_____________________ _____________________
Internal Advisor External Advisor
_____________________ _____________________
Examiner-1 Examiner-2
i
ABSTRACT
The AHP model provides a comprehensive and rational framework for structuring the
supplier selection problem, for representing and quantifying the criterion involved in
supplier selection, for relating those criterions to the overall goal, and for evaluating
alternate suppliers using both subjective as well as objective criterion more effectively.
The criterion are ranked and prioritized using the pair wise judgments made by the
evaluator, thus the relative weights of each criteria calculated by the synthesis of these
judgments have a strong mathematical background. Moreover, after the suppliers are
ranked, sensitivity analysis is performed to find the most ‘critical’ criteria and the
effect of changes in importance on the overall ranking of the alternatives.
A Visual Basic application has been also designed to provide flexibility and ease in
creating multiple hierarchies, performing pair wise comparisons, creating automated
supplier ranking forms and to use different evaluation techniques for subjective and
objective criteria.
ii
ACKNOWLEDGMENT
We would like to express our deep and sincere gratitude to our internal advisor,
Associate Professor Sayed M. Irfan, for his wide knowledge and logical way of
thinking which has been of great value for us, and for his detailed and constructive
comments, and for his important support throughout this work.
We are also deeply grateful to our external advisor, Mr Fahham Hasan Qaiser,
Assistant Manager (Buying), Crescent Steel & Allied Products, for his understanding,
encouraging and personal guidance has provided a good basis for the present work.
We also express our warm and sincere gratitude to project coordinator, Assistant
Professor Ali Zulqarnain, and Chairperson Prof Dr. Amir Iqbal, Industrial &
Manufacturing Department, for their utmost support and co-operation and providing
help in difficult situations, without their support and motivation the fulfillment of the
project would have been difficult.
iii
TABLE OF CONTENTS
Chapter Page
Title
No. No.
1.INTRODUCTION………………………………………………………………..1
4. SENSITIVITY ANALYSIS……………………………………..………... 52
iv
4.1. SENSITIVITY ANALYSIS IN AHP 52
6. CONCLUSION………………………………………………………..…... 81
7. REFERENCES……………………………………………………………..83
8. APPENDIX………………………………………………………………… 85
v
LIST OF TABLES
Table Page
Table Caption
No. No.
vi
LIST OF FIGURES
Figure Page
Figure Caption
No. No.
1: Project Flow Chart………………………………………………….………..10
vii
20: Pair wise Comparisons Tab............................................................................72
viii
CHAPTER` 1
INTRODUCTION
To select the best supplier of Hot Rolled Steel Coils for Crescent Steel and Allied
Products Limited using the multi criteria approach of Analytic Hierarchy Process.
The selection has to be made among the three suppliers in the suppliers pool of
CSAPL, they are:
Hot Rolled Steel Coils are the basic raw material for steel pipe manufacturing. The
Hot Rolled Coils are de-coiled and formed to produce a cylindrical hollow body at a
pre-determined forming angle. Inside, and later, outside welding is performed by an
automatic submerged arc process [5].
The existing supplier selection method at CSAPL uses a weighted-point model. Nine
criteria are used for supplier ranking having an equal weight. Suppliers are rated using
a 1-5 scale and an average score is calculated. The major draw back of this method is
that all the criteria have an equal importance and the rating given to each supplier is
based purely on human judgment using a numeric scale.
1
1.2. ABOUT CRESCENT STEEL AND ALLIED PRODUCTS LIMITED
(CSAPL)
Crescent Steel and Allied Products Limited is a steel pipe manufacturing and coating
company which is listed on all the Stock Exchanges of Pakistan. The manufacturing
facilities consist of a Spiral Pipe Production line and a multi-layer Polyolefin, stand-
alone Epoxy Coating and Heat Shrink Tape Coating line, both located side by side at
the Sindh Industrial Trading Estate, Nooriabad in Jamshoro district of Sindh,
Pakistan.
The company has the unique distinction of having the authorization to use API
monogram of the American Petroleum Institute since its inception in 1987, the highest
international standard accredited for quality of steel line pipe. In 1997, the Company
was awarded ISO 9001 Quality Management Standard Certificate which it continues
to maintain as ISO 9001:2000. In addition, CSAPL has become the first Pakistani
company to have acquired oil and gas industry specific ISO 9000-2000 Quality
Management System certification from API [6].
Supplier selection and management is a key issue for any firm because the cost of raw
materials and component parts constitutes the main cost of a product, such that in
some cases it may account for up to 70% and most of the firms have to spend
considerable amount of their sales revenues on purchasing. In high technology firms,
purchased materials and services represent up to 80% of total product cost. Selecting
the right suppliers significantly reduces the purchasing cost and improves the
corporate competitiveness.
2
Supplier selection problem involves trade-offs among multiple criteria that involve
both quantitative and qualitative factors. Hence, supplier selection problem can be
modeled and solved by means of utilizing multi-criteria decision analysis [7].
Experts agree that no best way exists to evaluate and select suppliers, and thus
organizations use a variety of approaches. The overall objective of the supplier
selection and evaluation process is to reduce risk and maximize overall value to the
purchaser. An organization must select suppliers it can do business with over an
extended period of time [7].
The complexity of methods of selecting the best supplier is basically due to the
involvement of strategic decisions which need to incorporate tangible, intangible,
strategic, and operational factors into any analysis. A methodology that allows for the
synthesis of these factors and helps management structure the decision problem can
prove to be a valuable resource [8]. Few of the major Supplier Selection Techniques
along with their advantages and disadvantages are discussed in Table 1 [7]:
3
Cost Ratio • Subjectivity is reduced • Complexity and
• Flexibility requirement for a
developed cost
accounting system
• Performance Measures
(cost ratios) are
artificially expressed in
the same units
The basic characteristic of decision making is to choose the best option in a set of
competitive alternatives, which are evaluated under conflicting criteria. Analytic
Hierarchy Process (AHP) provides us with a comprehensive framework for solving
such problems. It was developed by Thomas L. Saaty in the 1970s. The AHP offers a
methodology to rank alternative courses of action based on the decision maker’s
judgments concerning the importance of the criteria and the extent to which they are
met by each alternative. For this reason, AHP is ideally suited for the supplier
selection problem.
AHP is designed to cope with the intuitive, the rational and the irrational, while
making multi-objective and multi-criterion decisions. It breaks down a problem into
its smaller constituent parts and then brings in simple pair-wise comparison
judgments to obtain the priorities in the hierarchy. There are three principles that one
4
can recognize in the problem solving. They are the principles of Decomposition,
Comparative Judgment and Synthesis of Priorities [9].
Among the many applications of AHP made by companies and governments, it has
been used by Xerox, IBM, British Airways, Ford Motor Company, Institute of
Strategic Studies in Pretoria, The Turkish Government, American Army and many
others [20].
5
1.5.3. The Fundamental Scale
Intensity of
Definition Explanation
Importance
2 Weak or slight
4 Moderate plus
6 Strong plus
6
above non zero numbers
assigned to it when
compared with activity j,
then j has the reciprocal
value when compared
with i
In order to determine the relative priorities or weights of the elements from the near
consistent pair wise positive reciprocal comparison matrix, a number of mathematical
methods can be used, as listed below [16]:
However Saaty justifies the use of Eigen value approach. This method is based upon
deriving the Eigen matrix of the comparison matrix which represents the priority
matrix [17]. With in the Eigen value approach several simplified methods have been
developed, we have used the Mean of Normalized value method which is also adopted
in many of the Operations Research text books and research papers [13; 18; 19]. The
mathematical basis of this method is beyond the scope of this report and therefore
only the steps to derive the priorities are given below.
7
1.5.5. Consistency
. .
.
. 1
.
Where,
C.I. is the consistency index of the comparison matrix. For a perfectly consistent
matrix the maximum positive Eigen value (λmax) will be equal to the order of the
matrix (n) [15], which increases with increasing inconsistency. Hence C.I. is the
measure of the deviation of λmax from n [13]. λmax can be calculated by first
multiplying the comparison matrix with the priority vector and then summing all
elements of the resultant vector which gives λmax or the maximum positive Eigen
value of the comparison matrix. The value of C.I. can be calculated as [15; 18; 19]:
. . 2
1
And R.I. is the Random Consistency Index or the average of the consistency indexes
of 50,000 randomly generated reciprocal matrices of a particular order [20]. The
values of R.I. are generalized using the formula [18]:
1.98 2
. . 3
In case the CR exceeds the acceptable limit of 0.1 then the decision maker is supposed
to identify the most inconsistent judgment, determine the range of values to which
that judgment can be changed corresponding to which the inconsistency will be
8
acceptable and consider if the judgment could be modified to a value in the acceptable
range [20].
In order to find the most inconsistent judgment we can make use of the fact that in a
perfectly consistent matrix (CR=0) each element (i, j) is given by Pi/Pj where Pi and Pj
are the elements of priority vector. Thus if ai, j is the element in ith row and jth column
of the comparison matrix then the ratio of each ai, j to its corresponding Pi/Pj shows the
level of inconsistency associated with that judgment. The greater this value, greater
will be the inconsistency [15]. Based on this the following formula could be derived
[15; 20]:
. 4
Where the value of represents the deviation from the consistent matrix due to the
comparison ai, j.
After studying the supplier selection procedure, Analytic Hierarchy Process and their
integration, the supplier selection process was broken down into smaller steps and
objectives and a detailed project flow chart was created as depicted in Figure 1. Based
on this a Gantt chart was also created.
9
Figure 1: Project Flow Chart
10
CHAPTER 2
The first step in formulating an AHP model is to structure the problem into the form
of a hierarchy with multiple levels. The first level of the hierarchy represents the
overall goal or the objective which in our case was to select the best supplier that suits
the requirements of CSAPL and provides maximum financial and long term benefits.
The second level of the Hierarchy contains the major factors that influence this goal.
The third level goes into much finer detail of each level 2 criterion. The final or fourth
level of the hierarchy contains the alternatives or the suppliers.
In order to define the primary criteria it was necessary to identify the basic
requirements of CSAPL regarding their raw material suppliers. This was done by
having detailed discussions with the buying personnel of CSAPL, Mr. Fahham Hasan
Qaiser (AM Buying); as a result the following areas of major importance were
identified.
• Cost
• Quality
11
• Delivery
• Financial Stability
• Technical Capability
• Supplier Relationship
These primary criteria were than further elaborated by defining their respective
secondary criterion. This initial hierarchy was discussed with the buying personnel of
CSAPL to identify any weaknesses and redundancy. As a result of these discussions a
number of criterion were added, removed and modified. This process was continued
until a final hierarchy was agreed upon. While selecting a criterion it was kept in
consideration that there was enough data and information available to compare and
evaluate suppliers on basis of that criterion and the meaning and purpose of each
criterion could be easily and clearly understood. Following is a brief description of all
the selected primary and secondary criterion.
2.1.1.Cost
Cost represents the price per unit, as well as the logistics cost and if the supplier is
willing to provide any payment flexibility including mode of payment, the amount of
advance payment required by the supplier, credit period etc. Based on this the primary
criterion was further divided into:
2.1.1.1.Unit Price
The actual cost per unit of the product as quoted in the quotation provided by the
supplier. This price also includes the cost of delivering the product to the port i.e.
logistics cost.
2.1.1.2.Payment Terms
Flexibility in payment terms provided by the supplier including mode of payment, the
amount of advance payment required by the supplier, credit period etc
2.1.2.Quality
12
past performance of deliveries also serves as an indicating factor to predict the quality
of future deliveries. The relationship between the supplier and firm does not end after
the shipment is received but in case of any quality problems the supplier will be
required to provide immediate and efficient response to these problems. Based on
these factors, Quality was further sub-divided into the following criterion:
2.1.2.1.Quality of Deliveries
Suppliers with ISO 9001 Quality Management System implemented are preferred by
CSAPL.
2.1.3.Delivery
Delivery employs that supplier must be able to deliver shipment on the pre negotiated
and agreed time. The delivery lead time is also critical to prevent any delays in
production due to unavailability of raw material. Delivery was sub divided into the
following level three criteria:
2.1.3.1.On-time Delivery
On-time delivery is the ability and past performance of the supplier to manufacture
and ship products on time.
Delivery lead time is the total time taken by supplier to manufacture and ship
products.
13
2.1.4.Financial Stability
The Financial Stability of the supplier is important to reduce risk of a supplier going
out of business and to select suppliers in sound financial conditions who are expected
to remain in business for the longer period of time. A supplier in poor financial
condition bore the risk of going out of business and not able to complete the order.
Most of the suppliers provide limited financial information due to its sensitivity and
confidentiality. Therefore the data available in supplier evaluation and rating forms
was used, this includes the total revenues and banking history.
2.1.4.1.Total Revenues
Total revenues provide information about supplier’s size and its total sales in the
recent years.
2.1.4.2.Banking History
Banking history indicates any problems in supplier’s transactions and cash inflow and
out flow.
2.1.5.Technical Capability
2.1.5.1.Annual Capacity
Annual capacity identifies the production capacity of the supplier and if it is sufficient
to meet the buyer’s requirements.
2.1.5.2.Technical Support
14
2.1.6.Supplier Relationship
2.1.6.1.Reciprocal Considerations
It is the understanding between the buyer and the supplier to provide each other
technical support and expertise from which both parties can benefit and to give
priority to their demands and needs.
2.1.6.2.Primary or Secondary
It is preferred to work with a primary supplier rather than a secondary one in order to
cut the middle man out. A primary supplier would have a better understanding of the
product and could provide better technical support.
Does working with CSAPL a financial interest for the supplier and will the size of its
order and business make CSAPL an important customer?
2.1.6.4.Supplier Response
It covers the interaction between the two firms. Does the supplier provide immediate
and complete information about the status of the order? How is the behavior and
attitude of supplier’s personnel?
In all six (6) primary (Level 2) and fifteen (15) secondary (Level 3) criterion were
selected. The number of primary and secondary criterion was kept to a minimum in
order to reduce the number of pair wise comparisons required to determine the weight
of each criterion. The final Hierarchy is shown in figure 2.
15
The proposed hierarchy covers almost all of the existing criteria of CSAPL while
adding a few new ones. A comparison of the existing and proposed criteria is shown
in Table 3.
7 Technical Innovativeness -
10 - Payment Terms
13 - Technical Support
15 - Supplier Response
16
Figure 2: The Supplier Selection Hierarchy
17
2.2. DETERMINATION OF CRITERION WEIGHTS
After structuring the hierarchy and finalization of primary and secondary criterion the
next step was to determine the relative weight or priority of each criterion. The method
employed for this purpose is laid down by Thomas L. Saaty in his books, The Analytic
Hierarchy Process (1980) and Decision making for Leaders: The Analytic Hierarchy
Process for Decision making in a Complex World (1982) and his various research papers
[13; 17; 20] and is also discussed in various Operations Research books [18; 19].
The core of this process is pair wise comparisons. In pair wise comparisons all criteria of
a cluster are compared against each other, two criteria at a time. The fundamental scale of
AHP as discussed in earlier chapter is used for making the comparison which converts
the semantic judgment of the decision maker into a predefined standard numerical value.
These values are fed into the positive pair wise reciprocal comparison matrix and solved
by using approximation to Eigen value method (Mean of Normalized Values Method)
discussed in Section 1.5.4 which extracts the numerical values of the priority or
importance of each criterion.
In order to facilitate the decision maker in performing the pair wise comparisons, MS
Excel spread sheets were developed for level 2 and level 3 of the hierarchy. The decision
maker could select the verbal judgment for the relative importance of one criterion as
compared to another from a drop down list. The preference of one criterion as compared
to another i.e. less preferred or more preferred could be selected from another drop down
list. The inputs are converted into the corresponding numerical values and the relative
weights and CR is calculated automatically by using the method discussed in Section
1.5.5.
In case the CR exceeds the acceptable limit of 0.1 then the decision maker is supposed to
identify the most inconsistent judgment, determine the range of values to which that
judgment can be changed corresponding to which the inconsistency will be acceptable
and consider if the judgment could be modified to a value in the acceptable range [20]. In
order to facilitate this process, an automated consistency chart was created which uses
real time inputs to identify the most inconsistent judgment and the range of acceptable
18
values of input for each comparison. It also serves as a tool for sensitivity analysis in
terms of CR to identify the effect of modifications in judgments upon the CR. The rank
of each comparison on basis of its inconsistency is also identified in the spread sheet (1
being most inconsistent). Green and Black colored cells represent input values
corresponding to each comparison, resulting in consistent or inconsistent matrix
respectively; yellow colored cells represent the current selection. In order to find the most
inconsistent judgment the method explained in Section 1.5.5 was used. A copy of this
form filled by buying personnel of CSAPL for level 2 and level 3 along with the
consistency chart is shown in Appendix.
2.2.1.Results of Comparisons
The spread sheets created for pair wise comparisons were provided to the buying
personnel of CSAPL. A brief introduction of the pair wise comparisons, fundamental
scale, consistency ratio etc was also provided in order to better understand the process.
After pair wise comparisons the following results were obtained for the Primary or Level
2 Criteria:
The obtained results reflected the priorities of the purchasing department of CSAPL and
were approved by the purchasing personnel of CSAPL. The results show that Financial
Stability (45.676%) is the most important criteria followed by Delivery (15.845%) and
Cost is the least preferred criterion (4.384%). This might appear contrary to the general
understanding that cost, quality and delivery are the most important criteria respectively.
However Price is regarded as only one of the elements in the marketing mix. Surveys
indicate that only in minority of transactions, price is the dominant factor. Price is the
19
most important factor only if all other things are equal (which is not often the case), or
when there are financial difficulties on part of the buyer [22]. But since in case of
purchase of raw materials by CSAPL the contracts are of million of dollars, thus the
financial stability of the supplier becomes extremely important to make sure that the
supplier is financially strong enough to complete the contract efficiently and stay in
business in order to develop a long term relationship.
Following are the calculations for criteria weights, consistency ratio and the degree of
inconsistency in the judgments. The calculations for determination of weights was based
on the formulas discussed earlier and are given in Sections 1.5.4 and 1.5.5.
• Cost (CC)
• Quality (Q)
• Delivery (D)
• Financial Stability (FS)
• Technical Capability (TC)
• Supplier Relationship (SR)
20
Technical
8 Quality is Strongly more preferred than
Capability
Supplier
9 Quality is Equally preferred as
Relationship
Financial
10 Delivery is Strongly less preferred than
Stability
Technical
11 Delivery is Strongly more preferred than
Capability
Supplier
12 Delivery is Equally preferred as
Relationship
Financial Technical
13 is Strongly more preferred than
Stability Capability
Financial is Moderately to Strongly more preferred Supplier
14
Stability than Relationship
Technical Supplier
15 is Strongly less preferred than
Capability Relationship
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Cost 0.0438
Quality 0.1418
Delivery 0.1585
21
Financial Stability 0.4568
Technical Capability 0.0532
Supplier Relationship 0.1459
0.2741783
0.9352507
(Pair wise Reciprocal 1.0229241
Comparison Matrix) x (Priority
Vector) = 3.0708695
0.3215035
0.9580888
In order to determine the comparisons with most inconsistency, the method defined in
Section 1.5.5 was used.
22
2.4. CALCULATIONS FOR WEIGHTS OF LEVEL 3 CRITERIA
Following are the calculations for criteria weights, consistency ratio and the degree of
inconsistency in the judgments. Calculations are performed using the method defined
earlier in Sections 1.5.4 and 1.5.5.
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
(UP) (PT)
(UP) 1 3
(PT) 0.3333333 1
(UP) (PT)
(UP) 0.7500 0.7500
(PT) 0.2500 0.2500
From the Normalized Matrix the weights of criteria can be calculated, which are:
23
Lambda max 2
Consistency Index 0
Random Consistency Index 0
Consistency Ratio 0
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
24
Response to Quality Problems 0.3333
Lambda max 3
Consistency Index 0
Random Consistency Index 0.66
Consistency Ratio 0
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
(OD) (DLT)
(OD) 1 4
(DLT) 0.25 1
(OD) (DLT)
(OD) 0.8000 0.8000
(DLT) 0.2000 0.2000
From the Normalized Matrix the weights of criteria can be calculated, which are:
25
On Time Delivery 0.8000
Delivery Lead Time 0.2000
Lambda max 2
Consistency Index 0
Random Consistency Index 0
Consistency Ratio 0
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
(TR) (BH)
(TR) 1 4
(BH) 0.25 1
(TR) (BH)
(TR) 0.8000 0.8000
(BH) 0.2000 0.2000
From the Normalized Matrix the weights of criteria can be calculated, which are:
26
Banking History 0.2000
Lambda max 2
Consistency Index 0
Random Consistency Index 0
Consistency Ratio 0
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
(AC) (TS)
(AC) 1 0.3333333
(TS) 3 1
(AC) (TS)
(AC) 0.2500 0.2500
(TS) 0.7500 0.7500
From the Normalized Matrix the weights of criteria can be calculated, which are:
27
Consistency of the Comparison Matrix can be calculated as follows:
Lambda max 2
Consistency Index 0
Random Consistency Index 0
Consistency Ratio 0
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
28
After Normalization of the Comparison Matrix we obtain the following Normalized
Matrix:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Table 4 shows the Local and Global weight of each criterion. The local weight represents
the contribution of each sub criterion to its respective upper level criterion, where as
global weight gives the contribution of each criterion to the overall objective i.e. selecting
the best supplier. The pie chart in Figure 3 represents the relative weights of primary
criteria, whereas the bar chart in Figure 4 shows the relative global weights of the
secondary criteria.
29
Table 4: Local and Global weights of Primary and Secondary Criteria
Local Global
S. No. Primary Criterion Secondary Criterion
Weight Weight
1 Financial Stability 45.676%
i Total Revenues 80% 36.541%
ii Banking History 20% 9.135%
2 Delivery 15.845%
i On time Delivery 80% 12.676%
ii Deliver Lead Time 20% 3.169%
3 Supplier Relationship 14.592%
i Supplier Response 53.18% 7.760%
ii Desire for Business 22.12% 3.228%
iii Reciprocal Considerations 19.15% 2.794%
iv Primary or Secondary 5.54% 0.809%
4 Quality 14.179%
i Quality of Deliveries 33.33% 4.726%
ii ISO Certification 33.33% 4.726%
Response to Quality
iii 33.33% 4.726%
Problems
5 Technical Capability 5.325%
i Technical Support 75% 3.994%
ii Annual Capacity 25% 1.331%
6 Cost 4.384%
i Unit Price 75% 3.288%
ii Payment Terms 25% 1.096%
30
Technical Cost
Capability 4.38%
5.32%
Quality Financial
14.18% Stability
45.68%
Supplier
Relationship
14.59%
Delivery
15.85%
40%
35%
30%
Global Weight
25%
20%
15%
10%
5%
0%
Primary or Secondary
Banking History
On time Delivery
Annual Capacity
Quality of Deliveries
Payment Terms
Technical Support
Unit Price
ISO Certification
Total Revenues
Supplier Response
Reciprocal Arrangement
31
CHAPTER 3
Supplier ranking is the process in which each supplier is assessed and evaluated on the
basis of each lowest level criterion. Some of these criteria were qualitative or subjective
whereas others were quantitative or objective therefore different techniques within the
scope of AHP were employed namely Pair wise Comparisons, Pair wise Comparisons
using Equivalent Scale and Direct Method [23]. Spread sheets were developed for
supplier evaluation and ranking which were provided to the buying personnel of CSAPL.
The excel spread sheets provided means for pair wise comparisons as well as direct input
of numeric data based on the nature of criterion. These inputs were then used in the
comparison matrices to calculate relative weights or performance of each supplier. These
weights were multiplied by the weight of their respective criterion and summed to give
the overall weight of the supplier. The fifteen lowest level criteria are shown in Table 5
along with their evaluation methods:
32
Since criteria like Payment Terms, Response to Quality Problems, Banking History,
Technical Support, Reciprocal Considerations, Desire for Business, Supplier Response,
Quality of Deliveries and On time Delivery are either purely or predominantly subjective
criterion or it was difficult to measure them quantitatively. Therefore the most
appropriate choice of evaluation method was pair wise comparisons. Two suppliers were
compared at a time on basis of each criterion, using the fundamental scale and their
relative performance was calculated.
Total Revenues and Annual Capacity were measured using actual data. The evaluation
method used is direct input. In the case of direct input of numeric data we use the ratio of
two numeric values in the comparison matrix [24] therefore for each comparison the ratio
between the corresponding values of each supplier were calculated and substituted in the
comparison matrix to calculate the weights. Total capacity and annual revenues both have
a direct relation with the objective.
In case of case of Unit Price and Delivery Lead Time, actual quantitative data is also used
but direct method is not suitable in this case since they are not purely objective. When we
use direct method, we are not concerned with the actual values but how large or small
they are relatively. In case of these two criterions this is not true since the actual unit
price and time taken for delivery is also important to the decision maker.
For example if two suppliers have quoted a price of 25,000 Rs and 50,000 Rs
respectively then the input in comparison matrix for this comparison will be 2. However
if the prices quoted are 100,000 Rs and 200,000 Rs respectively, the input in the matrix
will still be 2. But the absolute price difference in both cases is 25,000 Rs and 100,000 Rs
which does not have the same meaning for the decision maker. That’s why the use of
direct method for evaluating suppliers on basis of price is not suitable. Moreover the
decision to prefer one supplier over another on the basis of cost difference might be
effected by the firm’s financial conditions or decision maker’ s own judgment and
perception. Hence it was necessary to develop an equivalent scale in which each value
corresponds to an absolute price or time difference (in case of delivery lead time). This
also saves decision maker’s effort to compare each supplier, rather the decision maker
33
could respond to questions like “What price difference would make one supplier equally
preferred as another?” or “What price difference would make one supplier strongly
preferred over another?” and so on. The answers to these questions formed the equivalent
scale for the comparisons, the intermediate values were interpolated. Hence the absolute
price difference values corresponding to the 1-9 scale were developed and used for the
comparisons
The question of whether a supplier has ISO 9001 Certification will result in a YES or NO
answer but we must also know that how much a supplier with ISO 9001 Certification is
preferred over a supplier without the certification. The decision maker was asked this
question and the result was used in the pair wise comparisons. Since the resulting priority
is a relative weight therefore the weight of supplier depends not only whether the supplier
has ISO Certification or not but also on how many other suppliers are being compared
and how many of them have ISO Certification.
The data for supplier evaluation was provided by the buying personnel of CSAPL by
filling in the Supplier Evaluation Form developed for this purpose. A copy of this form is
shown in Appendix. The calculations performed for supplier evaluation are as follows:
34
Baosteel 765.00 $/unit
TISCO 750.00 $/unit
Hadeed 830.00 $/unit
The equivalent scale for Price per Unit was created by using the answers of the following
questions given by the decision maker.
What Price difference between two Suppliers will make;
Fundamental Scale in
terms of Absolute Cost
Difference
1 14.5
2 21.375
3 28.25
4 35.125
5 42
6 50.25
7 58.5
8 66.75
9 75
On the basis of the comparisons the following Comparison Matrix can be constructed
After Normalization of the Comparison Matrix we obtain the following Normalized Matrix:
35
Baosteel TISCO Hadeed
Baosteel 0.3200 0.3103 0.4444
TISCO 0.6400 0.6207 0.5000
Hadeed 0.0400 0.0690 0.0556
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.3583
TISCO 0.5869
Hadeed 0.0548
On basis of Payment Terms, the pair wise comparisons performed by the decision maker
are:
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.2409
TISCO 0.2106
Hadeed 0.5485
36
Lambda Max 3.0222
Consistency Index 0.0111
Random Consistency Index 0.66
Consistency Ratio 0.0168
On the basis of Quality of deliveries, the pair wise comparisons performed by the
decision maker are:
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.4737
TISCO 0.0526
Hadeed 0.4737
37
3.2.4. Calculations for “ISO 9001 Certification” Criterion
Baosteel Yes
TISCO Yes
Hadeed Yes
How much a Supplier with ISO 9001 Certification is preferred over a Supplier without ISO 9001
Certification?
Very Strongly to Extremely
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.3333
TISCO 0.3333
Hadeed 0.3333
38
3.2.5. Calculations for “Response to Quality Problems” Criterion
On the basis of Quality of deliveries, the pair wise comparisons performed by the
decision maker are:
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.4429
TISCO 0.1698
Hadeed 0.3873
On the basis of Quality of deliveries, the pair wise comparisons performed by the
decision maker are:
39
Baosteel is Strongly more preferred than TISCO
Baosteel is Moderately to Strongly more preferred than Hadeed
TISCO is Moderately less preferred than Hadeed
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.6651
TISCO 0.1038
Hadeed 0.2311
Baosteel 60 Days
TISCO 60 Days
Hadeed 35 Days
The equivalent scale for Price per Unit was created by using the answers of the following
questions given by the decision maker
40
What Delivery Lead Time difference between two Suppliers will make;
Fundamental Scale in
terms of difference in
delivery lead time (Days)
1 5
2 7.5
3 10
4 12.5
5 15
6 17.5
7 20
8 22.5
9 25
On the basis of the comparisons the following Comparison Matrix can be constructed
41
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.0909
TISCO 0.0909
Hadeed 0.8182
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.4992
TISCO 0.3429
Hadeed 0.1579
42
3.2.9. Calculations for “Banking History” Criterion
On the basis of Quality of deliveries, the pair wise comparisons performed by the
decision maker are:
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.4615
TISCO 0.0769
Hadeed 0.4615
Lambda Max 3
Consistency Index 0
Random Consistency Index 0.66
Consistency Ratio 0
43
Baosteel 20.00 Million Tons/Year
TISCO 10.00 Million Tons/Year
Hadeed 5.50 Million Tons/Year
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.5634
TISCO 0.2817
Hadeed 0.1549
On the basis of Quality of deliveries, the pair wise comparisons performed by the
decision maker are:
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
44
Baosteel TISCO Hadeed
Baosteel 1 3 2
TISCO 0.3333 1 0.3333
Hadeed 0.5 3 1
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.5247
TISCO 0.1416
Hadeed 0.3338
On the basis of Quality of deliveries, the pair wise comparisons performed by the
decision maker are:
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
45
After Normalization of the Comparison Matrix we obtain the following Normalized
Matrix:
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.3333
TISCO 0.3333
Hadeed 0.3333
Lambda Max 3
Consistency Index 0
Random Consistency Index 0.66
Consistency Ratio 0
Baosteel Primary
TISCO Primary
Hadeed Primary
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
46
Baosteel TISCO Hadeed
Baosteel 1 1 1
TISCO 1 1 1
Hadeed 1 1 1
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.3333
TISCO 0.3333
Hadeed 0.3333
On the basis of Quality of deliveries, the pair wise comparisons performed by the
decision maker are:
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
47
Hadeed 0.2000 0.2000 0.2000
From the Normalized Matrix the weights of criteria can be calculated, which are:
Baosteel 0.2000
TISCO 0.6000
Hadeed 0.2000
Lambda Max 3
Consistency Index 0
Random Consistency Index 0.66
Consistency Ratio 0
On the basis of Quality of deliveries, the pair wise comparisons performed by the
decision maker are:
The Pair wise Reciprocal Comparison Matrix for these comparisons can be expressed as:
From the Normalized Matrix the weights of criteria can be calculated, which are:
48
Baosteel 0.5889
TISCO 0.2519
Hadeed 0.1593
The over all performance of the three suppliers in terms of local and global weights for
the primary and secondary criterion is shown in Tables 6 and 7 and Figures 5 and 6.
Where A, B and C represents Baosteel, TISCO and Hadeed respectively. Analysis of the
results shows that Shanghai Baosteel Group Corporation (Baosteel) is the best supplier
with a considerable margin. The performance of Baosteel is better than the other three
suppliers in 5 out of 6 primary and 11 out of 15 secondary criteria. Hence Baosteel is the
recommended supplier. Table 6 gives the supplier performance on basis of each
primary criterion.
49
60%
SupplierPerformance 50%
40%
30%
20%
10%
0%
Cost Quality Delivery Financial Technical Supplier
Stability Capability Relationship
PrimaryCriteria
50
8 Total Revenues 36.54 49.92 34.29 15.79 18.24 12.53 5.77
9 Banking History 9.14 46.15 7.69 46.15 4.22 0.70 4.22
10 Annual Capacity 1.33 56.34 28.17 15.49 0.75 0.37 0.21
11 Technical Support 3.99 52.47 14.16 33.38 2.10 0.57 1.33
Reciprocal
12 2.79 33.33 33.33 33.33 0.93 0.93 0.93
Considerations
Primary or
13 0.81 33.33 33.33 33.33 0.27 0.27 0.27
Secondary
14 Desire for Business 3.23 20.00 60.00 20.00 0.65 1.94 0.65
15 Supplier Response 7.76 58.89 25.19 15.93 4.57 1.95 1.24
Total Weight 47.79 25.66 26.56
90%
80%
70%
Supplier Performance
60%
50%
40%
30%
20%
10%
0%
Secondary Criteria
SENSITIVITY ANALYSIS
Sensitivity analysis identifies the impact of changes in the priority of criteria on the
suppliers' performance. After obtaining the initial solution with the given weights of the
criteria, sensitivity analyses were carried out to explore the response of the overall
ranking of alternatives to the changes in the relative importance of each primary criterion.
This is important because a decision maker might be very interested to know the
consequences of variation in weights of criteria. Some influence could have been
underestimated or a slight variation in one criteria weight could lead to a completely
different decision. A comprehensive sensitivity analysis would ensure the robustness of
the system and eliminate risk of selecting the ‘wrong’ supplier because some
comparisons in the model were not made accurately [25].
Sensitivity Analysis was performed only on the Level 2 criteria so that the decision
maker is not over burdened with information and only the very important and critical
criteria are considered. In order to perform Sensitivity Analysis the weight of a criterion
is perturbed slightly at a time and its effect on the overall supplier ranking is observed. If
a slight variation in an important criterion results in rank reversal it shows that there is
uncertainty in the model. An increase in a criterion weight would result in a
corresponding decrease in the weight of remaining criteria in the same cluster and vice
versa, this change is proportional to the weight of these criteria. Therefore the priorities
of criteria after perturbation could be easily calculated using the following simple relation
[10]:
(5)
% 6
! " ! # ! $
52
% 7
! " ! # ! $
Where:
P1, P2 ... Pn = priorities of criterion 1, 2 and n after perturbation.
P1a, P2a ... Pna = actual priorities of criterion 1, 2 and n.
n = number of criterion.
x = amount of perturbation.
If weight of criterion 1 is increased, then the weights of all remaining criteria will be
decreased and vice versa.
The widely used AHP software Expert Choice™ provides different means of sensitivity
analysis. For our study we selected two methods, namely Performance Sensitivity and
Gradient Sensitivity analysis and developed MS Excel spreadsheets to perform these
analyses for our primary level criteria.
In performance sensitivity the decision maker has the freedom to modify the weights of
primary criteria and to observe their impact on overall ranking. In this way an infinite
combination of priorities can be analyzed and important what if questions can be
answered. In order to modify weights in the Excel spread sheets, form controls are
embedded with VBA coding which perturb the relative weights. Different degrees of
precision can also be selected. The results are displayed graphically in the form of charts.
Using the form controls on the performance sensitivity analysis spread sheet, the weight
of each criterion is maintained at 16.667%. In this situation the overall weights of
supplier performance are: 46.03%, 26.58% and 27.4% for Baosteel, Hadeed and TISCO
respectively. The resulting ranking is the same as the current ranking. Therefore it shows
that even when all the criterion have the same weight, Baosteel is clearly the best
supplier. Figure 7 shows the result for Scenario 1.
53
Figure 7: Performance Sensitivity Analysis for Scenario 1
We want to see what will happen if cost was the most important criterion rather than
Financial Stability, to find out we increase the overall weight of cost to 45%, the
remaining priorities are 24.84%, 15.27%, 6.45% ,6.09%, 3.25% for quality, delivery,
supplier relationship, financial stability and technical capability respectively. The result
shows that still Baosteel is the best supplier followed by TISCO and Hadeed with overall
performance of 40.64%, 32.75% and 26.6% respectively. The ranks of TISCO and
Hadeed have been interchanged but still Baosteel is the better supplier with quite a
considerable margin. Figure 8 shows the result for Scenario 2.
Similarly a number of different scenarios can be created and their results can be analyzed.
The two scenarios considered show that the system is not effected by small changes and
the best supplier remains to be Baosteel.
54
Figure 8: Performance Sensitivity Analysis for Scenario 2
In gradient sensitivity only one criterion is observed at a time. The weight of a certain
criteria is plotted on the x-axis and the overall supplier priority on the y-axis. The overall
weight of each supplier is represented by a straight line. This graph represents whether
the weight of supplier is increasing or decreasing with changing the weight of that
criterion. Most importantly it identifies the point at which rank reversal is taking place.
Individual graphs for each primary criterion were plotted. In order to plot gradient
sensitivity of a particular criterion, the relative weights of all the other criteria in the
cluster are calculated using eqns. 5 to 7, when that criterion has weights of 0% and 100%
respectively. These perturbed weights are used to calculate the overall supplier
performances and a graph is then plotted using these values.
Financial Stability is the most important Primary Criteria with a global weight of 45.68%.
At this weight Baosteel is the best supplier, followed by Hadeed and TISCO. The
sensitivy analysis in Figure 9 shows that no matter how much the weight of Financial
Stability is increased or decreased, Baosteel continues to be the best supplier. With
55
increasing importance of financial stability the overall performance of Baosteel also
improves. Also, if the weight is increased beyond 51.91%, TISCO will become more
preferred over Hadeed but still behind Baosteel. Hence on basis of financial stability
Baosteel is clearly the best supplier.
Table 8 shows the calculations, where F.S. stands for relative weight of Financial
Stability.
56
4.3.2. Gradient Sensitivity of Delivery Criterion:
Delivery is the second most important Primary Criteria with a global weight of 15.85%.
At this weight Baosteel is the best supplier, followed by Hadeed and TISCO. The
sensitivy analysis in Figure 10
1 shows that no matter how much the weight of Delivery
criterion is increased or decreased, Baosteel continues to be the best supplier. Also, if the
weight is reduced beyond 12.66% then TISCO will become more preferred over Hadeed
but still behind Baosteel. Hence on basis of Delivery Baosteel is clearly the best supplier.
Table 9 shows the calculations, where D stands for relative weight of Delivery.
57
4.3.3. Gradient Sensitivy of Supplier Relationship Criterion:
Supplier Relationship is the third most important Primary Criteria with a global weight of
14.59%.
%. At this weight Baosteel is the best supplier, followed by Hadeed and TISCO.
The sensitivy analysis in Figure
Fi 11 shows that no matter how much the weight of
Supplier Relationship criterion is increased or decreased, Baosteel continues to be the
best supplier.
lier. With increasing importance of Supplier Relationship the performance of
Baosteel decreases.. Also, if the weight is increaded beyond 19.76%
% then TISCO will
become more preferred over Hadeed but still behind Baosteel. Hence on basis of Supplier
Relationship Baosteel is clearly the best supplier.
Table 10 shows the calculations, where S.R. stands for relative weight of Supplier
Relationship
58
4 Quality 16.601 0.000
5 Technical Capability 6.234 0.000
6 Cost 5.133 0.000
Quality has a global weight of 14.18%. At this weight Baosteel is the best supplier,
followed by Hadeed and TISCO. The sensitivy analysis in Figure 12 shows that no matter
how much the weight of Supplier Relationship criterion is increased or decreased,
Baosteel
osteel continues to be the best supplier.
supplier Also, if the weight is reduced beyond 10.40%
then TISCO will become more preferred over Hadeed but still behind Baosteel. Hence
on basis of Quality Baosteel is clearly the best supplier.
59
4 Quality 0.000 1.000
5 Technical Capability 6.204 0.000
6 Cost 5.108 0.000
Table 12 shows the calculations, where T.C. stands for Technical Capability.
60
Table 12:: Calculations for Gradient Sensitivity of Technical Capability Criterion
Perturbed
S. Overall Supplier Performance
Criteria Weights
No.
T.C. = 0 T.C. = 1 Suppliers T.C. = 0 T.C. = 1
1 Financial Stability 48.245 0.000 Baosteel 0.475 0.534
2 Delivery 16.736 0.000 TISCO 0.261 0.177
3 Supplier Relationship 15.412 0.000 Hadeed 0.264 0.289
4 Quality 14.976 0.000
5 Technical Capability 0.000 1.000
6 Cost 4.630 0.000
The sensitivity analysis shows very clearly that the system is robust and small changes in
criteria weights does not affect the overall ranking of the suppliers and hence it confirms
that any inconsistency in the pair wise comparison phase resulting in slightly different
criteria weights does not affect the overall system. As a result we can say that Baosteel
(China) is the best supplier with a considerable margin and must be selected with full
confidence.
62
CHAPTER 5
In order to store the data entered by the user and to retrieve it for future use, the
application is linked with Microsoft Excel. MS Excel not only stores the data but also
performs various mathematical and logical operations thus saving coding space and
processing time in Visual Basic. All the information entered by the user is stored in an
MS Excel file with each work sheet assigned to store some particular information related
to each Hierarchy.
The Application is structured in the form of several windows associated for different
phases of the AHP process. After starting the application user will see a main window
from where he can navigate to the other windows. The application can be divided into
three major modules or windows:
63
2) Manage Hierarchies: Hierarchies can be added, deleted and edited
3) Evaluate Suppliers: Evaluate suppliers on the basis of lowest level criteria
The above mentioned three windows can only be opened, one at a time and can only be
invoked from the main window (main menu). The structure of the application is shown
graphically in Figure 15.
64
Splash Screen
Window 1.2.3.1:
Adjust Inconsistency Window 1.3.2.1:
Tab 1.2.4: View Results
Result
65
5.5 SOFT WARES USED
• Windows XP or higher
• MS EXCEL 2007 or higher
Following is a brief introduction of the interface along with description of each window
and tab and the functionality of the controls present on them:
After installation the Application can be invoked by selecting Start> All Programs>
Supplier Selection using AHP> Supplier Selection using AHP
Invoked by: Start> All Programs> Supplier Selection using AHP> Supplier Selection
using AHP
66
Figure 16: Splash Screen
The main menu shown in Figure 17, allows user to navigate through the different child
windows or modules. User can select from the following buttons:
67
Opens child window to edit/delete/create Hierarchies and calculate relative weight
of criteria by performing pair wise comparisons
3) Evaluate Suppliers and View Results:
Opens child window to Evaluate Suppliers (Alternatives) on the basis of the
lowest level criteria
4) About:
Opens about window showing Information about the application
5) What is AHP?
Opens child window showing Information about Analytic Hierarchy Process
Invoked by “Maintain Supplier Information” Button (Window 1), this Window shown
in Figure 18, allows user to maintain a data base of Suppliers by entering Supplier Name,
Contact Information and Product(s)/Service(s) Supplied. The suppliers added in this
module could be subsequently associated to a Hierarchy in Window 1.3.1. A maximum
of forty suppliers can be added. Database can be managed using the following buttons:
68
1. Add: To add a new supplier to the database
2. Remove: Removes the selected entry
3. Edit: Allows user to edit the selected entry
4. Clear All: Removes all Suppliers from the data base
5. Finish: Returns to Main Menu (Window 1)
Invoked by “Edit/Create Hierarchies” Button (Window 1), this module allows user to
manage Hierarchies by creating and editing new Hierarchies, adding criterion to a
Hierarchy and performing pair wise Comparisons. It consists of the following tabs:
Invoked by “Edit Create Hierarchies” Button (Window 1), this Tab shown in Figure 19,
allows user to specify and edit hierarchy name, date on which it was created, names of
people who performed the comparisons and notes related to the Hierarchy. The list box
displays the names of Hierarchies created. A maximum of four Hierarchies could be
created. The user can use the following buttons on this Tab:
69
Figure 19: Edit/Create Hierarchies Tab
Invoked by “Next” Button (Tab 1.2.1), this Tab shown in Figure 20, allows user to add
criteria to a Hierarchy. Criteria could be structured in three levels (Level 2, Level 3, and
Level 4). Level 2 can have maximum of ten Primary criteria. Level 3 can have maximum
of seven Secondary criteria under each Primary Criteria and Level 4 can have maximum
of 4 Tertiary criteria under each Secondary Criteria. In all maximum of 360 criteria could
be defined. If the number of criteria exceeds the allowed limit the user will be prompted
by a message box. The List Box contains the defined criteria in a Hierarchical manner.
User can use the following controls on this Tab:
70
Figure 20: Define Hierarchy Tab
Invoked by “Next” Button (Tab 1.2.2), this Tab shown in Figure 21, allows user to
perform Pair Wise Comparisons to calculate the relative weights of the criteria. The Tab
displays the Reciprocal Positive Pair Wise Comparison Matrix of only one cluster at a
time. The rows and columns of the matrix are assigned to the criteria of the cluster.
Comparisons can be performed by clicking the element on the intersection of the row and
71
column of the desired criteria in the upper diagonal of the matrix. The horizontal scroll
bar in the frame above the matrix can be used to enter the semantic comparison of the
two criteria based on the fundamental scale of AHP. The weights of criteria (in
percentage) and the consistency ratio are calculated with each change in the comparisons.
If the consistency ration exceeds the allowable limit of 0.10 its value is highlighted and
the “Fix” button is enabled. Clicking the “Next Matrix” and “Previous Matrix” button
displays the next and previous matrices respectively. The user will be asked to save or
discard the changes made to the current matrix before proceeding to another one.
72
4) Exit: Returns to Main Window (Window 1)
5) Back: Invokes previous Tab (Define Hierarchy)
6) Next: Invokes next Tab (Results)
7) Horizontal Scroll Bar: To input comparisons in the matrix
Invoked by “Fix” Button (Tab 1.2.3), this Window shown in Figure 22 helps user to
identify most inconsistent judgment of the current comparison matrix, identify the range
of possible input values and assist in selecting an appropriate value of the input. A series
of text boxes display the projected value of consistency ratio, in case the current value of
the comparison is replaced with the seventeen possible values. The user can use the
following buttons on this Window:
73
3) Update: Recalculate the most inconsistent comparisons
4) Close: Close “Adjust Inconsistency” Window
Invoked by “Next” Button (Tab 1.2.3), this Tab shown in Figure 23 displays the Local
and Global Weights of the criteria. The user can use the following buttons on this Tab:
74
This module allows user to associate suppliers with a hierarchy and evaluate them on the
basis of the criteria selected. This window consists of the following Tabs:
Invoked by “Evaluate Suppliers and View Results” Button (window 1), this Tab shown
in Figure 24 allows users to add alternatives (suppliers) to a Hierarchy. In the left hand
side list box titled “Hierarchies” the existing hierarchies are displayed along with their
related notes. The list of existing suppliers in the data base is displayed in the bottom left
list box titled “List of Suppliers”. User can add suppliers by selecting the appropriate
hierarchy and then selecting the suppliers from the list of suppliers and Clicking “Add”
button. The selected suppliers are added to the list box titled “Selected Suppliers”. A
maximum of 10 suppliers can be added to a hierarchy, if more than ten suppliers are
added the user will be prompted “You cannot add more than 10 suppliers”. The selection
must be saved before proceeding. To continue with evaluating the suppliers, the user
must select the hierarchy and click “Next”. The user can use the following buttons on this
Tab:
1) Remove: Removes the selected supplier from the selected supplier list
2) Save: Saves the selection
3) Add: Adds the suppliers selected from the list of suppliers to the selected supplier list
4) Exit: Returns to main window (window 1)
5) Next: Invokes next Tab (Select Criteria to Evaluate)
75
Figure 24: Select/Edit Suppliers Tab
Invoked by “Next” Button (Tab 1.3.1), this Tab shown in Figure 25 allows user to
associate an evaluation method with a criteria and to proceed towards evaluating it. The
list box displays the lowest level criteria of the hierarchy selected in the previous tab
along with their Global Weights. The “Select Evaluation Method” frame in the bottom
right position shows the evaluation methods which are Pair wise Comparisons,
Quantitative and Factor Rating.
1.2.5.2.1. Pair wise Comparisons: In pair wise comparisons the suppliers could be
evaluated by comparing them against each other using the fundamental scale of Analytic
Hierarchy Process. The interface is similar to the one described in Tab 1.2.3.
76
Figure 25: Evaluation Method Selection Tab
In order to proceed with the evaluation, user must select a criterion and select the
appropriate evaluation method and click “Evaluate”. In case a different evaluation
method is associated with the criteria, the user will be prompted “Input using another
evaluation method is associated with this criteria. Are you sure you want to change
evaluation method” selecting ok will change the evaluation method to the selected
77
method. Results of the evaluation can be seen by clicking “Results” button. The user can
use the following buttons on this Tab:
Invoked by “See Results” Button (Tab 1.3.2), this window shown in Figure 26 displays
the results of evaluation of suppliers in a list wise and graphical manner.. In order to
document the results, user can use the “Compile Results” button which will generate an
automated MS Excel file, a specimen of which is shown in Figure 27. User will be asked
to enter the name of the company, date, name of evaluator and the product for which the
evaluation was performed; this information will be added into the Excel file along with
78
the lowest level criteria, its global weight, mode of evaluation and the performance of
each supplier on these criteria.
Figure 27: Example of automated Excel spread sheet for Supplier Ranking
Invoked by “About” Button (Window 1), shown in Figure 28 this window displays basic
information about the application.
79
Figure 28: About Window
The application was packaged using Package and Deployment Wizard available in the
Visual Basic 6.0 Tools.
80
CONCLUSION
The approach developed in this report is more formal and systematic than the current
supplier selection approach. It allows for a wide range of evaluation criteria including
both quantitative and qualitative ones which is a major advantage of this method. Unlike
other methods, which require the priorities of criterion to be provided directly by the
decision maker, the priorities or weights are determined systematically on the basis of
simple pair wise comparisons which reflect human judgment.
After selection of the appropriate criteria and their pair wise comparisons, the most
important primary criteria was found to be Financial Stability (45.67%) followed by
Delivery (15.84%), Supplier Relationship (14.59%), Quality (14.17%), Technical
Capability (5.32%) and Cost (4.38%).
Weights of secondary criteria were also calculated in a similar way. The three suppliers
namely Baosteel (China), TISCO (China) and Hadeed (KSA) were evaluated on basis of
the fifteen secondary criteria, the selection of the evaluation method depended upon the
nature of criterion. The result of evaluation identified Baosteel as a clear leader, it also
showed that cost is not the only or the most important deciding factor while selecting a
supplier. The overall performances of the three suppliers indicated Baosteel as a clear
leader with an overall performance of 47.79% followed by Hadeed and TISCO with
overall performance of 26.56% and 25.66%.
Next sensitivity analysis was performed to identify the effect of criteria weights on the
overall decision. The result of sensitivity analysis also verified the decision; only very
large variation in the weight of Cost criterion would result in a rank reversal. On basis of
the sensitivity analysis it is clear that Baosteel remains to be the best supplier. The
calculations and sensitivity analysis were also verified using Expert Choice™ Software.
Hence it can be concluded on basis of the calculations and the sensitivity analysis
performed that Shanghai Baosteel Group Corporation is the best and most efficient
supplier of Hot Rolled Steel Coils for this particular scenario of supplier selection for
Crescent Steel and Allied Products Limited.
81
The major advantages which our model brings to the supplier selection procedure are:
Apart from developing the mathematical model of supplier selection we also developed a
Visual Basic application, designed specifically for the supplier selection problem using
the Analytic Hierarchy Process approach. The objective was to provide flexibility and
ease in model formulation for future purchasing requirements and to analyze multiple
hierarchies simultaneously for varying needs and requirements.
Analytic Hierarchy Process, although a well known and widely used multi objective and
multi criteria decision making technique around the world, is not widely implemented
and used in Pakistan. This project shows how AHP could be used in effective decision
making and to improve the overall performance of an organization. AHP can be
successfully used for other applications like employee selection, contractor selection, site
selection, to measure the performance of individual or departments etc. and a wide range
of other applications.
82
REFERENCES
[1] http://en.wikipedia.org/wiki/Baosteel
[2] http://en.wikipedia.org/wiki/Hadeed
[3] http://en.wikipedia.org/wiki/Taiyuan
[4] http://www.tisco.com.cn/en/
[5] http://www.crescent.com.pk/home.htm
[6] http://www.crescent.com.pk/spiral_welded_steel_pipes.htm
[7]M.J.S. Bello, A case study approach to the supplier selection process in, Management
Systems Engineering, University of Puerto Rico, Mayagüez, pp. 142, 2004.
[8]http://goliath.ecnext.com/coms2/gi_0199-1562209/A-model-for-strategic-
supplier.html
[9]T.K.S. Babu & S. Kamana, Analytical Hierarchy Process for Vendor Evaluation - A
Case with a Research Institute, South Asian Journal of Management, volume 12,
Issue 1, pages 101-116,2005
[10] http://www.expertchoice.com/
[11] http://www.apian.com/software/decisionpad/
[12] http://www.bestshareware.net/download/rightchoicedss-professional.htm
[13]T.L. Saaty, How to make a decision: The analytic Hierarchy Process. European
Journal of Operational Research 48 (1990) 9-26.
[14]T.L. Saaty, The Analytic Hierarchy and Analytic Network Measurement Processes:
Applications to Decisions under Risk, European Journal of Applied and Pure
Mathematics, Volume 48, Issue 1, pages 122-196, 2008.
[15]S.I. Gass, Decision Making, Models and Algorithms, First edition, John Wiley &
Sons Inc., 1985.
[16]A. Ishizaka & M. Lusti, How to derive priorities in AHP: a comparative study,
Central European Journal of Operations Research, Volume 14, Issue 4, pages 387-
400, 2006.
[17]T.L. Saaty, Decision-making with the AHP: Why is the principal eigenvector
necessary, European Journal of Operational Research, Volume 145, Issue 1, pages
85-91, 2003.
83
[18]H.A. Taha, Operations Research An Introduction, Eighth edition, 2007.
[20]T.L. Saaty, Relative Measurement and Its Generalization in Decision Making; Why
Pairwise Comparisons are Central in Mathematics for the Measurement of
Intangible Factors The Analytic Hierarchy/Network Process, Royal Academy of
Sciences of Madrid, Volume 102, Issue 2, pages 251-318, 2008.
[22]P. Baily & D. Farmer, Purchasing Principles and Management, Fifth edition, Pitman
Publishing, 1986.
[23] http://en.wikipedia.org/wiki/Analytic_Hierarchy_Process
[25]Tahriri, F., et al., AHP approach for supplier evaluation and selection in a steel
manufacturing company, Journal of Industrial Engineering and Management,
Volume 1, Issue 2, 2008.
[26]P. Wright, Begining Visual Basic 6, First edition, Wrox Press Ltd., 2000.
84
APPENDIX
85
Pair-wise Comparisons for the Level 2 Of Hierarchy
Hierarchy
Select the Best
Level 1 Supplier
Technical Supplier
Level 2 Cost Quality Delivery Financial Stability
Capability Relationship
Note
Pair wise Comparisons are performed in the work sheet "Pair wise Comparisons-Level 2"
For each pair-wise comparison, the relative importance of criterion 1 as compared to criterion 2 can be selected
from the drop down menu
The preference of criterion 1 as compared to criterion 2 can be selected from the next drop down menu
The Consistency Ratio represents the level of inconsistency in the judgments. Acceptable level is of 0.1, the closer
its value is to 0 the more consistent will be the judgments
If the value of Consistency Ratio is greater than 0.1 the three most inconsistent judgments are displayed which
could be modified to increase the consistency
The work sheet "Consistency Chart" could be used to identify the value of consisteny ratio if the input to any
judgment is modified
Pair Wise Comparison Form
Level 2 (Primary Criteria)
Performed By
Name: Fahham Hasan Qaiser
Designation: Assistant Manager Buying
Date: 7th of June, 2009
Signature:
Relative Weights
Cost 4.38%
Quality 14.18%
Delivery 15.85%
Financial Stability 45.68%
Technical Capability 5.32%
Supplier Relationship 14.59%
Consistency Ratio 0.088305284
Criterion 1 is less preferred over criterion 2
Most In-
S.No. Criterion 1 Criterion 2 Very Strongly Strongly to Moderately
consistent Extremely Very Srongly Strongly Moderately
to Extremely very Strongly to Strongly
Legend
Inconsistent
Consistent
Current Selection
Criterion 1 Criterion 1 is more preferred over criterion 2
and 2 are
Equally to Equally to Moderately Strongly to Very Strongly
equally Moderately Strongly Very Strongly Extremely
moderately moderately to Strongly very Strongly to Extremely
preferred
0.091253 0.1098972 0.15077201 0.18840743 0.22262012 0.25400476 0.28310178 0.31033977 0.33605224 0.36049962
0.08370144 0.0989735 0.1349571 0.168755 0.19979641 0.22849996 0.25529477 0.28053153 0.30448652 0.32737688
0.15291973 0.2144774 0.30350808 0.37499343 0.43858789 0.49786833 0.55451976 0.60944743 0.66318274 0.71605862
0.08830528 0.0739712 0.07828179 0.08963128 0.10216603 0.11475217 0.12705921 0.13898169 0.15049367 0.16160077
0.09330697 0.1150428 0.15876191 0.19782261 0.23280381 0.26458599 0.29385844 0.32113632 0.34680628 0.37116242
0.09743719 0.0883053 0.10299034 0.12415114 0.14614175 0.16800955 0.18954627 0.21072594 0.23157342 0.25212607
0.07791098 0.0932813 0.12842161 0.16550039 0.20393599 0.24326367 0.28318859 0.32352776 0.36416433 0.40502099
0.08968782 0.0896878 0.07663093 0.07768882 0.08232058 0.08830528 0.09489852 0.10178739 0.10882168 0.11592294
0.09645065 0.0883053 0.10122483 0.11969546 0.13881021 0.15777564 0.17643289 0.19477241 0.21282336 0.23062246
0.06755474 0.0785795 0.11090037 0.14671191 0.18433722 0.22305858 0.26248763 0.30239795 0.3426496 0.38315175
0.13194231 0.0933704 0.07831121 0.07854909 0.08268116 0.08830528 0.09461425 0.10126707 0.10809875 0.11502216
0.0998998 0.0883053 0.09911374 0.11665503 0.13518202 0.15371778 0.17203623 0.19009593 0.20790834 0.22549934
0.31245739 0.2116 0.1431022 0.1145734 0.09855054 0.08830528 0.0813176 0.07640866 0.07294232 0.07053878
0.14164186 0.1045333 0.08734473 0.08503974 0.08830528 0.09470641 0.10313578 0.11297462 0.12383943 0.13547672
0.08193773 0.0980532 0.13585813 0.17060936 0.20190808 0.23037047 0.25656552 0.28094014 0.30383758 0.3255224
Pair Wise Comparison Form
Level 3 (Secondary Criteria)
Performed By
Name: Fahham Hasan Qaiser
Designation: Assistant Manager Buying
Date: 7th of June, 2009
Signature:
COST
S. No. Criterion 1 Relative Importance Criterion 2
1 Unit Price Moderately more important than Payment Terms
Consistency Ratio 0
QUALITY
S. No. Criterion 1 Relative Importance Criterion 2
1 Quality of Deliveries Equally important as ISO 9001 Certification
2 Quality of Deliveries Equally important as Response to Quality Problems
3 ISO 9001 Certification Equally important as Response to Quality Problems
Consistency Ratio 0
DELIVERY
S. No. Criterion 1 Relative Importance Criterion 2
1 On time Delivery Moderately to Strongly more important than Deliver Lead Time
Consistency Ratio 0
FINANCIAL STABILITY
S. No. Criterion 1 Relative Importance Criterion 2
1 Total Revenues Moderately to Strongly more important than Banking History
Consistency Ratio 0
TECHNICAL CAPABILITY
S. No. Criterion 1 Relative Importance Criterion 2
1 Annual Capacity Moderately less important than Technical Support
Consistency Ratio 0
SUPPLIER RELATIONSHIP
S. No. Criterion 1 Relative Importance Criterion 2
1 Reciprocal Arrangement Moderately to Strongly more important than Primary or Secondary
2 Reciprocal Arrangement Equally important as Desire for Business
3 Reciprocal Arrangement Moderately less important than Supplier Response
4 Primary or Secondary Very Strongly less important than Desire for Business
5 Primary or Secondary Strongly to Very Strongly less important than Supplier Response
6 Desire for Business Moderately to Strongly less important than Supplier Response
Consistency Ratio 0.098546061
Results of Pair Wise Comparisons
Level 3
Relative Priorities
COST
Unit Price 75.00%
Payment Terms 25.00%
Consistency Ratio 0
QUALITY
Quality of Deliveries 33.33%
ISO 9001 Certification 33.33%
Response to Quality Problems 33.33%
Consistency Ratio 0
DELIVERY
On time Delivery 80.00%
Deliver Lead Time 20.00%
Consistency Ratio 0
FINANCIAL STABILITY
Total Revenues 80.00%
Banking History 20.00%
Consistency Ratio 0
TECHNICAL CAPABILITY
Annual Capacity 25.00%
Technical Support 75.00%
Consistency Ratio 0
SUPPLIER RELATIONSHIP
Reciprocal Arrangement 19.15%
Primary or Secondary 5.54%
Desire for Business 22.13%
Supplier Response 53.18%
Consistency Ratio 0.098546061
Supplier Evaluation Form
Evaluated Suppliers:
Supplier 1: Baosteel
Supplier 2: TISCO
Supplier 3: Hadeed
How much a Supplier with ISO 9001 Certification is preferred over a Supplier without ISO 9001
Very Strongly to Extremely
On the basis of Response to Quality Problems:
Baosteel is Moderately more preferred than TISCO
Baosteel is Equally preferred as Hadeed
TISCO is Equally to Moderately less preferred than Hadeed
Consistency Ratio 0.01563252
What Delivery Lead Time difference between two suppliers will make:
Both suppliers equally preferred: 5 Days
One supplier strongly preferred: 15 Days
One supplier extremely preferred: 25 Days
Total Revenues:
Baosteel 21.50 Billion USD
TISCO 14.77 Billion USD
Hadeed 6.80 Billion USD
Annual Capacity:
Baosteel 20.00 Million Tons/Year
TISCO 10.00 Million Tons/Year
Hadeed 5.50 Million Tons/Year
Primary or Secondary:
Baosteel Primary
TISCO Primary
Hadeed Primary
1. Cost Criterion:
2. Quality Criterion
3. Delivery Criterion
4. Financial Stability Criterion