted value converted into Grey Fuzzy
Reasoning Grade (GFRG) by using defuzzifier.

030001-3
Measurement for tensile strength
and flexural strength

Normalization (grey relation


generating)

Estimate Grey Relation


Coefficient (GRC)

Fuzzifier Membership
Function
Fuzzy
Fuzzy Knowledge
Determine DFRG
inference Based
Fuzzy Rule
Get Optimal Control Factor Defuzzifier
Level for GFRG

FIGURE 1. Steps of grey-fuzzy-logic method

Out of these elements, the fuzzifier uses membership functions to fuzzify the GRC of input parameters, (X1 = grey
relation coefficient for tensile strength, X2 = grey relation coefficient for flexural strength). Membership function was
used to map the values of inputs (X1 and X2) and output (Y1=GFRG) parameters in the range of 0 to 1. Fuzzy reasoning
can be performed by the inference engine to generate a fuzzy value of the fuzzy rules. The rules writing procedure can
be described as follows. In this experimental work twenty fuzzy rules can be written for two inputs and one output are
developed for inference.
Rule 1: if X1 is A1; and X2 is B1; then Y is C1; else
Rule 2: if X1 is A2; and X2 is B2; then Y is C2; else
Rule n: if X1 is An; and X2 is Bn; then Y is Cn
Ai, Bi, and Ci are fuzzy subsets as per the corresponding membership functions

RESULTS AND DISCUSSION


Two responses, tensile strength and flexural strength, were measured as quality characteristics from every
experiment. The experimental results of two responses variables were summarized by Table 3. The results showed
that tensile and flexural strength of each repetition from M1 to M5 has ±5% deviation. Therefore, these experimental
results required further processing.
Fig. 2 shows the Scanning Electron Microscope (SEM) pictures of cross section specimens of 3 ply, 5 min, 40 bar,
210C and 5 ply, 5 min, 40 bar, 210C. The pictures shows that there is no gap around fibers which means Kenaf
fibers have been linked well with polypropylene matrix. This condition is agree with the purposes of PP-g-MA
employment as coupling agent.
The pre-processed data of experimental results, the grey relational coefficients and the overall grey relational grade
for each of the combination of parameters is given in Table3. For all the responses, ‘larger the better’ criterion is
preferred. All sequences after application the data preprocessing through Eq. (1) and Eq (2) are shown in Table 4.
Result of data pre-processing shown that experiment 6 is the best result for tensile strength and experiment 8 is the
best result for flexural strength.

030001-4
TABLE 3. Experimental Results
Tensile Strength (MPa) Flexural Strength (MPa)
Experiment
M1 M2 M3 M4 M5 M1 M2 M3 M4 M5
1 27.48 33.81 28.37 40.08 41.51 70.55 77.11 63.03 75.69 76.28
2 27.21 29.98 49.63 54 35.6 56.96 34.82 51.55 68.92 50.33
3 36.38 47.61 47.94 42.72 43.85 34.47 36.42 47.88 54.78 51.63
4 22.85 20.56 12.53 15.86 23.49 66.68 77.66 76.04 87.50 78.14
5 34.1 38.43 42.82 30.55 21.26 37.02 42.67 39.30 51.58 45.12
6 65.81 64.16 58.52 65.04 56.26 78.74 69.10 44.83 62.28 66.76
7 32.93 28.27 21.38 25.16 24.83 75.40 76.62 59.93 79.14 61.63
8 36.73 45.44 40.36 38.99 36.07 78.29 85.51 70.19 73.93 79.13
9 38.93 50.83 54.59 58.96 61.46 48.24 49.43 50.34 64.34 62.17
10 32.79 46.25 61.79 50.86 41.61 55.70 43.99 55.68 51.34 65.11
11 47.22 40.14 35.18 29.17 40.3 65.69 62.23 71.52 59.33 62.98
12 49.58 44.42 48.71 47.84 42.73 60.88 43.47 55.27 41.98 49.80
13 51.24 51.93 40.22 66.92 28.83 50.39 39.82 48.14 41.05 42.19
14 34.62 42.71 40.88 27.76 25.5 85.43 58.88 32.83 93.25 74.09
15 34.12 37.24 30.46 26.19 28.59 52.28 51.42 52.43 58.74 54.46
16 19.57 17.3 30.07 34.3 26.04 56.96 34.82 51.55 68.92 50.33

a) 3 ply, 5 min, 40 bar, 210C b)5 ply, 5 min, 40 bar, 210C


FIGURE 2. Representative SEM images of specimen with different parametric combinations.

030001-5
TABLE 4. Normalized, GRC and GRG value
Mean Normalized Value GRC
Tensile Flexural
Experiment Tensile Flexural Tensile Flexural GRG
Strength Strength
Strength Strength Strength Strength
(MPa) (MPa)
1 34.25 72.532 0.3541 0.8577 0.4363 0.7784 0.6074
2 39.284 52.516 0.4715 0.2736 0.4861 0.4077 0.4469
3 43.7 45.036 0.5744 0.0554 0.5402 0.3461 0.4432
4 19.058 77.204 0.0000 0.9940 0.3333 0.9881 0.6607
5 33.432 43.138 0.3351 0.0000 0.4292 0.3333 0.3813
6 61.958 64.342 1.0000 0.6187 1.0000 0.5673 0.7837
7 26.514 70.544 0.1738 0.7997 0.3770 0.7139 0.5455
8 39.518 77.41 0.4769 1.0000 0.4887 1.0000 0.7444
9 52.954 54.904 0.7901 0.3433 0.7043 0.4323 0.5683
10 46.66 54.364 0.6434 0.3276 0.5837 0.4265 0.5051
11 38.402 64.35 0.4509 0.6189 0.4766 0.5675 0.5220
12 46.656 50.28 0.6433 0.2084 0.5836 0.3871 0.4854
13 47.828 44.318 0.6706 0.0344 0.6029 0.3412 0.4720
14 34.294 68.896 0.3552 0.7516 0.4367 0.6681 0.5524
15 31.32 53.866 0.2858 0.3130 0.4118 0.4212 0.4165
16 25.456 52.516 0.1491 0.2736 0.3701 0.4077 0.3889

Two inputs and one output (GFRG) fuzzy-logic system is used. Grey relational coefficients for tensile and flexural
strength are inputs to the fuzzy logic system. The linguistic membership function for instance Lowest, Low, Medium,
High and Highest are used to represent the grey relational coefficients (GRC) of input variables. Likewise the output
grey relational grade is being represented by the membership functions such as Tiny, Very Small, Small, Medium,
Large, Very Large, Huge. The membership functions of the input and output are indicated in Figure 3.

a) Membership Function of Input b) Membership Function of Output


FIGURE 3. Membership function of input and output

The inference engine (Mamdani fuzzy inference system) performs fuzzy reasoning with fuzzy rules for generating
a fuzzy value. A total of 25 numbers of fuzzy rules are used for this purpose. These fuzzy rules are shown in the form
of ‘if–then’ control rule as shown Figure 4. Maximum–minimum compositional operation by tracking the fuzzy
reasoning yields a fuzzy output. At last, the defuzzifier converts the fuzzy predicted values into a GRFG by using
MATLAB (R2016a) fuzzy logic toolbox. This GFRG values are tabulated in Table 5.

030001-6
FIGURE 4. Fuzzy logic rules viewer

TABLE 5. GFRG
Experiment GFRG Rank
1 0.6564 4
2 0.4555 13
3 0.4651 12
4 0.7284 3
5 0.4200 16
6 0.8400 1
7 0.5680 7
8 0.8289 2
9 0.5812 5
10 0.5252 9
11 0.5497 8
12 0.5077 10
13 0.4997 11
14 0.5629 7
15 0.4575 14
16 0.4575 15
The performed experiment design as shown by Table 5 demonstrates that the parameters setting of 6 has the highest
GFRG. In addition, the average of the GFRG for each level of the parameters setting is summarized and given in Table
6. The difference between maximum and minimum value of GFRG of parameters is also calculated and provided in
Table 6. The percentage contribution is shown by the highest rank. Number of layer will give more contribution for
quality characteristic.
Additionally, analysis of variance (ANOVA) was conducted to determine the effect of process parameters on the
grey fuzzy relational grade. The significant factors are identified with the help of the analysis of variance (ANOVA).
ANOVA analysis was performed with a 95% confidence level and 5% significance level. F values of the control
factors indicate the significance of control factors with ANOVA analysis. As listed on Figure 5, ANOVA analysis
was carried out to see the parameters’ effects on the experimental result. The results of the ANOVA indicated that all
factors are not statistically significant to the GFRG values at the 95% reliability level. P value results which are higher
than 0.05 for each other, are the reason for those results.

030001-7
TABLE 6. Response table for GFRG.
Factor Level Delta Rank
Coded Uncoded 1 2
A Layer 0.6117 0.4986 0.1131 1
B Temperature 0.5568 0.5535 0.0033 4
C Pressing Time 0.5590 0.5512 0.0078 3
D Pressure 0.5097 0.6006 0.0909 2
Total average value 0.5551

FIGURE 5. ANOVA of GFRG

CONCLUSION
The current research work utilized hybrid optimization technique using GRA and fuzzy-logic method for
simultaneously optimizing characteristics of Kenaf-Polypropylene (tensile strength and flexural strength). This
technique directly combined the multiple responses into a single performance characteristic called grey fuzzy
relational grade. Based on the above experimental investigation as well as analysis, it was found out that the best
combination values were 3 ply, 210ºC, 5 minutes pressing and 50 bar (experiment 6). They were the recommended
levels of the controllable parameters concurrently considered. Analysis of variance (ANOVA) was utilized to
determine the effect of factors on GFRG. According to the results of ANOVA for GFRG, all factors had no statistically
significance on the GFRG values, but pressure had the highest % contribution. The Grey integration based fuzzy
algorithm combined with full factor on the experimental design can be applied in the multi criteria optimization.
Therefore, the multi response conversion of quality characteristics data into a single GFRG can be effectively
employed to optimize the factor level for sheet forming of Kenaf-Polypropylene composites.

ACKNOWLEDGMENTS

This study was partially supported by “Sarana Penelitian Industri Terapan (SPIRIT)” FY 2017, Pusdiklat Industri,
Ministry of Industry, Republic of Indonesia. The experiment in this study take placed in Laboratory of Technology of
Polymer, Agency for the Assessment and Application of Technology, Tangerang, Indonesia.

REFERENCES

1. C. Alves, P. M. C Ferrao, A. J. Silva, L. G. Reis, M. Freitas, L. B. Rodrigues and D. E. Alves, Journal of Cleaner
Production 18, 313-327 (2010).
2. E. E. Tiguh, “Finite Element Analysis of Internal Door Panel of a Car by using Bamboo Fiber Reinforced Epoxy
Composite”, Master thesis, Addis Ababa University, 2016.
3. H. Ku, H. Wang, N. Pattarachaiyakoop and M. Trada, Composites: Part B 42,856-873 (2011).

030001-8
4. L. Mohammed, M. N. M. Ansari, G. Pua, M. Jawaid and M. S. Islam, International Journal of Polymer Science,
ID 243947 (2015).
5. J. Holbery and D. Houston, JOM 58 (11), 80-86 (2006).
6. R. Jeyapaul, P. Shahabudeen and K. Krishniah, International Journal Advanced Manufacturing Technology 26,
1331-1337 (2005).
7. Y. M. Chiang and H. H. Hsien, Computer and Industrial Engineering Journal 56: 648-661 (2009).
8. S. Dewangan, S. Gangopadhyay and C. K. Biswas, Engineering Science and Technology: an International
Journal 18, 361-368 (2015).

030001-9

You might also like