Available online at www.sciencedirect.

com

ScienceDirect
Materials Today: Proceedings 5 (2018) 12161–12173 www.materialstoday.com/proceedings

ICMMM - 2017
Supplier Selection and Order Allocation in Supply Chain
Karuna kumar .Ga, M.SrinivasaRaob Dr. KesavaRao V. V. S c
a
Assistant professor , Department of Mechanical Engineering , Gudlavalleru Engineering College, Gudlavalleru, Andhra Pradesh 521356, India
b
DGM, vizag steel plant, Visakhapatnam A.P., India.
c
Professor, Department of Mechanical Engineering , Andhra University College of Engineering, Visakhapatnam A.P., India.

Abstract
Today, there are growing challenges on business environment because of increase in competition and customer expectations
.selection of supplier and later order allocation is an essential subject matter in the area of supply chains. An well-ordered and
flexible supply chain allows the firm to choose the correct suppliers at the right time for the right kind of materials, with reduced
purchase cost, thereby improving competitiveness.
Supply chains have the basic characteristics like multiple conflicting criteria, demand uncertainties, lead time and
delivery uncertainties. In the literature isolated studies on supplier selection using mathematical models, multiple criteria decision
making techniques and multiple stakeholder requirements are commonly found. Integration of these aspects give promising
results. The correlated Analytic Hierarchy (CAHP) considers the correlation effect between criteria in the Analytic Hierarchy
process. Linear physical programming (LPP) is a multi-objective optimization method that develops an aggregate objective
function of the criteria in a piecewise, goal-programming fashion. LPP model enables decision maker to think about multiple
criteria (i.e., cost, customer service, and rejections) and to express criteria preferences in terms of degrees of desirability. This
paper proposes an integrated method for dealing with such problems using correlated Analytic Hierarchy– and linear physical
programming techniques. The method proposed demonstrates selection of appropriate suppliers and allocates orders optimally
among them .finally model calculation is presented.
© 2017 Elsevier Ltd. All rights reserved.
Selection and/or Peer-review under responsibility of International Conference on Materials Manufacturing and Modelling (ICMMM - 2017).

Keywords: Analytic Hierarchy process ,linear physical programming, supplier selection, order allocation

Introduction

Today, there are growing challenges on business environment because of increase in competition and customer
expectations .selection of supplier and later order allocation is an essential subject matter in the area of supply
chains.. An well-ordered and flexible supply chain allows the firm to choose the correct suppliers at the right time
for the right kind of materials, with reduced purchase cost, thereby improving. Competitive advantage stems from
the many discrete activities a company performs in designing, producing, marketing, delivering, and supporting its
products. Supplier (vendor) selection is a significant issue in supply chain management (SCM) field for many
enterprises, therefore its objective is an identification of suppliers .Basically, there are two aspects in the issue of the
supplier selection problem: such as specification of criteria used for evaluation of suppliers, and second aspect is
the method applied to rank these suppliers. Evaluation of a supplier depends on several criteria such as price,
quality, delivery, reputation etc. These criteria can influence the outcome of the decision-making with an influence

2214-7853© 2017 Elsevier Ltd. All rights reserved.

Selection and/or Peer-review under responsibility of International Conference on Materials Manufacturing and Modelling (ICMMM - 2017).
12162 Karuna kumar .G et al / Materials Today: Proceedings 5 (2018) 12161–12173

on each other. Therefore, selecting the right suppliers is a key to the procurement process and represents a major
opportunity for companies to reduce costs the concept of utility function to reflect human persuading, such as
maximum satisfactory, less cost etc forms multiple attribute decision making (MADM) problems.
The objective of the present study includes identifying the criteria for supplier selection Study of the factors whether
they influence each other to find the correlation matrix To find the weights considering the correlation by multi
objective programming. To find the relative weighted factors to find the scores among the suppliers using AHP and
finally the quantity to be ordered on each supplier using linear physical programming.

2. MULTI ATTRIBUTE DECISION MAKING PROBLEMS

2.1The analytic hierarchy process (AHP)

Chen et al., 2006 (1)) suggested a fuzzy multiple criteria decision-making method for supplier selection
problems. A mixed integer non-linear programming model was proposed to address the multi-criteria sourcing
problem by Ghodsy pour & O’Brien, 2001[2]).A survey of Multiple Attribute Decision Making Methods (MADM)
and Applications was done by Hwang, C.L., and Yoon, K., (1981)[3]. the basic features that govern MADM are
construction of hierarchy system considering the nature of problem, selection of proper multiple criteria decision-
making technique, finding relative weights with respect to each alternative and finally choosing the best alternative
based on scores calculated against each alternative(ref fig1).Saaty [4] has devised method to calculate the relative
weights. The AHP is used to calculate the weights of independent criteria and related with the upper-level criteria.

Figure 1; Multi Criteria Decision

The effects of outer-dependence, inner dependence and feedback are also taken into consideration by saaty and
developed analytical network process. In addition to this there exists a reality correlation between criteria such as
quality of product versus after sales service, initial cost of vehicle versus reliability etc.

2.2Multi objective programming:

Linear programming techniques of operation research address optimization of single objective. But in real situation,
problems often bounded by multi objectives, which can be expressed mathematically as
Objective function Max z(x) = [z1(x),z2(x),……,zn(x)] -----------[1]
and Subject to constraints -g(x) ≤ b, and x ≥ 0.
Such Multi objective can be solved by optimizing each objective as follows:
Objective function Maxzj(x) --------- [2]
and Subject to constraints , g(x) ≤ b, and x ≥0.
 Karuna kumar .G et al./ Materials Today: Proceedings 5 (2018) 12161–12173 12163

Assuming we can get the ideal point as z∗ the point located on the Pareto solutions is closest to the ideal point as the
optimal solution .Hence the distance between objective values and the ideal point can be calculated using lp norm
concept which can be expressed as
1/
Min dp= ∑ =1 ∗
( )− ( ) , p=1,2,……,∞ --------[3]
s.t g(x) ≤ b,and x ≥ 0,
The minimum value of the jth goal is denoted by (x), = weightage of j th objective and lp –norm p; x is variable
and b refers to boundary limit

2.3. Correlation
. Correlation is a statistical relationships involving dependence, in simple terms linear relationship between
twovariables.. Correlations can be used for prediction of relationship and its value varies from -1 to +1 .Tthe Pearson
correlation coefficient. It is calculated multiplying the standard deviations and the result is divided by
the covariance of the two variables. The sum of squares forfirst variable say , the sum of square for second
variable say , and the sum of the the product of mean deviations ( ). The sum of squares for variable X is:
=∑ − ..........(4)
The sum of squares for variable Y is:
=∑ −
...................(5)
Finally, the sum of the product of mean deviations (SSXY) is:
= ∑( − ) − ----------- (6)
The correlation coefficient r is = ----------(7)
( )( )

3. Literature survey:

There are comprehensive literature reviews performed for supplier selection application by .Dickson [5], Weber et
al. [6], De Boer et al. [7] and Sanayei et al. [8]. Ayhan [9].Dickson‟s [4] stated 23 criteria for supplier selection.
Cheraghi et al. [10] updated Dickson‟s criteria with 13 more . As a brief of all criteria price, quality, and delivery
performances are found as the most significant selection criteria’s .various MCDM are put into practice , which can
be catagorised broadly into into three

1) Value Models: AHP and multi attribute utility theory (MAUT) fall in this group.
2) Goal, Models: Goal programming , TOPSIS.VIKOR belong to the group.
3) Outranking Methods: PROMETHE and ELECTRE etc belong to this group.

Ghodsypour, S.H., and O‟Brien, C [11] proposed A Decision Support System for Supplier Selection Using an
Integrated AHP and Linear Programming. Kilic,[ 12] suggested An integrated approach for supplier selection in
multi item/multi supplier environment .

Xia, W. and Wu, Z., [13] considered Supplier selection with multiple criteria.AHP, which was first developed by
Saaty integrates experts‟ opinions and evaluation scores into a simple elementary hierarchy system.Yahya and
Kingsman [14] used AHP to determine priorities in selecting suppliers. The book by Lambert and Gupta [15]
presented the importance of the area of disassembly.

Gungor and Gupta [16] cited a comprehensive study in manufacturing industry .Kongar and Gupta [17] presented a
multi-criteria decision making approach. Imtanavanich and Gupta [18] modelled the supply chain problem with
stochastic yields LPP technique is used in solving the supply chain problem by Imtanavanich and Gupta [19].
Massoud and Gupta [20] considered the multi-period order problem .Kongar and Gupta [21] proposed a LPP model
for environmenta land financial goals.
12164 Karuna kumar .G et al / Materials Today: Proceedings 5 (2018) 12161–12173

4. Methodology:

The study is done in three phases for supplier selection. Before proceeding to various phases the various factors for
supplier selection is considered in the phase one the interrelation ship among factors is considered. In the second
phase the weights among the factors considered. In the third phase linear physical programming is considered.

4.1 Mathematical model:

the general form of AHP can be depicted as shown in Figure2.The ahp procedure involves devising the

Fig 2 :ahp diagram

Hierarchical system by decomposing the problem and compare the comparative weight between the attributes of the
decision elements using satty score. calculation of the relative weight estimation and determine the best alternatives
after aggregation. If we wish to compare a set of attributes pair wise according to their relative weights
(importance), where the weights (Wij) ratio of pair wise compared value value between i and j ,are
W = [wij]n×n , --------------------- [8]
Where Wij = wij-1, wij= wikwkj ,andwij= wi/wj.
1 1 1

1
1 1

= = =
1

Or(w-nI)w = 0 --------[9]
The approximate weights we can find by calculating the eigen vector w with respect to which satisfies
Aw = w ----[10]
Where is the largest eigenvalue of the matrix A. in addition , calculate the consistency indexes (C.I.) to check
−
if the consistency condition is almost satisfied for A:C.I = Where the largest eigenvalue
−1
...11
and n denotes the numbers of the attributes. Satty suggested that the value of the C.I. should not exceed 0.1 for a
confident result. On the other hand as per .Liu, Hsiang [22] the problem of deriving the relative weights among
criteria in the AHP is equivalent to solving the following mathematical programming to obtain wi:

Min ∑ =1 −

s.t ∑ = 1, ∀1 ≤ ≤ , (12)
Where . denotes the p-norm and p {1,2,….}..
 Karuna kumar .G et al./ Materials Today: Proceedings 5 (2018) 12161–12173 12165

4.2. Correlated AHP :

Criteria 1 Criteria 2

Sub Sub
criteria 1 criteria 3

Sub Sub
criteria 2 criteria 4

Fig 3 : Relation between criteria

As shown in Figure3, it can be seen that Criteria 1 and 2 are considered to affect the decision of the problem in
order to consider the correlation effect in the AHP, we should first quantify the correlation matrix between criteria.
Ex correlation matrix R = correlation between i and jIn addition, it should be highlighted that R == R , ∀ ,

R= = , = , = , =

because the correlation effect is symmetric .Then, we assume that if Criterion is highly correlated to Criterion ,
they have similar weights or influence to the problem. Hence, if we obtain the correlation matrix between criteria,
we can objective to maximize the correlation, that is, ′ Rw.if there is no correlation it will be 0.

5 .Biobjectiveproblem( as per Hsiang-Hsi Liu,[22])

Max wʹRw ,
min − + − (13)

s.t.∑ =1 (14)

Where denotes the given estimated weight ratio of the upper-level Criteria and , denotes the true weight of
the upper-level th criterion, In the upper-level th criterion can be divided into lower-level criteria

6.Linear physical programming(lpp)

Linear Physical Programming (LPP), as a multi-objective optimization method, Developed by Messac et al. [24],
LPP simplifies physical programming procedure by defining preference functions as piece-wise linear functions
[10]. LPP has been successfully applied to different multi-objective problems ma etc[25].LPP has the ability to
avoid the weight assignment by Providing a preference function.• DM(decision maker) determines a suitable
preference function and specifies ranges of different degrees of desirability(ideal, desirable, tolerable, undesirable,
highly undesirable, and unacceptable) for each criterion the physical programming algorithm requires that the
decision maker expresses his/her preferences with respect to each criterion using one of the eight different classes.
The first four classes are “Soft class function “. The later four are “Hard class functions”.
12166 Karuna kumar .G et al / Materials Today: Proceedings 5 (2018) 12161–12173

Table 1: Ranges and corresponding constraints for the problem

Soft Classes Hard Classes
Class-1S Smaller is better Minimization Class-1H smaller gu<tu,max
Class-2S Larger is better Maximization Class-2H larger gu>tu,min
Class-3S Value is better Target optimization Class-3H equal gu = tu,max
Class-4S Range is better Range optimization Class-4H in range tu,max<gu<tu,min
The properties of class functions are as follows
Alower value of a class function which is always positive ,continous ,piecewise,linear and convex

Table 2: Class Functions

Class 1-S Class 2-S

Range Preference Constraint(tolerance Range Preference Constraint(tolerance
Index Level values) Index Level values)
1 Ideal ≤ +1 1 Ideal ≥ −1
+ + −
2 Desirable 1 ≤ ≤ 2 2 Desirable 1 ≤ ≤ −2
+ + −
3 Tolerable 2 ≤ ≤ 3 3 Tolerable 2 ≤ ≤ −3
+ + −
4 Undesirable 3 ≤ ≤ 4 4 Undesirable 3 ≤ ≤ −4
Highly Highly
+
−
4 ≤ ≤ −5
5 Undesirable 4 ≤ ≤ +5 5 Undesirable
−
+ 5 ≥
6 Unacceptable 5 ≤ 6 Unacceptable

Table 1 demonstrates the ranges and corresponding constraints for the problem gu= goaltu = target value of
u th criteria preferred over a higher value thereof. The value of a class function, zu, at given range-intersection (say,
desirable-tolerable) is the same for any class-type as seen below
≡ ( ) ≡ ( )∀ ; (2 ≤ ≤ 5); ≡0
--------------------- (15)

The magnitude of the class function’s vertical excursion across any range must satisfy the One vs. Others criteria-
rule (OVO-rule).The OVO-rule entails inter-criteria preference for each soft criterion, guThe change inzu across the
sth range is given by
≡ 2 − −1 ; (2 ≤ ≤ 5); 1
≡ 0 ---------------------------- (16)

Therefore, OVO rule can be expessed mathematically by

≡ β( − 1) −1 ; (3 ≤ ≤ 5); ( > 1); β > 1, --------------- (17)
where nsc denotes the number of soft criteria, and  denotes the convexity parameter. Basically the larger the  value
we have, the more convexity the class function we get. one needs a given initial value of z2. In practice, small
positive value, such as 0.1, is appropriate. Considering the convexity requirement parameter, we define the sth
range of the uth criterion as
̃ + = + − +( −1) ; ̃ − = − − −( −1) ; (2 ≤ ≤ 5);s : soft criteria ------ (18)
With this quantity, the magnitude of the slopes of the class function of the generic uth criterion (wus) can be
calculated as
+ −
+
= / ̃ ; − = / ̃ (2 ≤ ≤ 5);s : soft criteria ------------------ (19)
Once the slopes are known, the convexity requirement can be verified by the relationship
+ −
min = min( , ) > 0; (2 ≤ ≤ 5);s : soft criteria ------------------ (20)
,
 Karuna kumar .G et al./ Materials Today: Proceedings 5 (2018) 12161–12173 12167

+ −
Where = + − +( −1) ; = − − −( −1) ; + = − = 0; (2 ≤ ≤ 5);s : soft criteria ----(21)
+ −
It is known that as long as all weight values, 1 and 1 , are positive (or min is positive), the class function will be
piecewise linear and convex. In order to be more descriptive, these weights are then normalized. Normalized
weights can be represented mathematically as
+ + + −
( −1) )/ ∑ =2( ( −1) )/ ∑ =2(
+ 5 + − − 5 − −
=( − − ( −1) ); =( − − ( −1) ),
----(22)
+ −
1 = 1 = 0; (2 ≤ ≤ 5);s : soft criteria
These weights are applied in the LPP model. the magnitude of the convexity parameter  can be increased till
convexity is satisfied.. the importance of linear physical programming is explained by Messac A.,etc[26].The Linear
Physical Programming Weight (LPPW) algorithm depicted by Imtanavanich( 19) is outlined as follows quote “
1. Initialize:  = 1.1, +1 = 0, −1 = 0, 2 = 0.1u = 0, s = 0, nsc = number of criteria
2. Set u = u + 1 ;
3. Set s = s + 1
+ − + −
4. Evaluate (in sequence): , ̃ , ̃ , + , , and min . If min turns out to be a negative value, then
increase  (eg. increase by 0.1), rest u and s value, and go to step 2.
5. If s 5 then go to step 36 ..If unsc then go to step 2
For each criterion, the DM determines which one of the four soft and hard classes; and desirability range limits
are defined. LPPW algorithm is used to generate the weights .”unquote
In brief the methodology applied here. a)Calculation of ahp weights to check the consistency b)Calculation of
each supplier scores using correlated ahp Formulation of lpp problem
+ + − −
Objective functions −min +
= ∑ =1 ∑5=2( + )------(23)
, ,
+ +
s.t. − ≤ +( −1) ; ≥ 0; ≤ +5 for all u in class 1S, 3S and 4S, -------(24)
u = 1, 2, …,nsc, s = 2, …, 5
+ − ≤ −( −1) ; − ≥ 0; ≤ −5 for all u in class 2S, 3S and 4S, ----(25)
u = 1, 2, …,nsc, s = 2, …, 5
and gvtv,max for all v in classes 1H, v = 1, 2, …, nhc-------------------------------------------------(26)
gvtv,min for all v in classes 2H, v = 1, 2, …, nhc ------------------------------------------(27)
gv = tv,value for all v in classes 3H, v = 1, 2, …, nhc -------------------------------------------(28)
tv,mingvtv,min for all v in classes 4H, v = 1, 2, …, nhc---------------------------------------------(29)
xminxxmax--------------------------------------------------(30)
where x = quantity to be ordered ; = quantity to be ordered on ith supplier;gsxi = global score of ith supplier
ismx = individual score of f selected factor of supplier m
∑ ∗ >target value of global score; ----------------------------------(30)
∑ ∗ >target value of individual score; for all m =1 ,2,…..n ----(31)

7 .Model:

A study was conducted to identify the key selection criteria in a manufacturing industry in order to place orders on
each manufacturer called as supplier and constructed a decision hierarchy, identified the factors whether they
influence each other to find the correlation matrix between the criteria. Normalized weights of factors are found
considering the correlation by multi objective programming. The scores among the suppliers using correlated AHP
are found. Finally the quantity to be ordered on each supplier using LPP taking into consideration of target level
score obtained.
12168 Karuna kumar .G et al / Materials Today: Proceedings 5 (2018) 12161–12173

Phase one: The selection of criteria scores are obtained as per satty guide lines
7.1.AHP weight calculation: The meaning of the terminology used.
 Operation speed: . It indicates generally which supplier will deliver fast because of the process capability
 Operating Readiness: Preparation of stores which can be used straightaway without a bit of damage.
 Operation accuracy: It includes many aspects like adherence to transportation time, on-time delivery
 Order processing: Order processing starts from picking, packaging and packed items delivery
 Operating cost: Operating costs are the expenses for conducting the a business or facility
 Storage cost &Transportation cost: it includes the cost of moving and storing possessions
 Information technology: it is the application of computers and data acquisition and data management for
conduct of business or other manufacturing and allied activities.
 Storage Technology: the technology implemented for storage
 Transportation technology: the technology implemented for transporting
 Customer satisfaction: It is a measure of how goods and services supplied by a company
 Compatibility : it is the ability of the manufacturer, its vendors and their customers work in collaboration
 Financial easiness: Ensures continuity in services., better cash flow, sound balance sheet are indicators

Table3 : Factors of suppliers selection

C1. Operation speed
Operating efficiency (B1): C2. Operating readiness
C3. Operation accuracy
C4. Transportation cost
Cost(B2) C5. Storage cost
C6. Order processing cost
C7. Information technology
Technology level satisfaction (B3): C8. Storage technology
C9. Transportation technology
C10. Customer satisfaction
Service Quality (B4) C11. Compatibility
C12.Financial easiness
.
Table 4: Relation between various factors(first level) Table5: column sum of the various factors
B1 B2 B3 B4 A B1 B2 B3 B4
B1 1 2 3 2 B1 1 2 3 2
B2 ½ 1 2 1 B2 ½ 1 2 1
B3 1/3 ½ 1 ½ B3 1/3 ½ 1 ½
B4 ½ 1 2 1 B4 ½ 1 2 1
SUM 2.3333 4.5 8 4.5

operating efficiency, cost, technology level and service quality are represented by B1, B2, B3 and B4 respectively.
Step 1:column sum s( table 5). After considering the relations between different aspects .the column Sum for the
normalization purpose is performed.,ex Sum 1 = B11+B21+B31+B41 = 1+0.5+0.333+0.5 = 2.3333
Step2: column normalization see table 6: B11= 1 2.3333=0.4285; … the remaining all calculated.
 Karuna kumar .G et al./ Materials Today: Proceedings 5 (2018) 12161–12173 12169

Table 6: column normalization of factors Table7: Row sum of the various factors
A B1 B2 B3 B4 A B1 B2 B3 B4 SUM
B1 0.4285 0.4444 0.375 0.4444 B1 0.4285 0.4444 0.375 0.4444 1.6923
B2 0.2142 0.2222 0.25 0.2222 B2 0.2142 0.2222 0.25 0.2222 0.9086
B3 0.1428 0.1111 0.125 0.1111 B3 0.1428 0.1111 0.125 0.1111 0.49
B4 0.2142 0.2222 0.25 0.2222 B4 0.2142 0.2222 0.25 0.2222 0.9086
TOTAL 3.9995

Step3: Row sums (table 7):row wise totalling ex Sum1 = (0.4285+0.4444+0.375+0.4444) = 1.6923 .
Step4: (table 8). The individual row sums are divided by the total sum to get weights.W1 = s1/s = 1.69/3.99 = 0.4231
Step5:Consistencyindex:The consistency index (CI) measures the consistency set of data. consistence ratio (cr).oo4
which is acceptable for first level. Ahp weights are considered for next level B1,B2,B3,B4Phase 2:7.2 : Correlation
Matrix :By considering individual operation speed and corresponding Order processing cost rate are in correlation
with each other (table 14).The correlation is calculated as mentioned below. The correlation coefficient is r =
.
= =0.213 ≈ 0.2all correlation coefficients(table13)
( )( ) ( )( )
.3 multi objective programming: Multi Objectives function is performed to find out correlation weights
Min n= -((.2 * w11 * w21) +(.3 * w11 * w22 ) +(.2 *w11 *w23 ) +( .4 *w12 * w21) +(.5 *w12 * w22) + (.3 * w12 * w23) + (.2 *w13 * w21)
+(.4 *w13 *w22 ) + (.3 *w13 *w23));

Table8: Row normalization of the factors Table 9: Relation between internal factors of B1
A B1 B2 B3 B4 Weights
B1 0.4285 0.4444 0.375 0.4444 0.4231 B1 C1 C2 C3 Normal ahp
Weights
B2 0.2142 0.2222 0.25 0.2222 0.2271
B3 0.1428 0.1111 0.125 0.1111 0.1225 C1 1 3 2 0.55
B4 0.2142 0.2222 0.25 0.2222 0.2271 C2 1/3 1 1 0.21
C3 ½ 1 1 0.2422

. Table 9the consistency ratio are found in order Cr = .001

Table 10: Relation between internal factors of B2 Table 11: Relation between internal factors of B3

B2 C4 C5 C6 Normal
ahp B3 C7 C8 C9 W
Weights C7 1 .25 2 0.20
C4 1 2 3 0.55 C8 4 1 4 0.66
C5 1/2 1 1 .25 C9 .5 1/4 1 0.14
C6 1/3 1 1 .20

The table 10 consistency ratio are found in order Cr =.001Table 11 the consistency ratio found in order . Cr =.05

Table 12: Relation between internal factors of B4 Table 13 correlation table

B4 C10 C11 C12 Normal
C4 C5 C6
C10 1 .2 2 .19
C1 .2 .3 .2
C11 5 1 3 0.69
C2 .4 .5 .3
C12 ½ 1/3 1 0.12
C3 .2 .4 .3
The consistency ratio found in order .cr =.09
Table 14. correlation data for c1 and c6

C1 C6 C1 C6 C1 C6
Operation Order Operation Order Operation Order
Serial Serial Serial
Speed Processing Speed Processing Speed Processing
No. No. No.
(Days) Cost (Rs) (Days) Cost (Rs) (Days) Cost (Rs)
1 3 4 7 4 11 13 4 8
2 4 5 8 3.5 8 14 5 9
3 5 6 9 4.5 9 15 3 11
4 4 7 10 3 4 16 4 10
5 5.5 10 11 4 5 17 3 7
6 3 9 12 5 7 18 4.5 8
12170 Karuna kumar .G et al / Materials Today: Proceedings 5 (2018) 12161–12173

Min m = - 2− + 3− + 2− + 1 2− + 2− + 1− + 1 3− + 1 2− + 1 2− +
1 1 1
2− + 1− + 2− + 3− + 2− + 3− + 1− + 2− + 1− + 2− +

3− + 1 2− + 1− + 1 3− + 1− + 0.25 − + 2− + 4− + 4− + 1
2− +

0.25 − + 0.2 + 2− + 5− + 3− + 1 1 W1+w2+w3+w4=1; W1 =w11 + w12

2− + 3−
+ w13; W2 =w21 +w22 +w23; W3 =w31 +w32 +w33;
w4 =w41+w42 +w43;W1 > =0; w2 >=0; w3 >= 0; w4 >=0;W11 > =0; w12 >=0; w13 >= 0;W21 > =0; w22 >=0;
w23 >= 0; W31 > =0; w32 >=0; w33 >= 0;W41 > =0; w42 >=0; w43 >= 0; Where w1 ,w2,w3,w4 are weights of
b1,b2.b3,b4 respectively . and w11,w12,w13 w21,w22,w23,w31,w32,w33,w41,w42,w43are weights of
c1,c2,c3,c4,c5,c6,c7,c8,c9,c10,c11,c12,c23 respectively.

Table 15.The results of correlated weights

Top-Level Second Second Second
Weights Weights Weights Weights
Factors Level Level Level
B1 0.1142675 c1 0.062939 c5 0.04732 c9 0.029755
B2 0.1421024 c2 0.023208 c6 0.071051 c10 0.18665
B3 0.1814302 c3 0.027676 c7 0.060235 c11 0.2811
B4 0.5621999 c4 0.023731 c8 0.090715 c12 0.092201
The table(15) clearly demonstrates the correlated ahp values are different from normal ahp values. First normal ahp
is performed to check the consistency.

7.4 supplier wise ranks are calculated.

Each supplier is given weight with respect to each factor. On a scale of 1 ---10.Supplier weights are calculatedin
table 16 and 17

Table 16 Supplier global weighed scores

Phase 3
7.5 linear physical programming: the data considered for this part of section is mentioned in tables18,19,20.
7.6 Formulation of equation
Operating efficiency Goal = g1 = 0 .78 x1 +0.81 x2 +0.80 x3 + 0.81 x4
Technology satisfaction Goal = g2= 0.84 x1 +0.82 x2 +0.82 x3 +0.85 x4
service quality Goal = g3 = 0.7 x1 +0.7 x2 +0.5 x3 +0.6x4
cost Goal = g4 = 110 x1 +150 x2 +145 x3 +120 x4
Subject to Total quantity to be procuredx1 + x2 + x3 + x4= 1500 ;
the maximum limit that can be procured from supplier
 Karuna kumar .G et al./ Materials Today: Proceedings 5 (2018) 12161–12173 12171

x1<800 , x2<500 , x3<700 , x4<600 ; x1 ≥0 , x2 ≥0; , x3 ≥0 , x4 ≥0

x1 , x2, x3, x4 are the quantities to be ordered on suppliers1,2,3 and 4
5.5.2Based on linear physical programming, the equations are reformulated
Objective function Min z = ∑ ∗ + * + *
Where , = weights calculated as per weighted algorithm
+ 2
Step 1:  = 1.1, wt1 = 0, w− 1 ;
t1 = 0, z = 0.1; z =0 where t =1;U =0; s= 0; nsc = no of soft criteria =3;
Step 2 .set u= u+1 = 1 ;Step 3. set s = s+1 =1;Step 4 . ̌ 3 =  *(nsc-1) * ̌ 2=(1.1 *2 *.1) =2.2
̌ 4 =  *(nsc-1) * ̌ 3=(1.1 *2 *.2.2) =4.84
̌ 5 =  *(nsc-1) * ̌ 4=(1.1 *2 *.4.84) =10.648Step 5. _̌ = - = 50 ; ̌_ = 200 ; ̌_ = 200; _̌ =200;Step 6. _

= ̌ 2 / ̌_ =.1/50 =.002; _ = 2.2/200=.011 ; _ =4.84/200=.0242 ; _ =10.648/200 =.05324; = _ - _

=(2-
_ _ _ _
0)/1000 =.002 ; = - =.011-.002 =.009 ; = - =.0132 ; =(.053 -.024 ) =.0294
Since all values are greater than 0 the value are normalized . =. 04 ; = .17; =.24 ; =.55;
Similarly all other values. =. 06 ; = .2; =.14 ; =.6; =. 1; = .21; =.29 ; =.4;
Table 17 .Supplier weighed scores of individual factors

Table 18 .Data of product

Suppliers Cost Operating efficiency Technology satisfaction service quality Capacity in
(rupees) fraction fraction (fraction) numbers
Supplier 1 110 .78 .84 .7 800
Supplier2 150 .81 .82 .7 500
Supplier 3 145 .80 .82 .5 700
Supplier 4 120 .81 .85 .6 600

Table 19 Soft criteria

goals ideal desirable tolerable undesirable Highly undesirable unacceptable
operating efficiency >950 900-950 700-900 500-700 300-500 <300
Technology satisfaction >900 800-900 700-800 600-700 300-600 <300
service quality >900 800-900 700-800 500-700 400-500 <400

Table 20 Hard criteria

goals unacceptable acceptable
cost >200000 <200000

Goal constraints: Soft constraints for class 2s function (maximum is better)

g1 + ≥950;g1 + ≥900;g1 + ≥ 700;g1 + ≥ 500;g1≥ 300g2 + ≥900;g2 + ≥ 800;
g2 + ≥700;g2 + ≥600;g2≥ 300 ;g3 + ≥900;g3 + ≥800;g3 + ≥ 700;g3 + ≥500;
g3≥400;
, , , , , , , , , , all are deviation variables greater than zero..
Hard constraint; The total cost shall be less than 200000 rupees; G4 <200000;
12172 Karuna kumar .G et al / Materials Today: Proceedings 5 (2018) 12161–12173

The ahp total score of supplier and sub factor scores shall be greater than boundary limits.

X1* . 0.2651 + x2 *0.26134+x3 *0.21553+x4 *00.25804>300;

X1 * 0.2214+ x2 *0.13838+x3 *0.19373+x4 *0.166054>250;
X1 * 0.35526+ x2 *0.56841+x3 *0.42631+x4 *0.568408>785;
X1 * 0.20829 + x2 *0.42631+x3 *0.17853+x4 *0.23804>430;
X1 * 0.64541+ x2 *0.64541+x3 *0.46101+x4 *0.553206>865;

Total quantity to be ordered is 1500 units; iex1 + x2 + x3 + x4= 1500

the limitation of production capacityx1<800 , x2<500 , x3<700 , x4<600 ; x1 ≥0 , x2 ≥0; , x3 ≥0 , x4 ≥0 ;

Table 21. Results: The quantities to be ordered against each supplier and goal values achieved are mentioned
x1 x2 x3 x4 goal1 goal2 goal3 goal 4
143 500 257 600 1208 1250 938 199595

8. Conclusion

The problem is solved using software lingo 11 Multi objective technique is used to find out the correlated
weights of criteria in supplier selection. Implementation of linear physical programming technique which has the
capability to represent decision maker preference by using a utility function and to manage problem in multi criteria
environment for order allocation is presented. The study gives ample scope for future: such as the model can be
further extended accommodating more variables as power requirements, infrastructure requirements, product
recycling etc. this can be extended to new areas with fuzziness in consideration

References

[1] Chen, C. T., Lin, C. T., & Huang, S. F. (2006). A fuzzy approach for supplier evaluation & selection in supply chain management.
International Journal of Production Economics, 102, 289–301.
[2] Ghodsypour, S. H., &O’brien, C. (2001). The total cost of logistics in supplier selection, under conditions of multiple sourcing, criteria and
capacity constraint. International Journal of Production Economics, 73(1), 15–27.
[3] Hwang, C.L., and Yoon, K., (1981) Multiple Attribute Decision Making Methods and Applications: A State of the Art Survey, Springer-
Verlag, USA
[4] Saaty, T.L., (1980)The Analytic Hierarchy Process, McGraw-Hill, New York, USA.
[5] Dickson,G.W.(1966)“An AnalysisofVendorSelection Systems and Decision”.Journal ofPurchasingVol.2(1), 5-17.
[6] Weber, C.A., Current J.R. and Benton, W.C. (1991) “Vendor Selection Criteria and Methods”,European Journal of Operational Research
Vol.50(1), 2-18.
[7] De Boer, L., Labro, E. and Morlacchi, P.,(2001) “A Review of Methods Supporting SuppliersSelection”,European Journal of Purchasing
and Supply Management Vol. 7(2), 75-89
[8] Sanayei, A., Mousavi, S.F. and Yazdankhak, A., (2010) “Group Decision Making Process forSuppliers Selection with VIKOR Under Fuzzy
Environment”, Expert Systems with ApplicationsVol. 37 (1), 24-30.
[9] Ayhan, M.B. (2013). Fuzzy Topsis application for supplier selection problem. International
Journal of Information, Business and Management, Vol. 5(2), 159-174.
[10] Cheraghi, S. H., Dadashzadeh,M., & Subramanian, M., (2004) “Critical success factors forSupplier selection: An Update”, Journal of
Applied Business Research, Vol 20(2), 91–108.
[11] Ghodsypour, S.H., and O‟Brien, C., (1998) “A Decision Support System for Supplier Selection Using an Integrated Analytic Hierarchy
Process and Linear Programming”, International Journal of Production Economics, Vol. 56-57(20), 199-212..
[12] Kilic, H.S., (2013) “An integrated approach for supplier selection in multi item/multi supplier
environment”, Applied Mathematical Modelling, Vol. 37 (14-15), 7752-7763.
[13] Xia, W. and Wu, Z., (2007) “Supplier selection with multiple criteria in volume discountenvironments”, Omega, Vol. 35(5), 494-504..
[14] Yahya, S. and Kingsman, B., (1999) “Vendor Rating for an Entrepreneur DevelopmentProgramme: A Case Study Using the Analytic
Hierarchy Process Method”, Journal of theOperational Research Society Vol.50: 916-930.
[15] Lambert, A. J. D. and Gupta, S. M., "Disassembly Modeling for Assembly, Maintenance, Reuse, and Recycling," CRC Press, Boca Raton,
Florida, ISBN: 1-57444-334-8, 2005.
[16] Gungor, A. and Gupta, S. M., "Issues in environmentally conscious manufacturing and product recovery: a survey," Computers &
Industrial Engineering, Vol. 36, No. 4, 811-853, 1999.
[17] Kongar, E. and Gupta, S. M., "A multi-criteria decision making approach for disassembly-to-order systems," Journal of Electronics
Manufacturing, Vol. 11, No. 2, 171-183, 2002.
[18] Imtanavanich, P. and Gupta, S. M., "Calculating disassembly yields in a multi-criteria decision making environment for a disassembly-to-
order system," Proceedings of the 2005 Northeast Decision Sciences Institute Conference, Philadelphia, Pennsylvania, 2005.
 Karuna kumar .G et al./ Materials Today: Proceedings 5 (2018) 12161–12173 12173

[19] Imtanavanich, P. and Gupta, S. M., "Evolutionary computation with linear physical programming for solving a disassembly-to-order
system," Proceedings of the SPIE International Conference on Environmentally Conscious Manufacturing VI, Boston, Massachusetts, 30-
41, 2006.
[20] Massoud, A. Z. and Gupta, S. M., "Solving the multi-period disassembly-to-order system under stochastic yields, limited supply, and
quantity discount," Proceedings of 2008 ASME International Mechanical Engineering Congress and Exposition, Boston, MA, 2008.
[21] Kongar, E. and Gupta, S. M., "Solving the disassembly-to-order problem using linear physical programming," International Journal of
Mathematics in Operational Research, Vol. 1, No. 4, 504-531, 2009.
[22] Hsiang-Hsi Liu, Yeong-YuhYeh, and Jih-JengHuang(2014) “Correlated Analytic Hierarchy Process” Mathematical Problems in
EngineeringVolume 2014, Article ID 961714
[23] Fusunkucukbay and CeyhunAraz(2016) “ port folio selection problem –a comparison of fuzzy goal programming and linear physical
programming. an international journal of optimization and control theories and application vol 6 no 2 pp121-128
[24] Messac, A.; Gupta, S.; Akbulut, B. Linear Physical Programming: A New Approach to Multiple Objective Optimization. // Transactions on
Operational Research. 8, (1996), pp. 3959.
[25] Ma, X.; Dong, B. Linear Physical Programming-Based Approach for Web Service Selection. // Proceedings of the International Conference
on Information Management, Innovation Management and Industrial Engineering, Taipei, Taiwan, (2008), pp. 398-401.
[26] MessacA.,Gupta . S .M and Akbulut , B ., Linear Physical programming Effective Optimization for complex linear systems, Transactions of
Operation research ,8(2),39-59 (1996).