Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

Project Report

Smart decision-making Tool for Supply chain


"Supplier selection New methods in the Era of
4.0"

By:

▪ NAJI Doha
▪ EZ-ZERRIFI AMRANI Malak

Supervised by:

▪ Academic supervisor: Ms. DADDA Afaf


Ms. EZZINE Latifa
▪ Industrial Supervisor: Ms. LEBHAR Ikram

Academic year 2023-2024


Acknowledgments
We would like to express our sincere gratitude to our esteemed academic supervisor, Ms. Afaf
DADDA, for her unwavering guidance, invaluable insights, and continuous support throughout
the course of this project. Her expertise and encouragement have been instrumental in shaping
the trajectory of our work.

A special thanks to Ms. EZZINE Latifa, for her administrative support and facilitation, which
have contributed to the smooth progression of our project.

We would also like to extend our appreciation to our dedicated industrial supervisor, Ms.
LEBHAR Ikram, whose practical wisdom, industry knowledge, and mentorship have enriched
our project experience. Her commitment to excellence has been a driving force behind our
endeavors.

This acknowledgment is a testament to the collaborative spirit fostered by these remarkable


individuals, whose contributions have been pivotal to the success of this endeavor.

2
List of figures
Figure 1: Rating scale. ................................................................................................................................. 14
Figure 2: The proposed methodology .................................................................................................. 19

3
List of tables
Table 1:Random Index, with n the number of alternatives………………………………………….14
Table 2: QQOQCP method. ........................................................................................................................ 15
Table 3: Decision matrix ............................................................................................................................ 20
Table 4: Normalized matrix ...................................................................................................................... 20
Table 5: Weights for every criterion ..................................................................................................... 20
Table 6: Weighted normalized decision matrix ................................................................................ 21
Table 7: Value of PIS and NIS ................................................................................................................... 21
Table 8: The separation measures from the PIS and the NIS ...................................................... 21
Table 9: The relative closeness to the ideal solution ...................................................................... 21
Table 10: Final ranking using TOPSIS .................................................................................................. 22
Table 11: Decision matrix ......................................................................................................................... 22
Table 12: Values of f* and f-..................................................................................................................... 22
Table 13: Values of Rj and Sj .................................................................................................................... 22
Table 14: Values of Qj ................................................................................................................................. 23
Table 15: Final ranking using VIKOR .................................................................................................... 23

4
List of abbreviations
AHP: Analytic Hierarchy Process
AI: Artificial Intelligence
BWM: Best Worst Method
COPRAS: COmplex PRoportional Assessment
DM: Decision Making
DSC: Digital Supply Chain
IoT: Internet of Things
MAH: Maximize Agreement Heuristic
MCDM: Multi Criteria Decision Making
MULTIMOORA: Multi Objective Optimization based on Ratio Analysis plus the Full
Multiplicative Form
NIS: Negative Ideal Solution
PF: Proximity Factor
PIS: Positive Ideal Solution
SC: Supply Chain
TFN: Triangular Fuzzy Numbers
TOPSIS: Technique for Order Preference by Similarity to Ideal Solution
VIKOR: VlseKriterijumska Optimizacija I Kompromisno Resenje

5
Table of contents
Introduction....................................................................................................................................................... 7

Chapter one: Literature Review and project context ........................................................................ 9

I. Introduction ...................................................................................................................................... 10

II. Bibliographic research................................................................................................................... 10

1. Supply chain ........................................................................................................................... 10

2. Digital supply chain ........................................................................................................... 10

3. MCDM method ....................................................................................................................... 10

4. BWM method .......................................................................................................................... 10

5. MULTIMOORA method...................................................................................................... 10

7. MAH method ........................................................................................................................... 11

8. TOPSIS method ..................................................................................................................... 11

III. Project framing ............................................................................................................................... 15

IV. Conclusion ......................................................................................................................................... 16

Chapter two: Article analysis and real case study ........................................................................... 17

I. Introduction ...................................................................................................................................... 18

II. Article analysis.................................................................................................................................. 18

III. Real case study ................................................................................................................................. 20

1. TOPSIS Method ........................................................................................................................ 20

2. VIKOR Method: ........................................................................................................................ 22

IV. Conclusion .......................................................................................................................................... 23

General Conclusion ...................................................................................................................................... 24

Bibliography ................................................................................................................................................... 25

6
Introduction
A supply chain is a network of organizations, individuals, activities, resources, and
information involved in the creation and distribution of goods or services, from the source
of raw materials to the end customer. The supplier provides raw materials to the
manufacturer, the manufacturer ships finished products to the distributors, the
distributors pass on selected amounts of the products to retailers who ultimately sell the
products to consumers.

Whilst every stage is considered important and crucial in the process of production,
procurement is considered as a key element. This stage involves sourcing and purchasing
raw materials or components from suppliers. Evidently, manufacturers are required to
select a supplier. This choice has repercussions across the entire value network,
influencing product quality, operational efficiency, ultimately, the success of an
organization in the competitive market and most importantly client’s satisfaction.

Organizations commonly utilize MCDM techniques to select suppliers. These


methodologies provide a systematic and quantitative approach to evaluate and rank
suppliers based on multiple criteria, aligning decision-making with organizational goals
and priorities. It takes into consideration multiple criteria simultaneously, allowing
decision makers to weigh and balance various factors such as cost, quality, reliability, and
sustainability.

Nowadays, in the dynamic landscape of Industry 4.0, the traditional contours of supply
chain management are undergoing a profound metamorphosis. Organizations are striving
to adapt to the demands of this era, it has become imperative to revisit and redefine the
very essence of the supply chain that fuels industrial ecosystems. With all the last
repercussions, the re-evaluation of methods for selecting suppliers has become essential.

Amidst all these changes, a novel and forward-looking approach in supplier selection
surfaced by employing a combination of MCDM techniques and machine learning. In fact,
combining the analytical rigor of MCDM with the computational capabilities of AI
enhances the precision and efficiency of their decision-making processes. MCDM offers a
structured framework for evaluating suppliers based on multiple criteria, whilst AI
techniques, such as machine learning algorithms, contribute predictive analytics and
data-driven insights. This synergy enables organizations to not only assess supplier

7
performance across diverse parameters but also to anticipate future trends and potential
risks.

8
Chapter one: Literature Review and project context

9
I. Introduction
In this section, we concentrate on two vital components of our project. Initially, we
undertake extensive research, exploring key domains like machine learning, deep
learning, neural networks, IoT, and various Multi-Criteria Decision-Making (MCDM)
methods. This research phase is crucial for establishing a comprehensive theoretical
foundation essential for our project's evolution. Additionally, we delve into framing our
project using systematic methods, ensuring a clear roadmap for its development.

II. Bibliographic research


1. Supply chain
A supply chain covers all phases of commercial goods production, from gathering raw
materials to the final delivery of finished products to consumers. It involves a chain of
linked businesses and activities that move the product from supplier to end customer.

2. Digital supply chain


A digital supply chain refers to the use of digital technologies, data, and advanced analytics
to enhance and optimize the various processes and components of a traditional supply
chain. It involves the integration of digital technologies throughout the entire supply chain
to improve efficiency, visibility, and responsiveness.

3. MCDM method
Multi Criteria Decision Making refers to a systematic and structured approach used to
resolve complex problems involving conflicting criteria. It offers decision-makers a formal
methodology to rank alternatives and make informed choices, even under highly intricate
conditions.

4. BWM method
Best Worst Method is a valuable Multi-Criteria Decision-Making (MCDM) technique that
facilitates the assignment of importance weights to evaluation criteria by making pairwise
comparisons between the best alternative and all others, as well as between the worst
alternative and all others.

5. MULTIMOORA method
MULTIMOORA is a method used to solve a wide range of management-related
optimization problems characterized by the presence of conflicting objectives. It consists

10
of three main phases with MOORA being one of them. It includes three main phases: a
ratio system approach MOORA, a reference point approach, and a full multiplicative form
approach.

6. COPRAS method
COPRAS Method is a valuable Multi-Criteria Decision-Making (MCDM) technique that
facilitates the assignment of importance weights to evaluation criteria by making pairwise
comparisons between the best alternative and all others, as well as between the worst
alternative and all others.

7. MAH method
MAH is a method used to aggregate in a final consensus ranking. Its techniques are used
to evaluate and make decisions when faced with multiple conflicting criteria.

8. TOPSIS method
TOPSIS is a method used for ranking alternatives and provides a convenient approach to
untangle MCDM. It is built on fundamental concepts of positive ideal solutions (PIS) and
negative (NIS) and it requires the best alternative to be the one situated at both the
smallest distance from the PIS and the farthest distance from the NIS. [2]

Here are the steps involved in the TOPSIS method:

a) Construct the decision matrix


First, we establish a measurement scale for the criterion, for example, a scale ranging from
1 to 5.

1: not interesting and 5: perfectly interesting.

b) Calculate normalized matrix


This step transforms various attribute dimensions into non-dimensional attributes which
allows comparisons across criteria. Since various criteria are usually measured in various
units, the scores in the evaluation matrix should be transformed to a normalized scale
using Euclidean distance:

𝑥𝑖𝑗
𝑟𝑖𝑗 =
√∑𝑛𝑖=1 𝑥𝑖𝑗 2

11
c) Calculate the weighted normalized decision matrix
We assign weights to each criterion 𝑤𝑗 and then we calculate the weighted normalized
values in the following way:

𝑣𝑖𝑗 = 𝑤𝑗 × 𝑟𝑖𝑗

N.B: ∑ 𝑤𝑗 = 1 !!

d) Determine the PIS and NIS


The PIS is the solution that maximizes the benefit criteria and minimizes the cost criteria
whereas the NIS maximizes the cost criteria and minimizes the benefit criteria.

e) Calculate the separation measures from the PIS and the NIS
The separation of each alternative from the PIS is given as:

𝐸𝑖+ = √∑(𝑟𝑗+ − 𝑟𝑖𝑗 )2


𝑗=1

The separation of each alternative from the NIS is given as:

𝐸𝑖− = √∑(𝑟𝑗− − 𝑟𝑖𝑗 )2


𝑗=1

If the alternatives representing the farthest distance from the NIS and closest to the PIS
were identical, we would have concluded our calculation at this stage, but since that is not
the case, we should add an additional step.

f) Calculate the relative closeness to the positive ideal solution


PF: The measures the rate between 0 and 1 of proximity from the NIS to the PIS.


𝐸−
𝑆 = −
𝐸 + 𝐸+

9. VIKOR method
The VIKOR is a multi-criteria decision-making (MCDM) technique used to select the best
alternative from a set of options. It is particularly useful when there are conflicting criteria
and a need to find a compromise solution. [3]

12
Here are the steps involved in the VIKOR method:

Let M ([𝑥𝑖𝑗 ]) be a Decision matrix, I the number of criteria and j the number of alternatives.

a) Determine the values of the criteria ‘i’ 𝒙∗𝒊 and 𝒙−


𝒊

𝑥𝑖∗ = max(𝑥𝑖𝑗 ) 𝑥𝑖− = min(𝑥𝑖𝑗 )

b) Calculate the values 𝑹𝒋 et 𝑺𝒋


𝑛
𝑥𝑖∗ − 𝑥𝑖𝑗 𝑥𝑖∗ − 𝑥𝑖𝑗
𝑆𝑗 = ∑ 𝑤𝑖 ∗ 𝑒𝑡 𝑅𝑗 = max (𝑤𝑖 ∗ )
𝑥𝑖 − 𝑥𝑖− 𝑥𝑖 − 𝑥𝑖−
𝑖=1

c) Calculate the values 𝑺∗ , 𝑺− et 𝑹∗ , 𝑹−


𝑆 ∗ = min 𝑆𝑗 , 𝑆 − = max 𝑆𝑗 , 𝑅 ∗ = min 𝑅𝑗 , 𝑅 − = max 𝑅𝑗

d) Calculate the values 𝑸𝒋 with =0.25


𝑆𝑗 − 𝑆 ∗ 𝑅𝑗 − 𝑅 ∗
𝑄𝑗 =  + (1 − )
𝑆− − 𝑆∗ 𝑅− − 𝑅∗

e) Propose a solution A for which Q is minimum and verifies the two conditions 𝑪𝟏 and 𝑪𝟐
• 𝐶1 : Acceptability condition
𝑄(𝐵) − 𝑄(𝐴) ≥ 𝐷𝑄
1
where B the second best alternative and 𝐷𝑄 = | 1−𝑗 | and j being the number

of alternatives.
• 𝐶2 : Stability condition
A should be the best solution based on the S and R rankings

10. AHP Method


The Analytic Hierarchy Process (AHP) is a structured technique for decision-making that
involves breaking down a complex problem into a hierarchy of criteria and alternatives,
and then systematically evaluating and comparing them. AHP is widely used in various
fields for decision analysis and is especially effective when dealing with complex, multi-
criteria decision problems. [4]

Here are the key steps in the AHP method:

13
a) Define the Decision Hierarchy:
Develop a hierarchical structure with a goal at the top level, the attributes/criteria at the
second level, and the alternatives at the third level.

b) Pairwise Comparison matrix:


Determine the relative importance of different attributes or criteria for the goal.

-> Pair-wise comparison matrix.

Pair-wise comparison matrix is created with the help of a scale of relative importance.

Figure 1: Rating scale.

c) Normalize the matrix and calculate the relative weights:


-Normalize the matrix by dividing each column by the sum of the values in this column.

𝑎𝑖1 𝑎𝑖2
; ;… with aij is the value of each cell.
∑𝑎𝑖1 ∑𝑎𝑖2

-Calculate the criteria weight by averaging all the elements in the row.

𝑎1𝑗 𝑎2𝑗
; ;… with j is the number of alternatives.
𝑗 𝑗

d) Calculate the consistency:


-We use the non-normalized pair-wise comparison matrix and we multiply each cell with
the criteria weight associated.

-Then we calculate the weighted sum value by adding all the values in each row.

WSV1=∑a1j…

-Next, we calculate the ratio by dividing the weighted sum value by the criteria weight.

14
𝑊𝑆𝑉1
𝜆1𝑗 = …
𝑊1𝑗

-The average of the values calculated is called λmax

∑𝜆1𝑗
𝜆𝑚𝑎𝑥 =
𝑗

e) Calculate the consistency index (CI)


𝜆𝑚𝑎𝑥 − 𝑗
𝐶𝐼 =
𝑗−1

f) Calculate the Consistency Ratio. (CR)


𝐶𝐼
𝐶𝑅 = ; with RI: Random Index
𝑅𝐼

We should make sure that CR<0.10 in order to verify the consistency of the matrix.

Table 1: Random index with n number of alternatives.

n 1 2 3 4 5 6 7 8 9 10
RI 0.00 0.00 0.58 0.90 1.12 1.24 1.32 1.41 1.45 1.49

g) Make a Decision:
Make a decision based on the aggregated weights, and select the alternative that has the
highest overall score.

III. Project framing


1. QQOQCP method
Table 2: QQOQCP method.

Who Decision-makers

What Supplier selection

How MCDM methods

Why Selecting the best supplier

2. Long-term Objective
15
The long-term goal of this project is the integration of 4.0 algorithms in MCDM methods.
This helps in picking the best suppliers, lowering risks, encouraging new ideas, and
ultimately making the supply chain stronger and more successful over time.

IV. Conclusion
In summary, this chapter lays the groundwork for our project. The extensive bibliographic
research deepened our understanding of key concepts, and the QQOCQP method
systematically framed our project. We explored mathematical models and steps of
TOPSIS, VIKOR, and AHP methods. The long-term goal is a strategic and adaptive
approach to optimize supplier selections, mitigate risks, and foster innovation in the
organization's supply chain.

16
Chapter two: Article analysis and real case study

17
I. Introduction
In this chapter, we will be analyzing an article by the name: « An integrated and
comprehensive fuzzy multicriteria model for supplier selection in digital supply chains »,
published in 2021. In addition, we will be evaluating six suppliers using two MCDM
methods TOPSIS and VIKOR. The chosen supplier will be providing us with a management
platform.

II. Article analysis


The article processes a MCDM model for supplier selection. The proposed model defines
a practical and efficient solution by combining well-founded and widely used MCDM
techniques, including BWM, MULTIMOORA, COPRAS, and TOPSIS. These techniques are
integrated into a comprehensive model that allows for direct comparisons between
approaches and yields robust results.

It ranks the alternatives based on 12 criteria well chosen by a specialized team, the
criteria are as follows: Real-Time visibility, adopting advanced analytics, technical capability,
continuous collaboration, alignment of the supplier, agility, and flexibility, lack of tools and
technologies, lack of planning, lack of information sharing, lack of knowledge, lack of digital
collaboration, lack of technology integration.

The article introduces TFNs as a method to incorporate uncertainty in the decision-


makers' subjective evaluations. This allows the model to handle ambiguity and
imprecision associated with DMs' judgments. The integrated approach, which combines
multiple MCDM techniques, considers this uncertainty and provides a more robust
ranking procedure. Finally, the MAH is applied to enhance the consensus ranking,
ensuring that uncertainty and variations in DMs' evaluations are addressed.

The proposed methodology is presented as follows:

➢ Phase 1: Several selection criteria are reviewed, and the key ones for supplier
selection within a DSC environment are identified.
➢ Phase 2: Experts’ opinions are collected, and the importance weights of the
criteria are calculated by fuzzy BWM.
➢ Phase 3,4,5: Third, fuzzy MULTIMOORA, fuzzy COPRAS, and fuzzy TOPSIS are
used in three distinct and parallel phases (Phase 3, Phase 4, and Phase 5,
respectively) to rank the suppliers.

18
➢ Phase 6: The rankings obtained are integrated with MAH to achieve a consensus
ranking. [1]

Figure 2: The proposed methodology

19
III. Real case study
1. TOPSIS Method
1.1. Construct the decision matrix
On a scale of 1 to 5 we assign a grade to every supplier.

Table 3: Decision matrix

C1 C2 C3 C4 C5 C6
O1 3 4 3 3 1 4
K1 2 0 1 2 1 2
S1 2 2 2 2 1 2
Z1 5 5 1 4 4 5
J1 3 4 3 3 1 4
SP1 4 5 5 5 3 5

1.2. Calculating Normalized Matrix


Table 4: Normalized matrix

Rij C1 C2 C3 C4 C5 C6
O1 0,366508 0,431331 0,428571 0,366508 0,185695 0,421637
K1 0,244339 0 0,142857 0,244339 0,185695 0,210819
S1 0,244339 0,215666 0,285714 0,244339 0,185695 0,210819
Z1 0,610847 0,539164 0,142857 0,488678 0,742781 0,527046
J1 0,366508 0,431331 0,428571 0,366508 0,185695 0,421637
SP1 0,488678 0,539164 0,714286 0,610847 0,557086 0,527046

1.3. Calculate the weighted normalized decision matrix


We assign the following weights to each criterion.
Table 5: Weights for every criterion

C1 C2 C3 C4 C5 C6
Weight 0.3 0.15 0.15 0.2 0.1 0.1

We indeed can verify ∑ 𝑤𝑗 = ∑ 0.3 + 0.15 + 0.15 + 0.2 + 0.1 + 0.1 = 1.

20
Table 6: Weighted normalized decision matrix

Vij C1 C2 C3 C4 C5 C6
O1 0,109952 0,0647 0,064286 0,073302 0,01857 0,042164
K1 0,073302 0 0,021429 0,048868 0,01857 0,021082
S1 0,073302 0,03235 0,042857 0,048868 0,01857 0,021082
Z1 0,183254 0,080875 0,021429 0,097736 0,074278 0,052705
J1 0,109952 0,0647 0,064286 0,073302 0,01857 0,042164
SP1 0,146603 0,080875 0,107143 0,122169 0,055709 0,052705

1.4. Determine the PIS and NIS


Table 7: Value of PIS and NIS

A+ 0,183254 0,080875 0,107143 0,048868 0,074278 0,052705


A- 0,073302 0 0,021429 0,122169 0,021429 0,021082

1.5. Calculate the separation measures from the PIS and the NIS
Table 8: The separation measures from the PIS and the NIS

Vij E+ E-
O1 0,1062218 0,1010284
K1 0,1734377 0,0733574
S1 0,1505998 0,0830
Z1 0,0986661 0,1517
J1 0,1062218 0,1010
SP1 0,0840312 0,1464

The closest alternative to the PIS is SP1 and the farthest alternative from the NIS is Z1.

1.6. Calculate the relative closeness to the ideal solution


Table 9: The relative closeness to the ideal solution

Vij PF
O1 0,4874708

K1 0,2972401

S1 0,3552753

Z1 0,6059514

J1 0,4874708

SP1 0,6353464

21
The closest alternative to the ideal solution would be once again SP1, which confirms our
previous results.

1.7. Final ranking


Table 10: Final ranking using TOPSIS

Vij Ranking
O1 3
K1 5
S1 4
Z1 2
J1 3
SP1 1

2. VIKOR Method:
2.1. Construct the decision matrix
Table 11: Decision matrix

Alternatives C1 C2 C3 C4 C5 C6
O1 3 4 3 3 1 4
K1 2 0 1 2 1 2
S1 2 2 2 2 1 2
Z1 5 5 1 4 4 5
J1 3 4 3 3 1 4
SP1 4 5 5 5 3 5

2.2. Determine the values of the criteria ‘i’ 𝒇∗𝒊 and 𝒇−


𝒊
Table 12: Values of f* and f-

f* 5 5 5 5 4 5
f- 2 0 1 2 1 2

2.3. Calculate the values 𝑹𝒋 et 𝑺𝒋


Table 13: Values of Rj and Sj

Alternatives Sj Rj
O1 0.605 0.200
K1 1.100 0.300
S1 1.003 0.300
Z1 0.283 0.150
J1 0.605 0.200
SP1 0.133 0.100

22
2.4. Calculate the values 𝑆 ∗ , 𝑆 − et 𝑅∗ , 𝑅−
𝑆 ∗ = 0.133, 𝑆 − = 1.100, 𝑅 ∗ = 0.100, 𝑅− = 0.300

2.5. Calculate the values 𝑄𝑗 with  =0.25


Table 14: Values of Qj

Alternatives Qj
O1 0.4970
K1 1.0000
S1 0.9748
Z1 0.2263
J1 0.4970
SP1 0.000

SP1 is the alternative with the lowest value of Qj, hence it is the optimum choice amongst
the available alternatives.

2.6. Final ranking


Table 15: Final ranking using VIKOR

Alternatives Ranking
O1 3
K1 5
S1 4
Z1 2
J1 3
SP1 1
Verifying acceptability conditions:

1
𝑄(𝐵) − 𝑄(𝐴) = 0.226 > | | = 0.2
1−6

SP1 is the best solution based on the S and R rankings.

IV. Conclusion
In this chapter, we conducted an analysis of an article centered around a novel proposed
model that integrates various MCDM methods. Additionally, evaluated six alternatives
using two methods TOPSIS and VIKOR based on six criteria. We have obtained the same
raking using both techniques, with SP1 being ranked first, Z1 being second, O1 and J1
being third, S1 being fourth and lastly K1 being ranked fifth.

23
General Conclusion
In conclusion, the initial phase of this project has dedicated itself to a comprehensive
exploration of MCDM techniques and their application in the intricate domain of supplier
selection. Our exhaustive analysis not only delved into the theoretical underpinnings but
also culminated in the development of a robust methodology for supplier selection
employing techniques such as TOPSIS and VIKOR. This methodological foundation not
only enhances the precision of decision-making processes but also establishes a solid
framework for evaluating suppliers across multiple criteria. Furthermore, our findings
will be opening new avenues for future exploration, particularly in the integration of AI
algorithms into the decision-making process. The integration of AI holds the promise of
elevating the sophistication of supplier selection, presenting an exciting prospect for
further research and implementation in the subsequent phases of this project.

24
Bibliography
[1] Madjid Tavana, Akram Shaabani, Debora Di Caprii, and Maghsoud Amiri. (2021) An
integrated and comprehensive fuzzy multicriteria model for supplier selection in digital
supply chains, Sustainable Operations and Computers 2,149–169.

[2] Abdel YEZZA. (2017) La methode TOPSIS explique pas a pas, une variante propose.
Ph.D. thesis. Available on: Step by Step Topsis.pdf

[3] K. Selvakumari, and M. Ajitha Priyadharshini. (2017) Vikor Method for Decision
Making Problem Using Octagonal Neutrosophic Soft Matrix, International Journal of Latest
Engineering Research and Applications, 41-45.

https://fs.unm.edu/neut/VikorMethodForDecision.pdf

[4] Ahmed Dhouibi, Wahiba Bali Kalboussi, and Makrem Ben Jeddou. (2015) Application
de la methode AHP pour le choix multicritere des fournisseurs, Revue Marocaine de
recherche en management et marketing, 60-71.

https://revues.imist.ma/index.php/REMAREM/article/download/3772/4433.pdf

[5] Travail de recherche,


https://d1n7iqsz6ob2ad.cloudfront.net/document/pdf/532b44824d5ef.pdf

25

You might also like