Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Materials and Design 31 (2010) 1785–1789

Contents lists available at ScienceDirect

Materials and Design


journal homepage: www.elsevier.com/locate/matdes

A selection of material using a novel type decision-making method:


Preference selection index method
Kalpesh Maniya a,*, M.G. Bhatt b
a
Department of Mechanical Engineering, C.K. Pithawalla College of Engineering and Technology, Affiliated to Veer Narmad South Gujarat University, Surat 395007, Gujarat, India
b
Department of Production Engineering, Shantilal Shah Engineering College, Bhavnagar, Gujarat, Affiliated to Bhavnagar University, Bhavnagar, Gujarat, India

a r t i c l e i n f o a b s t r a c t

Article history: The aim of the current study is to implement a novel tool to help the decision-maker for selection of a
Received 7 October 2009 proper material that will meet all the requirements of the design engineers. Preference selection index
Accepted 8 November 2009 (PSI) method is a novel tool to select best alternative from given alternatives without deciding relative
Available online 12 November 2009
importance between attributes. In the present study, three different types of material selection problems
are examined. A validation and consistency test of preference selection index method is performed in
Keywords: present work by comparing results of PSI method with published results of graph theory and matrix
Material selection
approach (GTMA), and technique for order preference by similarity to ideal solution (TOPSIS) method,
Preference selection index method
Validation and consistency test
respectively. The research has concluded that the PSI method is logical and more appropriate for the
material selection problems.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction technique for order preference by similarity to ideal solution (TOP-


SIS), Vlse Kriterijumska Optimizacija Kompromisno Resenje (VI-
Recent developments in design, the selection of materials play KOR) method, analytical hierarchy process (AHP), graph theory
an important role for engineers. The material selection should and matrix representation approach (GTMA), etc. [7–10]. Rao
not be solely based on cost but also depends on different properties [9,10] presented graph theory and matrix representation approach
of material, availability, recycling, production method, disposal (GTMA) for the material selection. Shanian and Savadogo [11] pre-
method, design life, etc. selection of material depends on number sented ELimination and Et Choice Translating REality (ELECTRE)
of attributes or factors. Hence, selection of material is a multi attri- outranking method for the material selection. Shanian and Sava-
bute decision-making problem. dogo [12] applied TOPSIS method as multiple-criteria decision sup-
Selection of the appropriate material is an integral part of suc- port analysis for material selection of metallic bipolar plates for
cessfully implementation of an engineer’s design. The ability to se- polymer electrolyte fuel cell. Manshadi et al. [13] proposed numer-
lect the most appropriate material for a given application is the ical method for the material selection combining non-linear nor-
fundamental challenges faced by the design engineer. A systematic malization with modified digital logic method. Chan and Tong
and efficient approach to material selection approach is necessary [14] used grey relational analysis approach (GRA) for multi-criteria
in order to select the best alternative for a given application [1–5]. selection method. Rao [15] presented improved compromise rank-
The importance of materials selection in engineering design has ing method for material selection. Rao and Davim [16] described
been well recognized. The design decision-making regarding combined multiple attribute decision-making AHP/TOPSIS meth-
selecting appropriate materials is dictated by the specific require- odology. Prasenjit et al. [17] used compromise ranking and outran-
ments of an application, often the requirements on materials prop- king methods for material selection. Also material selection is
erties [6]. carried out using fuzzy decision-making, material design and
In the past lots of research had been reported for selection of selection using multi objective decision-making methods [18–21].
material using classical multi attribute decision-making methods. The objectives of current research it to implement a novel
A multi attribute analysis is a popular tool to select best alternative method named preference selection index (PSI) method for selec-
for given applications and the methods are simple additive tion of material for a given application. PSI method is a systematic
weighted (SAW) method, weighted product method (WPM), scientific method or tool for design engineers to select the appro-
priate material for the given application. To illustrate the PSI meth-
* Corresponding author. Tel.: +91 0261 2728282. od for material selection, one example is considered, which already
E-mail address: maniya777@yahoo.co.in (K. Maniya). solved using improved compromise ranking method by Rao [15]. In

0261-3069/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.matdes.2009.11.020
1786 K. Maniya, M.G. Bhatt / Materials and Design 31 (2010) 1785–1789

this research, validation and consistency test of proposed method where xij is the attribute measures (i = 1, 2, 3, . . .. , N and
for material selection are checked by comparing the results of PSI j = 1, 2, 3, . . . , M)
method with published results of GTMA and TOPSIS methods,
respectively. Step IV: Compute preference variation value (PVj).

2. Preference selection index (PSI) method In this step, preference variation value (PVj) for each attribute is
determined with concept of sample variance analogy using follow-
The proposed approach is new for selection of material. Mostly ing equation:
material selection is completed using multi attribute decision-
X
N
making methods. A literature review clearly indicates that in all PVj ¼ ½Rij  Rj 2 ð3Þ
these existing multi attribute decision-making methods it is neces- i¼1
sary to assign relative importance between attributes or attributes
where Rj is the mean of normalized value of attribute j and
weight and requires many complex calculations. In the proposed P
Rj ¼ N1 Ni¼1 Rij
method, it is not necessary to assign relative importance between
attributes, but in this method overall preference value of attributes
Step V: Determine overall preference value (Wj).
are calculated using concept of statistics. This method is useful
when there is conflict in deciding the relative importance between
In this step, the overall preference value (Wj) is determined for
attributes and that is the beauty of PSI method. Using overall pref-
each attribute. To get the overall preference value, it is required to
erence value, preference selection index (Ii) for each alternative is
find deviation (Uj) in preference value (PVj) and the deviation in
calculated and alternative with higher value of PSI is selected as
preference value for each attribute is determined using the follow-
best alternative. The detail steps for calculation of PSI are given
ing equation:
in the following methodology.
Uj ¼ 1  PVj ð4Þ
2.1. Methodology of PSI method
and overall preference value (Wj) is determined using following
equation:
Step I: Identify the goal; find out all possible the material alter- U
Wj ¼ PM j ð5Þ
natives, selection criteria and its measures for the given Uj
j¼1
application.
Step II: Formulate decision matrix.The solving each MADM The total overall preference value of all the attributes should be
P
problem begins with constructing decision matrix. Let, A = {Ai one, i.e. j Wj ¼ 1.
for i = 1, 2, 3, . . . , n} be a set of alternative, C = {Cj for
j = 1, 2, 3, . . . , m} be a set of decision criteria or attributes, xij is Step VI: Obtain preference selection index (Ii).
the performance of alternative Ai when it examined with crite-
ria Cj. Then the decision matrix is represented in tabular format Now, compute the preference selection index (Ii) for each alter-
as shown in Fig. 1. native using following equation:
Step III: The data normalization.
X
M
Ii ¼ ðRij  Wj Þ ð6Þ
The process of transforming attributes value into a range of 0–1 j¼1
is called normalization and it is required in multi attribute deci-
sion-making methods to transform performance rating with differ- Step VII:
ent data measurement unit in a decision matrix into a compatible
unit.
After calculation of the preference selection index (Ii), alterna-
If the expectancy is the-larger-the-better (i.e. profit), then the
tives are ranked according to descending or ascending order to
original attribute performance can be normalized as follows:
facilitate the managerial interpretation of the results, i.e. an alter-
xij native is ranked/selected first whose preference selection index (Ii)
Rij ¼ ð1Þ
xmax
j is highest and an alternative is ranked/selected last whose prefer-
ence selection index (Ii) is the lowest and so on.
If the expectancy is the-smaller-the-better (i.e. cost), then the
original attribute performance can be normalized as follows:
3. Illustration of example
xmin
j
Rij ¼ ð2Þ In this section, one example of material selection problem is
xij
considered to demonstrate the methodology of preference selec-
tion index method. This example is earlier illustrated by Rao [15]
using improved compromise ranking method. The quantitative
and qualitative data of material selection problem is shown in
Criteria (Cj) Table 1.
Alternatives (A i)
C1 C2 …. Cm The detailed steps involved in the application of PSI method for
A1 x11 x12 ….. x1n selecting optimal material for the given application are described
A2 x21 x22 ….. x2n below:
: : : : :
Step I: Decide the all the possible alternative materials for a
: : : : :
given application, its selection criteria, and its values. In present
An xm1 xm2 ….. xmn
study, five material alternatives with four attributes, and their
Fig. 1. Decision matrix. attribute measures are considered as same of Rao [15] and
K. Maniya, M.G. Bhatt / Materials and Design 31 (2010) 1785–1789 1787

Table 1 tween attributes when numbers of attributes are larger in


Quantitative data of material selection attributes [15]. selection process. In present work, PSI method gives same results
Material Tensile Material selection attributes with minimum and simple calculations without support of any
strength (MPa)
Young’s Density Corrosion other methods like AHP method, Entropy method that are used
modulus (GPa) (gm/cm3) resistance for determination of attribute’s weight. A PSI method can use for
1 1650 58.5 2.3 Average (0.5)
any numbers of attributes.
2 1000 45.4 2.1 Low (0.335)
3 350 21.7 2.6 Low (0.335)
4 2150 64.3 2.4 Average (0.5) 4. Validation of PSI method
5 700 23 1.71 Above
average (0.59) To validate proposed method for material selection, an example
based on graph theory and matrix approach (GTMA) for material
selection problem is considered as the same of Rao [9,10]. In this
the attributes are tensile strength, Young’s modulus, density problem, Rao [9,10] has taken six alternatives and four attributes
and corrosion resistance. for material selection as shown in Table 3.
Step II: In this step, decision matrix is formulated. A decision In GTMA, there is need to assign a relative importance between
matrix is nothing but representation of all the data in tabular material selection attributes. Generally in the GTMA a relative
format and it is shown in Table 1. importance between attributes are assigned using 8-scales pro-
Step III: In this step, attribute measures are normalized to con- posed by Chen and Hawang [7], scale proposed by Saaty [22],
vert in compatible unit using Eqs. (1) and (2). In present study, and the 11-point scale given by Venkatasamy and Agrawal [23].
density of material is non-beneficial attributes, i.e., lower value Rao [9,10] had assigned the relative importance between attribute
is desirable and the remaining three attributes are beneficial, by considering the 11-point scale given by Venkatasamy and Agra-
i.e., higher value is desirable. The normalized data of the mate- wal [23] as shown in Fig. 2.
rial selection attributes is shown in Table 2. This example is solved using PSI method, which is described in
Step IV: The values of preference variation (PVj) are calculated Section 2.1 and according to step III of PSI method a normalized va-
using Eq. (3) and these are given below: lue of material selection attributes are calculated by using Eqs. (1)
PV1 = 0.4571; PV2 = 0.3755; PV3 = 0.0703; PV4 = 0.1466. and (2). In this example, cost of material is non-beneficial attri-
Step V: Overall preference values (W) of attributes are calculated butes and the remaining three attributes are beneficial attributes.
using Eq. (5) and these are given below: The normalized data of the material selection attributes are shown
U1 = 0.5429; U2 = 0.6245; U3 = 0.9297; U4 = 0.8534 and in Table 4.
W1 = 0.1840; W2 = 0.2117; W3 = 0.3151; W4 = 0.2894. Result: By applying step IV to step VII of PSI method, results are
Step VI: Overall preference selection index (Ii) is calculated for obtained and it compared with the published result of GTMA ap-
each alternative using Eq. (6) and its values are given below: plied by Rao [9,10] as given in Table 5.
I1 = 0.8133; I2 = 0.6560; I3 = 0.4730; I4 = 0.8655; I5 = 0.7401. Discussion: For the given application, Rao [9,10] used GTMA and
Step VII: Best alternative is ranked by Ii is the one with maxi- get the ranking order of material is 5–4–2–3–1–6 where as PSI
mum value of Ii and ranking order is given as 4–1–5–2–3. As method suggests the ranking order of material is 5–4–2–3–1–6.
per the ranking order, alternative 4 is the first choice, alterna- Both the method suggests that material 5 is best alternative while
tive 1 is the second choice for a given application and alterna- material 6 is last alternative. The results of PSI method exactly
tive 3 is the last choice. match with the result of GTMA by Rao [9,10]. A GTMA must require
deciding the relative importance between attributes whereas the
proposed PSI method does not need to decide any relative impor-
Rao [15] solved the same problem earlier using improved com-
promise ranking method. Rao [15] obtained alternative 4 is the tance between attributes or deciding weight of attributes. Let,
best or first choice and alternative 1 is the second choice by apply-
ing improved compromise ranking method with use of analytical Table 3
hierarchy approach (AHP) for determination of attribute’s weight Quantitative data of material selection attributes [9,10].
or to assign relative importance between attributes. The results
Material Hardness Material selection attributes
of PSI method are matches with results obtained by Rao [15]. But (HB)
Machinability Cost ($/lb) Corrosion
Rao [15] made a calculation mistake and he explained the im-
rating % (MR) (C) resistance (CR)
proved compromise ranking methods only for four alternative
1 420 25 5 Extremely high (0.865)
materials. Hence, results comparison table is not shown for this
2 350 40 3 High (0.665)
material selection example. If decision makers use AHP method 3 390 30 3 Very high
for determination of attribute weight then it necessary to check 4 250 35 1.3 High (0.665)
consistency in judgments taken to assign relative importance be- 5 600 30 2.2 High (0.665)
tween attributes. It is difficult to assign relative importance be- 6 230 55 4 Average (0.5)

Table 2
Normalized (Rij) data of material selection attributes. HB MR C CR
Material Tensile
strength (MPa)
Material selection attributes HB − − 0.335 0.665 0.665
Young’s Density Corrosion
modulus (GPa) (gm/cm3) resistance MR 0.665 − − 0.745 0.745
D4×4 =
1 0.7674 0.9098 0.7435 0.8475 C 0.335 0.255 − − 0.335
2 0.4651 0.7061 0.8143 0.5678
3 0.1628 0.3375 0.6577 0.5678 CR 0.335 0.255 0.335 − −
4 1.0000 1.0000 0.7125 0.8475
5 0.3256 0.3577 1.0000 1.0000
Fig. 2. A relative importance between attributes [9,10].
1788 K. Maniya, M.G. Bhatt / Materials and Design 31 (2010) 1785–1789

Table 4 Table 7
Normalized data (Rij) of material selection attributes. Normalized value of attributes for wind turbine blades material.

Material Hardness Normalized value of material selection attributes Material Stiffness Material selection attributes
(HB) (GPa)
Machinability Cost ($/ Corrosion Tensile Density Elongation Maximum
rating % (MR) lb) (C) resistance (CR) strength (g)/cm3 at break (%) temperature
(MPa)
1 0.7000 0.4545 0.2600 1.0000
2 0.5833 0.7273 0.4333 0.7688 Steel 0.0857 0.0475 0.1933 1.0000 1.0000
3 0.6500 0.5454 0.4333 0.8613 Aluminum 0.0285 0.0225 0.5370 0.8000 0.8000
4 0.4167 0.6364 1.0000 0.7688 Glass-E 0.2086 0.875 0.5709 0.2000 0.7000
5 1.0000 0.5454 0.5909 0.7688 Carbon 1.0000 1.0000 0.8286 0.1200 1.0000
6 0.3833 1.0000 0.3250 0.5780 Aramid 0.3428 0.9000 1.0000 0.7333 0.5000

Table 5 Table 8
Result comparison of PSI method with GTMA. Result comparison of PSI method with TOPSIS method.

Material Published results of GTMA [9,10] Results of PSI method Material Published results of TOPSIS [24] Results of PSI method
VPF value Rank Index value (I) Rank Performance score (P) Rank Index value (I) Rank
1 1.2071 5 0.6321 5 Steel (A1) 0.5188 4 0.6058 3
2 1.3358 3 0.6436 3 Aluminum (A2) 0.5217 3 0.5820 4
3 1.3166 4 0.6393 4 Glass-E (A3) 0.4520 5 0.4965 5
4 1.6169 2 0.6973 2 Carbon (A4) 0.5606 2 0.7990 1
5 1.6353 1 0.7294 1 Aramid (A5) 0.5908 1 0.6584 2
6 1.1250 6 0.5884 6

which is solved using TOPSIS method. Suresh Babu et al. [24] have
there is a decision-making problem for selection of material alter- presented an illustrative problem for the material selection for typ-
native with M attributes and N alternative and it will be solved by ical wind turbine blades using TOPSIS method. The problem con-
graph theory and matrix approach. Subsequently, it is essential to sidering five alternatives and five attributes as shown in Table 6.
find determinant of M  M relative assignment matrix for N times. This problem is solved using PSI methodology, which is de-
It easy to get the determinant of 3  3 and 4  4 matrices but it is scribed in Section 2.1 and according to step III of PSI method, nor-
difficult to solve determinant of M  M matrix. It indicates lots of malized value of material selection attributes for wind turbine
calculation are required to find best and worst alternative using blade are calculated by using Eqs. (1) and (2). In this example, den-
GTMA, but similarly same best and worst alternative will be ob- sity of material is non-beneficial attributes and the remaining 4
tained using PSI with minimum calculation without considering attributes are beneficial attributes. The normalized data of the
any kind of relative importance and solving any size of relative material selection attributes are shown in Table 7.
assignment matrix. Hence, the application of GTMA becomes intri- Result: By applying step IV to step VII of PSI method, results are
cate when large numbers of alternatives and attributes are in- obtained and it is compared with the published result of TOPSIS
volved for selection of material alternative for the given considered by Suresh Babu et al. [24] for selection of material for
application. In addition, GTMA requires special computer program- wind turbine blades as given in Table 8.
ming to get determinant of matrix larger then 4  4 matrix Discussion: Suresh Babu et al. [24] used the TOPSIS method and
whereas PSI method is easy to understand, simple in calculations had considered attributes weight according to an importance and
and does not require special computer programming compares to capability of materials as Wj = [1,2,2,3,4]. Suresh Babu et al. [24]
GTMA. obtained the ranking an order of ideal solution is
A validation for any new approach is not just sufficient but it A5 > A4 > A2 > A1 > A3 whereas PSI method suggests the ranking
also requires checking of its consistency by applying the proposed an order of ideal solution is A4 > A5 > A1 > A2 > A3. Suresh Babu
method to other problems. Hence, consistency test is required to et al. [24] recommend that material alternative A5, i.e. aramid fiber
approve new methodology. is the first choice, and carbon fiber, aluminum, and steel are placed
in the second, third, and fourth choices, respectively. Whereas, PSI
5. Consistency test of PSI method method suggests material alternative A4, i.e. carbon fiber the first
choice, and aramid fiber, steel, and aluminum are placed in the sec-
To apply the new methods for material selection problem it is ond, third, and fourth choice, respectively. Both the method sug-
require checking the consistency of the new methods. To check gest that material alternative A3, i.e. Glass-E is the last choice.
consistency of proposed method, another problem is considered, According to the TOPSIS method aramid fiber is the best alternative
but due to poor compressive strength, poor machinability and poor
environmental stability and poor temperature strength, Suresh
Table 6 Babu et al. [24] have revised the results and selected the best alter-
Objective date of material for wind turbine blade [24]. native as carbon fiber material (i.e. A4) for wind turbine blades
Material Stiffness Material selection attributes analysis and worst material is A3. While PSI method already sug-
(GPa)
Tensile Density Elongation Maximum gests carbon fiber material (i.e. A4) is the best alternative and
strength (g/cm3) at break (%) temperature worst material is Glass-E (i.e. A3).
(MPa)
Steel 30 190 7.5 15 550 6. Concluding remarks
Aluminum 10 90 2.7 12 400
Glass-E 73 3500 2.54 3 350 In present study, PSI method is applied on three different types of
Carbon 350 4000 1.75 1.8 550
material selection problems without considering any relative
Aramid 120 3600 1.45 11 250
importance between attributes. The results obtained by PSI method
K. Maniya, M.G. Bhatt / Materials and Design 31 (2010) 1785–1789 1789

are compared with published results of improved compromise rank- [8] Hwang CL, Yoon KP. Multiple attribute decision-making: methods and
applications. New York: Springer; 1991.
ing method, graph theory and matrix approach and TOPSIS method.
[9] Rao RV. Decision making in the manufacturing environment using graph theory
These three methods are very different in functioning, and assign- and fuzzy multiple attribute decision-making methods. London: Springer-
ment of relative importance between attributes or in determination Verlag; 2007.
of weight of attributes compared to PSI method even though PSI [10] Roa RV. A material selection model using graph theory and matrix approach.
Mater Sci Eng A 2006;431:248–55.
method gives the same result for selection of material i.e. appropri- [11] Shanian A, Savadogo O. A material selection model based on the concept of
ate or best material for the given application. multiple attribute decision making. Mater Des 2006;27:329–37.
Finally, it is concluded that preference selection index method [12] Shanian A, Savadogo O. TOPSIS multiple-criteria decision support analysis for
material selection of metallic bipolar plates for polymer electrolyte fuel cell. J
is most appropriate and competent for selection of the best mate- Power Sources 2006;159:1095–104.
rial for any given application when large numbers of attribute are [13] Manshadi BD, Mahmudi H, Abedian A, Mahmudi R. A novel method for
involved in selection process. This method can be applied success- materials selection in mechanical design: combination of non-linear
normalization and a modified digital logic method. Mater Des 2007;28:8–15.
fully to any number of alternatives. The PSI method gives directly [14] Chan JWK, Tong TKL. Multi-criteria material selections and end-of-life product
optimal solution without assigning the relative importance be- strategy: grey relational analysis approach. Mater Des 2007;28:1539–46.
tween materials selection attributes and it is the beauty of PSI [15] Rao RV. A decision-making methodology for material selection using an
improved compromise ranking method. Mater Des 2008;29(10):1949–54.
method. In addition, calculations are uncomplicated, easy to [16] Rao RV, Davim JP. A decision-making framework model for material selection
understand, systematic and logical approach due to use of concept using a combined multiple attribute decision-making method. Int J Adv Manuf
of statistics compare to GTMA, VIKOR, TOPSIS, etc. preference Technol 2008;35:751–60.
[17] Chatterjee Prasenjit, Athawale Vijay Manikrao, Chakraborty Shankar. Selection
selection index method can be considered as a novel tool for selec-
of materials using compromise ranking and outranking methods. Mater Des
tion of materials by design engineers. 2009;30(10):4043–53.
[18] Liao TW. A fuzzy multi-criteria decision-making method for material selection.
J Manuf Syst 1996;15:1–12.
References [19] Ashby MF. Multi-objective optimization in material design and selection.
Mater Des 2000;48:359–69.
[1] Fisher PE, Lawrence W. Selection of engineering materials and [20] Khabbaz SR, Manshadi BD, Abedian A, Mahmudi R. A simplified fuzzy logic
adhesives. Taylor & Francis Group: CRC Press; 2005. approach for materials selection in mechanical engineering design. Mater Des
[2] Ashby MF, Johnson K. Materials and design: the art and science of materials 2009;30:687–97.
selection in product design. Oxford: Butterworth Heinemann; 2002. [21] Fayazbakhsh K, Abedian A, Manshadi BD, Khabbaz SR. Introducing a novel
[3] Farag M. Materials selection for engineering design. New York: Prentice-Hall; method for materials selection in mechanical design using Z-transformation in
1997. statistics for normalization of material properties. Mater Des 2009;30:
[4] Edwards KL. Selecting materials for optimum use in engineering components. 4396–404.
Mater Des 2005;26:469–73. [22] Saaty TL. Fundamentals of decision-making and priority theory with the
[5] Deng YM, Edwards KL. The role of materials identification and selection in AHP. Pittsburg: RWS Publications; 2001.
engineering design. Mater Des 2007;28:131–9. [23] Venkatasamy R, Agrawal VP. A digraph approach to quality evaluation of an
[6] Edwards KL, Deng YM. Supporting design decision-making when applying automotive vehicle. Qual Eng 1997;9:405–17.
materials in combination. Mater Des 2007;28:1288–97. [24] Suresh Babu K, Subba Raju NV, Srinivas Reddy M, Nageswara Rao D. The
[7] Chen SJ, Hwang CL. Fuzzy multiple attribute decision-making methods and material selection for typical wind turbine blades using a MADM approach &
applications. Lecture notes in economics and mathematical analysis of blades. In: Proceeding of MCDM, Chania, Greece: June 19–23;
systems. Berlin: Springer-Verlag; 1992. 2006.

You might also like