Professional Documents
Culture Documents
SCHOOL OF COMMERCE
GRADUATE PROGRAM
Addis Ababa
May, 2017
Modelling and IT Application in SCM
Table of Contents
1. INTRODUCTION......................................................................................................................2
4. METHODOLOGY........................................................................................................................4
5.1 DEVELOP A PAIRWISE COMPARISON MATRIX FOR EACH LOCATION AGAINST EACH CRITERION.............................5
5.2 SYNTHESIZATION...............................................................................................................................7
5.2.1 SUM UP THE VALUES IN EACH COLUMN OF THE PAIRWISE COMPARISON.......................................................7
5.2.2 DIVIDE EACH VALUE IN EACH COLUMN OF THE PAIRWISE COMPARISON MATRICES BY THE CORRESPONDING
COLUMN SUM AND GET THE BELOW NORMALIZED MATRICES.................................................................................8
5.2.3 AVERAGE THE VALUES, CONVERTED IN DECIMALS, IN EACH ROW OF THE NORMALIZED MATRICES TO HAVE THE SO
CALLED PREFERENCE VECTORS.........................................................................................................................9
5.2.4 COMBINE THE VECTORS OF PREFERENCES FOR EACH CRITERION, FROM ABOVE, INTO ONE PREFERENCE MATRIX
THAT SHOWS THE PREFERENCE FOR EACH LOCATION FOR EACH CRITERION..............................................................10
2|Page
Modelling and IT Application in SCM
1. Introduction
Zebidar’s Brewery S.C. plant is located some 168 km away from Addis Ababa in Gubreye
town near Wolkite. The brewery company was named after the mountain “Zebidar” that is
found in Gurage Zone, the same zone where the factory is established.
Zebidar is established by a joint venture between a Belgian Beer Company called Unibra,
and a local business owner called Jemar. The company launched its first products in January
2017 and hence is an infant to the beer market in Ethiopia. It is now working hard to claim
reasonable amount of share in the Ethiopian market.
Zebidar aimed to produce 350,000 hectoliter (hl) of beer per year and deliver it to
primarily targeted markets in Addis Ababa and the Southern part of Ethiopia. As part of
expansion plan and need to be more accessible, Zebidar found it difficult to deliver its
products only from the plant. Currently, the company is trying to address its clients’ needs
by supplying its products from the plant as well as from a distribution center (depot) that is
found in Addis Ababa. For that matter, the company rented a warehouse in Addis Ababa and
is using it as a distribution center.
This year, the foreign business partner, which is Unibra, has sold its major shares to BGI
Ethiopia and hence the company takes a new status in terms of new administration, board
of directors and even its business structure. The new management is planning to boost the
factory’s production from 350,000 hl per year to 650,000 hl per year and also planning a
strategy to reach out markets all over Ethiopia which has not been the case before.
Accordingly, the new management need to assess the significance of the Addis Ababa
distribution Center (Depot) and want to relocate its distribution center to other parts of
Ethiopia if that is found to be worth considering. The Distribution Manager is in charge of
assessing the case and suggest another potential site to relocate the depot. We contacted
this person in charge through one of our group member, who works for the same company
3|Page
Modelling and IT Application in SCM
under discussion, and assist on making the relocation decision through scientific model
analysis. When we shared the idea of scientific decision-making support tools with the
distribution manager, he happily accepted it and asked us to assess the situation and report
the results back to him so that he will recommend the best option to the top management.
The appropriate model we chose and applied to help out on this decision is the Multi
criteria decision making tool specifically the Analytical Hierarchy Process (AHP) model.
The objective of the study therefore is to apply the AHP model and advise which location is
best to be used as a depot and also rank the rest of the locations.
4|Page
Modelling and IT Application in SCM
Consequently, managers at any level are expected to make rational decisions to ensure that
organization are sustainable and function as intended. To do so, managers need to have
relatively comprehensive information about all the circumstances and the environment that
have impact on the decision. The information gathered can either be qualitative or
quantitative, that can be expressed in figures. Analyzing the information very well will help
managers to make rational and optimum decisions and choose the most viable option out
of the possible alternatives. Most of the time, decisions need to be made in view of different
conditions that have their own constraints which the manager should choose the best one
in favor of the organization. However, it is difficult to simply look at the decision criteria
and make a decision without applying a scientific model. Scientific models are designed to
support managers by analyzing the criteria for the decision and qualitatively recommend
the best option.
One of the Multi Criteria Decision Making models, The Analytical Hierarchy Process (AHP),
developed by Thomas Saaty, is a method for ranking decision alternatives and selecting the
best one when the decision maker has multiple objectives, or criteria, on which to base the
decision. By developing a numeric score to rank each decision alternative, the model
answers the question which one is best, as most of the time the decision maker has multiple
criteria to make a single decision.
4. Methodology
We paid a visit to the factory’s head office located around Kality and had a face to face
discussion and interview with the distribution manager of Zebidar Brewery Factory S.C.
5|Page
Modelling and IT Application in SCM
The manager listed out the alternative locations that the company wants to relocate the
Addis Ababa depot and the criteria based on which the factory wants to choose from those
locations. Accordingly, we have developed a pairwise matrix by which the manager
compared the alternatives and gave scores based on the preference scale for pairwise
comparisons. It was these data that we have used for computations throughout this paper.
All the Analytical Hierarchy Process (AHP) steps were suitably followed to reach out to the
results and made recommendations in the end to the distribution manager.
6|Page
Modelling and IT Application in SCM
5.1 Develop a pairwise comparison matrix for each location against each
criterion
Developing a pairwise comparison matrix using a preference scale is the first step in the
Analytical Hierarchy Process. The below matrix shows the summarized comparison made
for the locations against each criterion for location selection.
7|Page
Modelling and IT Application in SCM
5.2 Synthesization
The second mathematical step to arrive at the AHP recommended decision is called
Synthesization. Synthesizing the values has some simple steps which we will separately
discuss and apply below.
5.2.2 Divide each value in each column of the pairwise comparison matrices by
the corresponding column sum and get normalized matrices
Table 11. Normalized Matrix for Pairwise comparison for market share (Criterion 2)
Location Addis Ababa Hawassa Jima mekele Bahir Dar
Addis Ababa 60/117 112/165 48/77 9/58 5/15
Hawassa 60/468 28/165 16/77 21/58 4/15
Mekele 60/702 28/330 8/77 24/58 2/15
Jima 60/351 28/1155 8/616 3/58 3/15
Bahir Dar 60/585 28/660 8/154 3/174 1/15
Sum 1 1 1 1 1
9|Page
Modelling and IT Application in SCM
Table 12. Normalized Matrix for Pairwise comparison for Income level (Criterion 3)
Table 13. Normalized Matrix Pairwise comparison for Logistics cost (Criterion 4)
Table 14. Normalized Matrix with row average for market share (Criterion 2)
Location Addis Ababa Hawassa Jima mekele Bahir Dar Row
average
Addis Ababa 60/117 112/165 48/77 9/58 5/15
Hawassa 60/468 28/165 16/77 21/58 4/15
Mekele 60/702 28/330 8/77 24/58 2/15
Jima 60/351 28/1155 8/616 3/58 3/15
Bahir Dar 60/585 28/660 8/154 3/174 1/15
10 | P a g e
Modelling and IT Application in SCM
Sum 1 1 1 1 1
Table 15. Normalized Matrix with row average for Transportation cost (Criterion 3)
Location Addis Ababa Hawassa Bahir Dar Dire Dawa Nekemtie Dessie Averages
Addis Ababa 0.5398 0.7282 0.3684 0.3784 0.4348 0.4045 0.4757
Hawassa 0.1080 0.1456 0.2105 0.3243 0.3478 0.3371 0.2456
Bahir Dar 0.0771 0.0364 0.0526 0.0270 0.0435 0.0225 0.0432
Dire Dawa 0.0771 0.0243 0.1053 0.0541 0.0435 0.0337 0.0563
Nekemtie 0.1080 0.0364 0.1053 0.1081 0.0870 0.1348 0.0966
Dessie 0.0900 0.0291 0.1579 0.1081 0.0435 0.0674 0.0827
SUM 1 1 1 1 1 1 1
Table 16. Normalized Matrix with row average for Facility cost (Criterion 4)
Location Addis Ababa Hawassa Bahir Dar Dire Dawa Nekemtie Dessie Averages
Addis Ababa 0.0435 0.0233 0.0233 0.0341 0.0637 0.0545 0.0404
Hawassa 0.1304 0.0698 0.0465 0.0568 0.0764 0.0727 0.0754
Bahir Dar 0.1739 0.1395 0.0930 0.0568 0.0955 0.1091 0.1113
Dire Dawa 0.2174 0.2093 0.2791 0.1705 0.1911 0.1091 0.1961
Nekemtie 0.2609 0.3488 0.3721 0.3409 0.3822 0.4364 0.3569
Dessie 0.1739 0.2093 0.1860 0.3409 0.1911 0.2182 0.2199
SUM 1 1 1 1 1 1 1
5.2.4 Combine the vectors of preferences for each criterion, from above,
into one preference matrix that shows the preference for each
location for each criterion
11 | P a g e
Modelling and IT Application in SCM
Criterion
Location Market Potential Level of Transportation cost Facility cost
Competition
Addis Ababa 0.5609 0.4695 0.4757 0.0404
Hawassa 0.1447 0.1901 0.2456 0.0754
Bahir Dar 0.1117 0.1345 0.0432 0.1113
Dire Dawa 0.0816 0.0703 0.0563 0.1961
Nekemtie 0.0242 0.0283 0.0966 0.3569
Dessie 0.0768 0.1072 0.0827 0.2199
SUM 1 1 1 1
The above preference matrix shows the most favored location, best score, against a specific
criterion. Addis Ababa, for instance, is the most preferred location in terms of the three
criteria namely, Market potential, Level of competition and Transportation cost while
Nekemtie is the best preferred in terms of facility cost. The ranks for each criterion is
summarized as follows in the order of most preferred to least preferred.
Market Potential: Addis Ababa, Hawassa, Bahir Dar, Dire Dawa, Dessie, Nekemtie
Level of competition: Addis Ababa, Hawassa, Bahir Dar, Dessie, Dire Dawa, Nekemtie
Transportation cost: Addis Ababa, Hawassa, Nekemtie, Dessie, Dire Dawa, Bahir Dar
Facility cost: Nekemtie, Dessie, Dire Dawa, Bahir Dar, Hawassa, Addis Ababa
However, the manager should prefer the best location based on the cumulative effect of all
the criteria in contrast to the above results. Therefore, we proceed to the next level of
analysis.
12 | P a g e
Modelling and IT Application in SCM
Accordingly, the criteria should be ranked using the pairwise comparison and preference
scale the same way we did for the locations. The distribution manger used the preference
scale and did the same scoring for the criteria like he did for the locations which we use for
computing. Below are the developed pairwise matrix for the four criteria, and all other
matrices we applied for the locations above until we get a preference vector from
normalized matrix row averages.
Level of Transportation
Criterion Market Potential Competition cost Facility cost
Market Potential 1 3 7 7
Level of Competition 1/3 1 4 3
Transportation cost 1/7 1/4 1 1/2
Facility cost 1/7 1/3 2 1
Level of Transportation
Criterion Market Potential Competition cost Facility cost
Market Potential 1 3 7 7
Level of Competition 1/3 1 4 3
Transportation cost 1/7 1/4 1 ½
Facility cost 1/7 1/3 2 1
SUM 34/21 55/12 14 23/2
6.2 Compute the normalized matrix by dividing each value in each column
of the matrix by the corresponding column sum
Level of Transportation
Criterion Market Potential Competition cost Facility cost
Market Potential 21/34 36/55 1/2 14/23
Level of Competition 7/34 12/55 2/7 6/23
13 | P a g e
Modelling and IT Application in SCM
6.3 Develop the preference vector by computing the row averages for the
normalized matrix
Table 21. Normalized matrix with row average for the criteria shown in decimal
Level of Transportation
Criterion Market Competition cost Facility cost Row Average
Potential
Market Potential 0.6176 0.6545 0.5000 0.6087 0.5952
Level of Competition 0.2059 0.2182 0.2857 0.2609 0.2427
Transportation cost 0.0882 0.0545 0.0714 0.0435 0.0644
Facility cost 0.0882 0.0727 0.1429 0.0870 0.0977
SUM 1.00 1.00 1.00 1.00 1.00
From the above table we got the below preference vector for the selection criteria.
Criterion
Market Potential 0.5952
Level of Competition 0.2427
Transportation cost 0.0644
Facility cost 0.0977
The preference vector above quantitatively showed that the Market potential is the highest
priority criterion, Level of competition the second highest, Facility cost the third highest
and Transportation is the fourth and least priority criterion for the location decision.
Therefore, market potential is found to be the strongest criterion in terms of making the
location decision.
14 | P a g e
Modelling and IT Application in SCM
Using the AHP mathematical steps we have so far ranked the selection criteria in order of
priority and also the most preferred locations using each selection criterion. But at this
stage we will combine the two and compute an overall score by multiplying the criteria
preference matrix (table 17) by the preference vector (table 21) and sum up the products.
This step will give us the base to make our best recommendation for the location selection
decision. Below is how we do the computation.
Criterion
Location Market Level of Transportation Facility
Potential Competition cost cost
Addis Ababa 0.5609 0.4695 0.4757 0.0404 0.5952
Hawassa 0.1447 0.1901 0.2456 0.0754 0.2427
Bahirdar 0.1117 0.1345 0.0432 0.1113 0.0644
Diredawa 0.0816 0.0703 0.0563 0.1961 0.0977
Nekemtie 0.0242 0.0283 0.0966 0.3569
Dessie 0.0768 0.1072 0.0827 0.2199
By multiplying the above figures above, we get the below summary which is the overall
score used to rank the locations based on the magnitude of their scores.
15 | P a g e
Modelling and IT Application in SCM
Putting the above information together and ranking the alternatives, the below table
summarizes.
Table 22. Locations’ Overall Score and Rank
The above overall score reveals that Addis Ababa, Hawassa and Bahir Dar are the first three
locations recommended to have the depot while Dessie, Dire Dawa and Nekemtie are at the
bottom three position in their order of preference.
The whole process we have gone through above based on pairwise comparisons that
involved subjective judgement of the decision maker. As the decision maker makes
preferences, however, there may be a sort of inconsistency among his judgements. This
level of inconsistency is measured by a consistency index. The consistency index (CI) gives
information about logical consistency among pairwise comparison judgments. When CI
is=0, there is no logical inconsistency or the judgment is considered 100% consistent.
However, we don’t expect 100% consistency as judgements involve subjective human
elements. But for the model to be meaningful, the level of inconsistency should not be
greater than 10%. Below we will check the consistency of our analysis above by following
the appropriate steps of computing consistency index.
16 | P a g e
Modelling and IT Application in SCM
6.5.1 Multiply the pairwise comparison for each of the criteria by the
preference vector for the criteria and sum the products
6.5.2 Divide the results above by the corresponding values from the
preference vector and compute the sum and average of the results
17 | P a g e
Modelling and IT Application in SCM
We have said that due to subjective judgments made by decision makers, inconsistency is
experienced and that is normal. However, the level of inconsistency shouldn’t be out of
range as this means the whole analysis and thereby recommendation is meaningless. To
check if the level of inconsistency is acceptable or not we need to compare the consistency
index with the random index (RI), which is the consistency index of a randomly generated
pairwise comparison matrix. This RI is found from a table, below, for the same purpose and
we pick the RI for 4 criteria as we have used 4 criteria.
18 | P a g e
Modelling and IT Application in SCM
Generally, the degree of inconsistency is satisfactory and acceptable if the ratio is less than
0.1. If the ratio exceeds 0.1, the model would be considered to be meaningless. In our case,
the degree of inconsistency is within the range and hence the model results are meaningful.
Accordingly, we strongly recommend that Zebidar brewery Sc should have keep the current
distribution center (Depot) in Addis Ababa and considers the other locations according to
the sequence of their overall score-based preferences.
After we finalized the assignment, we have presented the results of the AHP to the
distribution manager and, in the first place, he was so grateful about our engagement
mentioning that he has learnt a lot from the assignment. He said the recommendation from
the model and some of his initial choices in mind didn’t match but will look into the
recommendation again and will most likely consider it.
19 | P a g e