Vishal
Fegade
SVKM's
NMIMS Mukesh Patel School of Technology Management & Engineering,
Mechanical Engineering Shirpur, India
E-mail: vishalfegade@gmail.com
Kshitij
Shrivastava
Indian
Institute of Technology Kharagpur, Ocean Engineering and Naval Architecture, India
E-mail: krshrivastava@gmail.com
A. V.
Kale
Yeshwantrao
Chavan College of Engineering, Mechanical Engineering, India
E-mail: avkale@ycce.edu
R. L.
Shrivastava
Yeshwantrao
Chavan College of Engineering, Mechanical Engineering, India
E-mail: rlshrivastava@gmail.com
Submission: 12/22/2018
Accept: 3/13/2019
ABSTRACT
The tremendous advancement in technology, productivity and improved standard of living has come at the cost of environmental deterioration, increased energy and raw material consumption. In this regard, remanufacturing is viable option to reduce energy usage, carbon footprint and raw material usage. In this manuscript, using computational intelligence techniques we try to determine the feasibility of remanufacturing in case of roller bearings. We collected used N308 bearings from 5 different Indian cities. Using Fuzzy-TOPSIS, we found that the roundness, surface roughness and weight play a vital role in design for remanufacturing of roller bearings. Change in diameter, change in thickness and change in width showed minimal influence. We also used Taguchi analysis to reassess the problem. The roundness of inner and outer race was found to be the most influential parameters in deciding the selection of bearing for remanufacturing. The results suggest the bearing designer to design the bearing in such a way that roundness of both races will be taken cared while manufacturing a bearing. However, using Taguchi the weight of the rollers was found to be of least influence. Overall, the predictions of Taguchi analysis were found to be similar to Fuzzy-TOPSIS analysis.
Keywords: Taguchi analysis; Fuzzy-TOPSIS; remanufacturing
1.
INTRODUCTION
In earlier day’s
manufacturers used to ignore the used products and these scraps would
eventually pose a landfill problem. However, due to recent stringent
environmental rules manufacturer has to consider the end of life strategies for
recycling of product (DINDARIAN et al., 2012). Sustainable remanufacturing is a
critical tool required to improve the efficiency of a product recovery.
To have the necessary
efficient and flexible remanufacturing system, part information and actual
condition of product become vital at end of life. For efficient
remanufacturing, challenges are the collection of cores, flexibility in process
and redistribution of products (LIU, et al., 2016). Selection of components in
remanufacturing always have some constraints with a specific objective, again
at the time of selection one cannot estimate the consequences of the process
accurately (PUROHIT; RAMACHANDRAN, 2015).
Remanufacturing is a
systematic process to bring the product back into function. It includes sorting
and inspection of components followed by disassembling the components for
reprocessing (FEGADE; SHRIVATSAVA; KALE, 2015). Parts disassembled can be
replaced if it cannot be repaired or reprocessed to meet the required quality
and functionality.
Remanufacturing is
beneficial to the environment as it diverts the scrap components from a
landfill. It also requires less energy and material as compared to new
component manufacturing (KENNE; DEJAX; GHARBI, 2012). Among all end of product
life recycling strategies, remanufacturing is perhaps the most potent one as
component returned through remanufacturing has specification nearly same as a
new one.
Additionally, it saves
time, material and energy imparted over the product. Higher quality of product
remanufacturing provides longer life extension and viable products.
Remanufacturing is a comprehensive industrial process by which a previously
sold, worn, or non-functional product or module is returned to a “like-new” or
“better than new” condition and warranted in performance level and quality
(AYDIN; KWONG; JI, 2015).
Remanufacturing, though
still an evolving industry, holds many ecological and economic benefits.
Remanufacturing is the complete or partial reconstruction or overhaul of a
product to the original stipulation of reusable and repairable parts and
replacing some completely worn out parts by new ones (BENKHEROUF; SKOURI;
KONSTANTARAS, 2016).
Product recovery is to
retrieve a product when the product no longer fulfills desired needs. Product
recovery involves recycling, reuse and remanufacturing (HILGER; SAHLING;
TEMPELMEIER, 2016). Integration of remanufacturing in closed loop supply chain
as a significant end of life product recovery system could potentially generate
benefits to both the business and the environment. For any company which
manufactures and sell both new and remanufactured products, optimizing the
product design is very difficult.
The design decisions
taken at early stages have more influence on profits in manufacturing and
remanufacturing (RAMACHANDRAN; AGARWAL, 2017). Remanufacturing has
environmental benefits as well as significant potential to influence product
economy in reverse logistic. Remanufacturing begins with identification and
inspection of cores (scrap products) further disassembly, reconditioning,
assembly and testing (ZHANG et al., 2011).
Inspections of cores
are the critical activity which leads to the effectiveness of the
remanufacturing. Remanufacturing is a generic term for technical renovation or
engineering activities of waste products. In remanufacturing, a large number of
inspections and evaluations of the failure conditions of the parts have to be
done, which are uneconomical and inefficient (SUGUMARAN; RAMACHANDRAN, 2007).
After the inspection
and selection of the parts, disassembly is the process which gives large impact
in any remanufacturing system (FORD; DESPEISSE, 2016). It depends upon the
volume of the returned product as volume increases this process becomes more
significant because the efficiency of remanufacturing depends upon the
effectiveness of the disassembly process (KWAK; KIM, 2015; SUBRAMANIAN;
FERGUSON; TOKTAY, 2013).
Selection of the
components for remanufacturing is largely affected by the disassembly process.
Parameters which influence disassembly of the component are also useful
criteria for the selection of the components for remanufacturing (HATCHER;
IJOMAH; WINDMILL, 2011).
In this paper, focus is
on the selection and optimization of suitable components for remanufacturing.
Optimization may be defined as the minimization of unnecessary traits and
maximization of necessary ones, to find the most effective and highest
attainable performance (SAHARE et al., 2018). In this study, different
parameters were considered which affects the viability in remanufacturing as
well as impacts the recovery of a product during remanufacturing.
Powerful and robust
computational intelligence techniques like Fuzzy-TOPSIS and Taguchi were used
in this research for selection of bearings for remanufacturing. Fuzzy-TOPSIS is
a perfect blend of multi-criteria decision methods and fuzzy set theory which
handles uncertain and incomplete information (MANICKAM, 2016).
Taguchi method is used
to decrease the sensitivity of engineering designs towards ungovernable noise
or factors (PANDEY; DUBEY, 2012). Taguchi method uses S/N ratio which shows
experimental data quality characteristics for better optimization results.
2.
HYBRID FUZZY-TOPSIS & TAGUCHI
OPTIMIZATION
Fuzzy refers to the
situation whose set of activity boundaries are not well defined. TOPSIS is a
widely adopted MCDM technique to solve multiple-criteria decision-making
problems in various fields (HAMDAN; SARHAN; HAMDI, 2012). Fuzzy-TOPSIS is
domain independent and thus may be applied virtually to any problem. Several
research groups like Shiraz, Mirac, Sengul and coworkers (SENGUL et al., 2015; SHIRAZ et al.,
2014; BAKIRCI et al. 2014; SENGUL et al., 2016; EREN, 2018; SHIVAKOTI et al.,
2017; DIYALEY et al., 2017; DAS et al., 2017; RAGAVENDRAN et al., 2018;
AIKHUELE et al., 2017; AIKHUELE et al., 2017; AZIZI et al., 2015) has regularly
used Fuzzy-TOPSIS or its variants in a wide range of problems.
Taguchi Analysis is
implemented in four steps which are – (i) Design of
Experiments (DOE) (ii) selection of model (iii) analysis of responses (iv)
desirability function analysis. The traditional experimental design results in
a very large number of experiments. DOE endeavors to plan systematic conduction
of experiments to acquire data in an intelligent and controlled manner with
minimum efforts.
The process can be
divided broadly into three parts as System; Input Factors and Responses. The
system can be considered as the heart of the process (KIVAK; SAMTACS; CICEK,
2012). Input factors are variable signals which serve as starting mechanisms of
the process. With the help of Fuzzy-TOPSIS, input factors such as various
diameters of outer race, inner race and roller diameter combinations were
finalized and utilized in Taguchi analysis.
The response is nothing
but the performance output of the system. In this case, responses were outer
race surface roughness, inner race surface roughness, roller surface roughness,
and outer race roundness, inner race roundness, roller roundness, outer race
weight, inner race weight and roller weight.
Data obtained from DOE
provides the information necessary to establish the relationship between the
specified input factors and the responses of the given process. The possible
values of input factors are termed as levels. The selection of the input
factors, their levels and responses are the most important and critical stage
in the analysis. In DOE when all possible combinations of given input factor
levels are considered, it is termed as a full factorial design (FFD) (GUNAY;
YUCEL, 2013).
However, in this study
only existing product combinations of bearings were considered for analysis. An
empirical model is selected to establish the relationship that exists between
the design input factors and the response. A regression model is selected here
and it also referred to as the main effect model. The main effect model needs a
smaller number of experiments than its extension. Each experiment corresponds
to a set of responses.
The response values
acquired from comparatively few experiments enables response prediction for
FFD. As this is a multi-response optimization problem, a popular simultaneous
optimization approach is employed. The optimal solutions were arranged in the
descending order of their combined desirability value. This feature of Taguchi
analysis can be used to acquire multiple optimal solutions. In this paper
Design-Expert version 11 software was used for implementation of Taguchi
Analysis.
3.
ROLLER BEARINGS SELECTION
The development of
roller bearings and its components has been a research objective for more than
five decades (WEINZAPFEL; SADEGHI, 2009). Roller bearings technically comprise
of four components namely outer ring, inner ring, roller elements and cage (LU
et al., 2013). We selected N308 bearing manufactured by Austin Engineering
Company, Gujarat, India for our analysis. We collected two used bearings each
from different cities having different climatic conditions namely Mumbai,
Delhi, Guwahati, Jaipur and Shimla for understanding the behavior of bearings
with climatic conditions. The specifications of N308 bearings are bore diameter
(d) 40 mm, outer diameter (D) 90 mm, width (B) 23 mm and weight (g) 662 grams.
Fig. 1 shows the different parts of the N308 Bearings.
Figura 1:
N308 Bearing Parts
Using Fuzzy-TOPSIS, we observed that
the roundness, surface roughness and weight play a vital role for design for
remanufacturing of roller bearings. Change in diameter, change in thickness and
change in width were the least preference (SELVARAJ; MARAPPAN, 2011; SERVAIS;
DUQUENNE; BOZET, 2013).
We carried out roundness test in
Carl Zeiss Rondcom 41C. Surface roughness was checked
in Carl Zeiss Surfcom S130. The weight was measured
using mLabs_SF400C, ASIN: B07B1XKKGW capacity 600g/10mg readability 0.01g/
minimum weight 0.05g, weight balance. We measured inside diameter of outer race
and outside diameter of inner race and the roller diameter was taken as a mean
of all the rollers diameters in the bearings. We used the three-point technique
to measure the diameters with a coordinate measuring machine (GmbH Carl Zeiss
Contour G2, Capacity: Size 700 X 700 X 600 mm).
4.
TESTING PARAMETERS SELECTION
We used Fuzzy-TOPSIS analysis for
testing parameter selection. Linguistic variables R1, R2, R3, R4, R5, R6 were
assigned to various impacts of testing, like remanufacture ability, quality,
level of integration, cost saving, End of life condition and durability.
Similarly, P1, P2, P3, P4, P5, and P6 were assigned for the testing parameters.
The linguistic variables and testing parameters are reported in Table 1.
Table 1: Linguistic variables and
testing parameters
Requirements |
Description |
Parameters |
Description |
R1 |
Remanufacture ability |
P1 |
Surface Roughness |
R2 |
quality |
P2 |
change in Diameter |
R3 |
Level
of Integration |
P3 |
change in Thickness |
R4 |
Cost saving |
P4 |
Change in Width |
R5 |
EOL condition |
P5 |
Weight |
R6 |
Durability |
P6 |
Roundness |
Table 2: Fuzzy number allotment for Requirements
and Testing’s parameters
Requirements
for Design of Remanufacturing |
Testing’s parameters |
||||||
|
Notation |
Fuzzy No |
|
Notation |
Fuzzy No |
|
|
Not important |
A |
(0,0, 0.1) |
Very poor |
P |
(0,0,0.1) |
|
|
Less important |
B |
(0,0.1,0.3) |
Medium poor |
Q |
(0,0.1,0.3) |
|
|
Medium Less important |
C |
(0.1,0.3,0.5) |
Poor |
R |
(0.1,0.3,0.5) |
|
|
Medium important |
D |
(0.3,0.5,0.7) |
Fair |
S |
(0.3,0.5,0.7) |
|
|
Medium high important |
E |
(0.5,0.7,0.9) |
Good |
T |
(0.5,0.7,0.9) |
|
|
High important |
F |
(0.7,0.9,1) |
Very Good |
U |
(0.7,0.9,1) |
|
|
Very high important |
G |
(0.9,1,1) |
Excellent |
V |
(0.9,1,1) |
|
|
The results are classified as very
poor, medium poor, poor, fair, good, very good and excellent (Table 2). The
notation and the fuzzy number of each of the six grades for both requirements
for the design of remanufacturing and testing's parameters were made with
respect to triangular membership function.
Table 3 highlights the opinions
given by three decision makers (DM1, DM2, DM3) on requirements for the design
of remanufacturing. We took DM as design engineers from bearing manufacturing
industries. The same decision maker's opinion regarding the requirements for
the design of remanufacturing with respect to testing’s parameters was taken in
order to enhance the design for remanufacturing (Table 4). Based on the
aggregate fuzzy numbers obtained, fuzzy numbers are assigned to the linguistic
variables suggested by the decision maker. The values of normalized fuzzy
decision matrix are weighted by multiplying them with the relevant aggregated
fuzzy number (Table 5).
Table 3: Design makers opinion
Requirements for design of remanufacturing |
DM1 |
DM2 |
DM3 |
Aggregate Fuzzy
No. |
R1 |
G |
F |
G |
(0.84,0.97,1) |
R2 |
E |
G |
F |
(0.7,0.87,0.97) |
R3 |
E |
F |
C |
(0.44,0.64,0.8) |
R4 |
F |
D |
E |
(0.5,0.7,0.87) |
R5 |
C |
E |
F |
(0.44,0.64,0.8) |
R6 |
G |
F |
F |
(0.76,0.93,1) |
Table 4: Decision maker’s opinion about requirements
for design of remanufacturing with respect to testing’s parameters
|
R1 |
R2 |
R3 |
R4 |
R5 |
R6 |
P1 |
T |
T |
T |
P |
R |
S |
P2 |
V |
S |
U |
R |
U |
S |
P3 |
U |
T |
U |
Q |
S |
S |
P4 |
V |
S |
U |
T |
Q |
R |
P5 |
T |
U |
U |
R |
T |
T |
P6 |
V |
V |
V |
V |
U |
U |
Table 5: Weighed normalized fuzzy
decision matrix
|
R1 |
R2 |
R3 |
R4 |
R5 |
R6 |
P1 |
(0.58,0.87,1) |
(0.35,0.60,0.60) |
(0.22,0.44,0.72) |
(0.45,0.7,0.87) |
(0.04,0.19,0.4) |
(0.22,0.46,0.7) |
P2 |
(0.75,0.97,1) |
(0.21,0.43,0.67) |
(0.30,0.57,0.8) |
(0.05,0.21,0.43) |
(0.30,0.57,0.8) |
(0.22,0.46,0.7) |
P3 |
(0.58,0.87,1) |
(0.35,0.60,0.60) |
(0.30,0.57,0.8) |
(0,0.07,0.26) |
(0.13,0.32,0.56) |
(0.22,0.46,0.7) |
P4 |
(0.75,0.97,1) |
(0.21,0.43,0.67) |
(0.30,0.57,0.8) |
(0.25,0.49,0.78) |
(0,0.06,0.24) |
(0.07,0.27,0.5) |
P5 |
(0.42,0.67,0.9) |
(0.49,0.78,0.97) |
(0.30,0.57,0.8) |
(0.05,0.21,0.43) |
(0.22,0.44,0.72) |
(0.38,0.65,0.9) |
P6 |
(0.75,0.87,1) |
(0.63,0.87,0.97) |
(0.39,0.64,0.8) |
(0.45,0.7,0.87) |
(0.30,0.57,0.8) |
(0.53,0.83,1) |
Next, the ranking of the testing
parameters was obtained using the relations below.
D*
represents the Fuzzy Positive Ideal Solutions (FPIS D*).
D* = Σ
½ [max (|1st -1|; |3rd -1|) + (2nd -1)]
Here, the values |1st -1| and |3rd
-1| from weighed normalized decision matrix are compared. The greater of the 2
values is added to (2nd -1).
D#
represents the Fuzzy Negative Ideal Solutions (FNIS D#).
D# = Σ
½ [max (|1stb b; - 0|, |3rd - 0|) + |2nd - 0|]
Here, the values |1st-0| and |3rd-0|
from weighed normalized decision matrix are compared. The greater of the 2
values is added to (2nd-0).
Relative closeness coefficient of
strategies(C*) = D#/ (D*+ D#)
Table 6: Ranking of Testing’s
Parameters
Piston |
D* |
D# |
C* = D-/(D*+D#) |
RANK |
P1 |
3.44 |
3.695 |
0.518 |
2 |
P2 |
3.48 |
3.805 |
0.522 |
4 |
P3 |
3.765 |
3.405 |
0.474 |
5 |
P4 |
3.815 |
3.39 |
0.470 |
6 |
P5 |
3.41 |
4.02 |
0.541 |
3 |
P6 |
2.235 |
4.76 |
0.680 |
1 |
From the Fuzzy-TOPSIS analysis, we
found that roundness ranks first followed by surface roughness, weight, change
in diameter, change in thickness, and change in width. The rankings and
relative closeness coefficient are reported in Table 6.
5.
TAGUCHI OPTIMIZATION ANALYSIS
The surface roughness of outer race
measured from inside face and inner race measured from outside face. Values
obtained from the testing are reported in Table 7. For this experiment, we used
the following notations— outer race diameter (F1); inner race diameter (F2);
roller diameter (F3); outer race surface roughness (R1); inner race surface
roughness (R2); roller surface roughness (R3); outer race roundness (R4); inner
race roundness (R5); roller roundness (R6); outer race weight (R7); inner race
weight (R8) and roller weight (R9). The data of 11 bearings were considered for
analyses. The signal to noise ratio is calculated using the difference between
the minimum and maximum response value and the standard deviation.
Signal-to-Noise ratio was set as per data in Table 8. An optimized design is
made by determining the parameters and their values. The data was feed in the
software where the parameters were specified with their respective values as
shown in Table 9.
Table 7: Testing results for
selected bearings
Bearing no |
F 1 |
F 2 |
F 3 |
R 1 |
R 2 |
R3 |
R4 |
R5 |
R 6 |
R 7 |
R 8 |
R 9 |
1 |
77.5247 |
53.5025 |
12.0071 |
0.131 |
0.079 |
0.106 |
3.5 |
6 |
0.001 |
283.203 |
207.141 |
11.2844 |
2 |
77.5398 |
53.4566 |
12.0007 |
0.341 |
0.234 |
0.213 |
13 |
12.4 |
0.003 |
275.365 |
208.407 |
10.3439 |
3 |
77.5083 |
53.5937 |
11.9957 |
0.267 |
0.658 |
1.156 |
4.7 |
9 |
0.0026 |
276.641 |
206.576 |
10.4011 |
4 |
77.5183 |
53.4100 |
12.0134 |
0.293 |
0.155 |
0.332 |
8 |
251 |
0.0017 |
279.293 |
204.695 |
10.4361 |
5 |
77.5458 |
53.4308 |
11.9583 |
1.381 |
0.675 |
1.45 |
120 |
30 |
0.0068 |
270.579 |
199.621 |
10.2388 |
6 |
77.5039 |
53.5413 |
11.9655 |
0.837 |
0.814 |
1.422 |
13 |
63 |
0.0024 |
276.919 |
201.181 |
10.415 |
7 |
77.5438 |
53.4570 |
11.8234 |
0.238 |
0.224 |
0.589 |
41 |
18 |
0.0068 |
282.843 |
199.934 |
9.8912 |
8 |
77.5596 |
53.4617 |
12.0431 |
0.809 |
0.255 |
0.129 |
15 |
17 |
0.0021 |
276.158 |
207.732 |
10.4629 |
9 |
77.5192 |
53.5189 |
12.0145 |
0.145 |
0.25 |
0.38 |
15.9 |
9 |
0.002 |
273.587 |
203.041 |
10.3922 |
10 |
77.5086 |
53.3786 |
12.0041 |
0.554 |
0.307 |
0.711 |
17 |
18 |
0.0036 |
273.912 |
206.006 |
10.1073 |
11 |
77.4961 |
53.4869 |
12.0081 |
1.409 |
0.509 |
0.347 |
68 |
9 |
0.0027 |
277.053 |
209.666 |
10.3417 |
Table 8: S/N ratio for all responses
Response |
Minimum |
Maximum |
Mean |
Std. Dev. |
S/N Ratio |
R1 |
0.131 |
1.409 |
0.5823 |
0.4683 |
10.76 |
R2 |
0.079 |
0.814 |
0.3782 |
0.2437 |
10.30 |
R3 |
0.106 |
1.45 |
0.6214 |
0.5015 |
13.68 |
R4 |
3.5 |
120 |
29.01 |
35.57 |
34.29 |
R5 |
6 |
251 |
40.22 |
71.73 |
41.83 |
R6 |
0.001 |
0.0068 |
0.0032 |
0.0019 |
6.80 |
R7 |
270.579 |
283.203 |
276.87 |
3.79 |
1.05 |
R8 |
199.621 |
209.666 |
204.91 |
3.49 |
1.05 |
R9 |
9.8912 |
11.2844 |
10.39 |
0.3412 |
1.14 |
Table 9: Constraints used for
optimization analysis
Name |
Goal |
Lower Limit |
Upper Limit |
R1 |
minimize |
0.361939 |
1.18701 |
R2 |
minimize |
0.281069 |
0.902219 |
R3 |
minimize |
0.325576 |
1.20416 |
R4 |
minimize |
1.87083 |
10.9545 |
R5 |
minimize |
2.44949 |
15.843 |
R6 |
minimize |
0.001 |
0.0068 |
R7 |
maximize |
270.579 |
283.203 |
R8 |
maximize |
199.621 |
209.666 |
R9 |
maximize |
9.8912 |
11.2844 |
Table 10: Ranking of bearings
Bearing No |
R1 |
R2 |
R3 |
R4 |
R5 |
R6 |
R7 |
R8 |
R9 |
Desirability |
1(new) |
0.131 |
0.079 |
0.106 |
3.500 |
6. |
0.001 |
283.203 |
207.141 |
11.284 |
0.968 |
2 |
0.341 |
0.234 |
0.213 |
13.000 |
12.4 |
0.003 |
275.365 |
208.407 |
10.344 |
0.654 |
8 |
0.809 |
0.255 |
0.129 |
15.000 |
17. |
0.002 |
276.158 |
207.732 |
10.463 |
0.637 |
9 |
0.145 |
0.250 |
0.380 |
15.900 |
9. |
0.002 |
273.587 |
203.041 |
10.392 |
0.581 |
3 |
0.267 |
0.658 |
1.156 |
4.700 |
9. |
0.003 |
276.641 |
206.576 |
10.401 |
0.483 |
10 |
0.554 |
0.307 |
0.711 |
17.000 |
18. |
0.004 |
273.912 |
206.006 |
10.107 |
0.472 |
6 |
0.837 |
0.814 |
1.422 |
13.000 |
63. |
0.002 |
276.919 |
201.181 |
10.415 |
0.006 |
Table 10 shows the best seven
bearings which were selected for remanufacturing by Taguchi analysis having a
positive desirability ratio ranging from 0.006 to 0.968. Bearing no. 1 is a new
bearing taken for reference purposes. This shows Taguchi analysis provided the
significant results in the selection of bearings for remanufacturing and
correspondingly all the bearing were graded as per the degradation of the
bearing and four bearings we rejected due to excessive degradation over the
lifetime. However, 60% of the old bearings were selected for remanufacturing by
the analysis directly without doing any modification.
Figure 2: (a) Desirability curve for
bearing no. 1(new) (b) Desirability curve for bearing no. 2
Figure 2(a) shows the
new bearing desirability curve and for that, we achieved a combined
desirability value of 0.968. This shows even the new bearing having a slight
variation from the optimized values. The desirability of all the testing
parameters was achieved 100% except for the inner race weight which has a
desirability of 0.74. This decrease in desirability is due to the manufacturing
tolerance range set by the original equipment manufacturer which is acceptable.
Figure 2(b) shows the
desirability curve for old used bearing no. 2. A combined desirability value of
0.635 was obtained for this case. The highest desirability value was for the
inner race roundness and the lowest was inner race weight. The result shows
that the inner race parameters are having both the lowest and highest
desirability value but both are in the acceptable range.
Figure 3: (a) Desirability curve for bearing no. 8, (b) Desirability
curve for bearing no. 9
Figure 3(a) shows the desirability
curve for old used bearing no. 8 and for that, a combined desirability value of
0.637 was achieved. The highest desirability value was for the roller surface
roughness and the lowest was inner race weight. The result shows that the inner
race parameters are having lowest desirability value and second highest
desirability value but both are in the acceptable range.
Figure
3(b) shows the desirability curve for old used bearing no. 9 having a combined
desirability value of 0.580. The highest desirability value was for the outer
race roundness and the lowest was outer race weight. The result shows that the
outer race parameters are having both the lowest and highest desirability value
but both are in the acceptable range.
Figure 4: (a) Desirability curve for bearing no. 3 (b) Desirability
curve for bearing no. 10
Figure 4(a)
shows the desirability curve for old used bearing no. 3 with a combined
desirability value of 0.483. The highest desirability value was for the outer
race roundness and the lowest was inner race surface roughness. The desirability
curve for bearing no. 10 was reported in Figure 4(b) and has a combined
desirability value of 0.471. The highest desirability value was for the inner
race roundness and the lowest was roller weight.
Figure 5: Desirability curve for bearing no. 6
Figure 5 shows the desirability
curve for old used bearing no. 6 and for that, we achieved a combined
desirability value of 0.0174. The highest desirability value was for the outer
race roundness and the lowest was inner race roundness. From the overall desirability
values, it has been observed that both inner race and outer race roundness have
not changed from the original dimensions. However the roller weight was getting
reduced in most of the old bearings. This phenomenon is due to wear. This study
reveals that the failure of the roller is usually because of the outer race
roundness.
6.
CONCLUSION
Using advanced computational
intelligence techniques like Fuzzy-TOPSIS and Taguchi analysis, we tried to
understand the feasibility of roller bearing for design for remanufacturing. In
our study, we achieved a combined desirability of 0.968 giving maximum importance
to all the testing parameters and compromising slightly in inner race weight
with the individual desirability of 0.748 on a scale of 1.
Overall in bearings which were
selected by Taguchi analysis, the roundness of inner race and outer race were
found to be the most influential parameters in deciding the selection of
bearing for remanufacturing. The results suggest the bearing designers to
design the bearing in such a way that the roundness of both races will be taken
care off while manufacturing a bearing. The weight of the rollers was found to
be the least influential parameter in deciding the selection of bearing for
remanufacturing.
The result obtained from Taguchi
analysis was seen to be similar to Fuzzy-TOPSIS analysis which infuses further
confidence in the findings. As expected in the analysis, the bearing ranking
one was the new one, this shows that roller bearing selection using Taguchi
desirability analysis is a significant one. The study successfully highlighted
the potential of remanufacturing of bearing strategy in Indian region since
more than 50% of the bearings used in the analysis were selected for
remanufacturing.
REFERENCES
AIKHUELE, D. O.; FAIZ, M. T.; ODOFIN,
S. M.; ANSAH, R. H. (2017) Interval-Valued Intuitionistic Fuzzy Topsis-Based
Model For Troubleshooting Marine Diesel Engine Auxiliary System. International
Journal Maritime Engineering,
v. 159, p. 107-114. DOI: http://umpir.ump.edu.my/id/eprint/16772
AIKHUELE, D. O.; SOROOSHIAN, S.;
ANSAH, R. H.; TURAN, F. M. (2017) Application of Intuitionistic Fuzzy Topsis
Model for Troubleshooting an Offshore Patrol Boat Engine. Polish Maritime Research, n. 24, p. 68-76. DOI: https://doi.org/10.1515/pomr-2017-0051
AYDIN, R.; KWONG, C. K.; JI, P.
(2015) A novel methodology for simultaneous consideration of remanufactured and
new products in product line design. International journal of production
economics, n. 169,
p. 127-140. DOI: https://doi.org/10.1016/j.ijpe.2015.07.022
AZIZI, A.; AIKHUELE, D. O.; SOULEMAN,
F. S. (2015) A Fuzzy TOPSIS Model to Rank Automotive Suppliers. Procedia
Manufacturing, n. 2,
p. 159-164. DOI: https://doi.org/10.1016/j.promfg.2015.07.028
BAKIRCI, F.; SHIRAZ, S. E.; SATTARY, A. (2014) BIST'da Demir, Celik Metal
Ana Sanayii Sektorunde Faaliyet Gosteren Isletmelerin Finansal Performans
Analizi: VZA Super Etkinlik ve TOPSIS Uygulamasi/Financial Performance Analysis
of Iron, Steel Metal Industry Sector Companies in The Borsa Istanbul: DEA Super
Efficiency and TOPSIS Methods. Ege Akademik Bakis, 14, 9.
BENKHEROUF, L.; SKOURI, K.;
KONSTANTARAS, I. (2016) Optimal control of production, remanufacturing
and refurbishing activities in a finite planning horizon inventory system. Journal
of Optimization Theory and Applications, n. 168, p. 677-698. DOI: https://doi.org/10.1007/s10957-015-0741-9
DAS, P. P.; GUPTA, P.;
GHADAI, R. K.; RAMACHANDRAN,
M.; KALITA, K. (2017) Optimization of
turning process parameters by Taguchi-based Six Sigma. Mechanics and Mechanical
Engineering, n. 21, p.
649-656.
DINDARIAN, A.; GIBSON, A. A.; QUARIGUASI-FROTA-NETO, J. (2012) Electronic
product returns and potential reuse opportunities: a microwave case study in
the United Kingdom. Journal of Cleaner Production, n. 32, p. 22-31. DOI: https://doi.org/10.1016/j.jclepro.2012.03.015
DIYALEY, S.; SHILAL, P.;
SHIVAKOTI, I.; GHADAI, R.
K.; KALITA, K. (2017) PSI and TOPSIS
Based Selection of Process Parameters in WEDM. Periodica Polytechnica Mechanical Engineering, n. 61, p. 255-260.
DOI: https://doi.org/10.3311/PPme.10431
EREN, M. (2018) forecasting of the
fuzzy univariate time series by the optimal lagged regression structure
determined based on the genetic algorithm. Economic Computation & Economic
Cybernetics Studies & Research, v. 52, n. 2
FEGADE, V.; SHRIVATSAVA, R. L.; KALE, A. V. (2015) Design for
Remanufacturing: Methods and their Approaches. Materials Today: Proceedings, n. 2, p. 1849-1858. DOI: https://doi.org/10.1016/j.matpr.2015.07.130
FORD, S.; DESPEISSE, M. (2016) Additive manufacturing
and sustainability: an exploratory study of the advantages and challenges. Journal
of Cleaner Production, n. 137,
p. 1573-1587. DOI: https://doi.org/10.1016/j.jclepro.2016.04.150
GUNAY, M.; YUCEL, E. (2013) Application of Taguchi
method for determining optimum surface roughness in turning of high-alloy white
cast iron. Measurement, n. 46,
p. 913-919. DOI: https://doi.org/10.1016/j.measurement.2012.10.013
HAMDAN, A.; SARHAN, A. A.; HAMDI, M. (2012) An optimization method of
the machining parameters in high-speed machining of stainless steel using
coated carbide tool for best surface finish. The International Journal of
Advanced Manufacturing Technology, n. 58, p. 81-91.
DOI: https://doi.org/10.1007/s00170-011-3392-5
HATCHER, G. D.; IJOMAH, W. L.; WINDMILL, J. F. (2011) Design for
remanufacture: a literature review and future research needs. Journal
of Cleaner Production, n. 19,
p. 2004-2014. DOI: https://doi.org/10.1016/j.jclepro.2011.06.019
HILGER, T.; SAHLING, F.;
TEMPELMEIER, H. (2016) Capacitated dynamic production and
remanufacturing planning under demand and return uncertainty. OR
spectrum, n. 38, p.
849-876. DOI: https://doi.org/10.1007/s00291-016-0441-3
KENNE, J.-P.; DEJAX, P.;
GHARBI, A. (2012) Production planning of a hybrid
manufacturing--remanufacturing system under uncertainty within a closed-loop
supply chain. International Journal of Production Economics, n. 135, p. 81-93. DOI: https://doi.org/10.1016/j.ijpe.2010.10.026
KIVAK, T.; SAMTACS, G.;
CICEK, A. (2012) Taguchi method based optimisation of drilling
parameters in drilling of AISI 316 steel with PVD monolayer and multilayer
coated HSS drills. Measurement, n. 45, p. 1547-1557. DOI: https://doi.org/10.1016/j.measurement.2012.02.022
KWAK, M.; KIM, H. (2015) Design for life-cycle profit
with simultaneous consideration of initial manufacturing and end-of-life
remanufacturing. Engineering Optimization, n. 47, p. 18-35. DOI: https://doi.org/10.1080/0305215X.2013.868450
LIU, M.; LIU, C.;
XING, L.; LIU, Z.; LI, X.;
LIN, L. (2016) Assembly process control method for remanufactured parts
with variable quality grades. The International Journal of Advanced
Manufacturing Technology, n.
85, p. 1471-1481. DOI: https://doi.org/10.1007/s00170-015-8026-x
LU, D.; ZHAO, W.;
LU, B.; ZHANG, J. (2013) Cage
speed of hydrodynamic rolling hybrid bearings. Tribology Letters, n. 51, p. 303-309. DOI: https://doi.org/10.1007/s11249-013-0154-6
MANICKAM, R. (2016) Back propagation
neural network for prediction of some shell moulding parameters. Periodica
Polytechnica Mechanical Engineering, n. 60, p. 203-208. DOI: https://doi.org/10.3311/PPme.8684
PANDEY, A. K.; DUBEY, A. K. (2012) Taguchi based fuzzy logic
optimization of multiple quality characteristics in laser cutting of Duralumin
sheet. Optics and Lasers in Engineering, n. 50, p. 328-335. DOI: https://doi.org/10.1016/j.optlaseng.2011.11.005
PUROHIT, P.; RAMACHANDRAN, M. (2015) Selection of Flywheel
Material using Multicriteria Decision Making Fuzzy Topsis. Indian Journal of Science and
Technology, 8.33.
DOI: https://doi.org/10.17485/ijst/2015/v8i33/80028
RAGAVENDRAN, U.; GHADAI, R. K.; BHOI, A.;
RAMACHANDRAN, M.; KALITA, K. A.
(2018) Sensitivity Analysis and Optimization of EDM Process Parameters. Transactions
of the Canadian Society for Mechanical Engineering. DOI: https://doi.org/10.1139/tcsme-2018-0021
RAMACHANDRAN, M.; AGARWAL, N. (2017) Identification of Most
Affected Parameter for Design for Remanufacturing of Scrap Piston by Taguchi
Desirability Function Analysis. International Conference on Man--Machine
Interactions, p. 320-329. DOI: https://doi.org/10.1007/978-3-319-67792-7_32
SAHARE, S. B.; UNTAWALE, S. P.; CHAUDHARI, S. S.; SHRIVASTAVA, R. L.; KAMBLE, P. D. (2018) Optimization of End
Milling Process for Al2024-T4 Aluminum by Combined Taguchi and Artificial
Neural Network Process. In Soft Computing: Theories and Applications,
p. 525-535. Springer. DOI: https://doi.org/10.1007/978-981-10-5699-4_49
SELVARAJ, A.; MARAPPAN, R. (2011) Experimental analysis of
factors influencing the cage slip in cylindrical roller bearing. The
International Journal of Advanced Manufacturing Technology, n. 53, p. 635-644. DOI: https://doi.org/10.1007/s00170-010-2854-5
SENGUL, U.; EREN, M. (2016) Selection of digital
marketing tools using fuzzy AHP-fuzzy TOPSIS. In Fuzzy Optimization and
Multi-Criteria Decision Making in Digital Marketing, p. 97-126. DOI:
https://doi.org/10.4018/978-1-4666-8808-7.ch005
SENGUL, U.; EREN, M.;
SHIRAZ, S. E.; GEZDER, V.; SENGUL, A. B. (2015) Fuzzy TOPSIS method for
ranking renewable energy supply systems in Turkey. Renewable Energy, n. 75, p. 617-625. DOI: https://doi.org/10.1016/j.renene.2014.10.045
SERVAIS, C.; DUQUENNE, M.;
BOZET, J.-L. (2013) Influence of tribology of cage material on ball
bearing cage instability. 15th European Space Mechanism and Tribology
Symposium (ESMATS), p. 71-78.
SHIRAZ, S. E.; SENGUL, U.;
EREN, M. (2014) Determination of extended fuzzy TOPSIS method of
criteria leading to supplier selection for industries. Asian Social Science, n. 10, p. 183. DOI: https://doi.org/10.1007/s13369-017-2673-1
SHIVAKOTI, I.; PRADHAN, B. B.; DIYALEY, S.;
GHADAI, R. K.; KALITA, K. (2017)
Fuzzy TOPSIS-Based Selection of Laser Beam Micro-marking Process Parameters. Arabian
Journal for Science and Engineering, n. 42, p. 4825-4831. DOI: https://doi.org/10.1007/s13369-017-2673-1
SUBRAMANIAN, R.; FERGUSON, M. E.; BERIL TOKTAY, L. (2013) Remanufacturing and
the component commonality decision. Production and Operations Management, n. 22, p. 36-53. DOI: https://doi.org/10.1111/j.1937-5956.2012.01350.x
SUGUMARAN, V.; RAMACHANDRAN, K. I. (2007) Automatic rule
learning using decision tree for fuzzy classifier in fault diagnosis of roller
bearing. Mechanical Systems and Signal Processing, n. 21, p. 2237-2247. DOI: https://doi.org/10.1016/j.ymssp.2006.09.007
WEINZAPFEL, N.; SADEGHI, F. (2009) A discrete element
approach for modeling cage flexibility in ball bearing dynamics simulations. Journal
of Tribology, n. 131,
p. 021-102. DOI: https://doi.org/10.1115/1.3063817
ZHANG, T.; CHU, J.;
WANG, X.; LIU, X.; CUI, P. (2011) Development pattern and
enhancing system of automotive components remanufacturing industry in China. Resources,
Conservation and Recycling,
n. 55, p. 613-622. DOI: https://doi.org/10.1016/j.resconrec.2010.09.015