Multi-Criteria Decision Making

Noted by Mỹ Dung K20 Logistics

MULTI-OBJECTIVE DECISION MAKING

There are k objectives considered simultaneously
Maximize 𝑍𝑗 = 𝑓𝑗 (𝑋) 𝑗 = 1, 2, … , 𝑘
Subject to:
𝑔𝑖 (𝑋) ≤ 𝑏𝑖 𝑖 = 1, 2, … , 𝑚

Where 𝑋 = (𝑥1 , 𝑥2 , … , 𝑥𝑛 )
𝑓𝑗 (𝑋) is the 𝑗𝑡ℎ objective function
𝑔𝑖 (𝑋) ≤ 𝑏𝑖 is the 𝑖𝑡ℎ constraint

Optimal solution:
- Definition: max/min value which achieving from all objectives simultaneously.
- In general, there is NO feasible optimal solution
- If objectives are not conflicting, then optimal solution can be found. However, by nature,
objectives are conflicting.
Efficient solution:
- A solution that no increase/decrease can be obtained at any objective without simultaneousely
decrease at least one of objectives.
- The solution is not unique.

Ideal solution:
- Optimal solution for each objective

1
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

How to do graphical method?

For each constraint:
Substitute 𝑥1 = 0 in the constraint, find 𝑥2 => we have point A(0, 𝑥2 )
Substitute 𝑥2 = 0 in the constraint, find 𝑥1 => we have point B(𝑥1 , 0)
Connect point A and B => we draw the constraint line
Choose the region base on the sign (≤ or ≥)

Example:

𝑥1 + 4𝑥2 ≤ 40

𝑥1 𝑥2
0 10
40 0

Connect point (0, 10) and point (40, 0)

=> we have a constraint line.

Which side is the feasible region?

Substitute any point on the graph, (ie point
(25,8) on the right side) on the constraint.

(25) + 4(8) = 57 > 40

However, the constraint says “≤”
=> Take the opposite side (the left side)

Some special cases:

1) 3𝑥1 + 𝑥2 = 0
Since when substitute 𝑥1 = 0 in the constraint, we get 𝑥2 = 0, and vice versa. So we have only 1 point (0, 0)
=> Cannot draw a line
=> To solve this, just substitute other value to get the second point.
𝑥1 𝑥2
0 0
1 -3

2) 3𝑥1 = 7 3) 5𝑥2 = 4
=> Whatever the value of 𝑥2 , 𝑥1 always equals to 7/3 => Whatever the value of 𝑥1 , 𝑥2 always equals to 4/5
=> Draw a vertical line at 7/3 (đường dọc) => Draw a horizontal line at 4/5 (đường ngang)
2
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

METHOD 1. GLOBAL CRITERION METHOD

Max 𝒇𝟏 (𝒙) = 0.4 𝑥1 + 0.3 𝑥2 , Max 𝒇𝟐 (𝒙) = 𝑥1
Subject to: 𝑔1 (𝑥) = 𝑥1 + 𝑥2 ≤ 400
𝑔2 (𝑥) = 2 𝑥1 + 𝑥2 ≤ 500
𝑥1 , 𝑥2 ≥ 0
Step 1: Obtain the ideal solutions / individual optimal solutions
Step 1.1: Draw graph of feasible region
Step 1.2: Determine corner points and substitute in all Z
Corner points 𝑍1 = 0.4𝑥1 + 0.3𝑥2 𝑍2 = 𝑥1
(max) (max)
A (0,0) 0 0
B (0,400) 120 0
C (100, 300) 130 100
D (250, 0) 100 250

Ideal solution (Z1): Z1 = 130, x1 = 100, x2 = 300

Ideal solution (Z2): Z2 = 250, x1 = 250, x2 = 0

Step 2: Construct a pay-off table

Z1 Z2 x1 x2
Z1 130 100 100 300
Z2 100 250 250 0

Step 3: Obtain the preferred solution

𝑘 𝑝
𝑍𝑗∗ (𝑥) − 𝑍𝑗 (𝑥)
𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒 𝑍 = ∑ [ ] , 𝑠. 𝑡 𝑡𝑜: 𝑔𝑖 (𝑥) ≤ 𝑏𝑖 ; 𝑖 = 1,2, . . , 𝑚
𝑍𝑗∗ (𝑥)
𝑗=1

𝑍𝑗∗ (𝑥): ideal solution of objective function 𝑗.

130 − 𝑍1 (𝑥) 250 − 𝑍2 (𝑥)
𝐶𝑎𝑠𝑒 𝑝 = 1: 𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒 𝑍 = +
130 250
Corner points Z1 Z2 Z
A (0,0) 0 0 2.00
B (0,400) 120 0 1.08
C (100, 300) 130 100 0.60
D (250, 0) 100 250 0.23

Since point D has the lowest Z, so the best efficient solution according to Global Criterion method is
x1 = 250, x2 = 0 with Z1 = 100, Z2 = 250

Case p = 2: (formulate model only)

130 − 𝑍1 (𝑥) 2 250 − 𝑍2 (𝑥) 2
𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒 𝑍 = ( ) +( )
130 250

3
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

!! Lưu ý: Đối với dạng max thì trình bày như vậy, còn nếu dạng minimize thì step 3 phải đổi vị trí TỬ SỐ thành
𝑍𝑗 (𝑥)−𝑍𝑗∗ (𝑥)
Ví dụ:
Min 𝒇𝟏 (𝒙) = 0.4 𝑥1 + 0.3 𝑥2 + 10, Max 𝒇𝟐 (𝒙) = 𝑥1
Subject to: 𝑔1 (𝑥) = 𝑥1 + 𝑥2 ≤ 400
𝑔2 (𝑥) = 2 𝑥1 + 𝑥2 ≤ 500
𝑥1 , 𝑥2 ≥ 0

Corner points 𝑍1 = 0.4𝑥1 + 0.3𝑥2 + 10 𝑍2 = 𝑥1

(min) (max)
A (0,0) 10 0
B (0,400) 130 0
C (100, 300) 140 100
D (250, 0) 110 250

Ideal solution (Z1): Z1 = 10, x1 = 0, x2 = 0

Ideal solution (Z2): Z2 = 250, x1 = 250, x2 = 0

Pay-off matrix

Z1 Z2 x1 x2
Z1 10 0 0 0
Z2 110 250 250 0

𝑍1 (𝑥) − 10 250 − 𝑍2 (𝑥)

𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒 𝑍 = +
10 250

4
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

METHOD 2. COMPROMISE PROGRAMMING

Step 1: Obtain the ideal and anti-ideal solutions

Corner points 𝑍1 = 0.4𝑥1 + 0.3𝑥2 𝑍2 = 𝑥1

(max) (max)
A (0,0) 0 0
B (0,400) 120 0
C (100, 300) 130 100
D (250, 0) 100 250

(Z1):
+ Ideal: Z1 = 130
+ Anti-ideal: Z1 = 0
(Z2):
+ Ideal: Z2 = 250
+ Anti-ideal: Z2 = 0

Step 2: Distance 𝑑𝑗
|𝑍𝑗∗ − 𝑍𝑗 (𝑥)|
𝑑𝑗 =
|𝑍𝑗∗ − 𝑍𝑗− |
|130 − 𝑍1 | |250 − 𝑍2 |
𝑑1 = ; 𝑑2 =
|130 − 0| |250 − 0|
Step 3: Calculate L for each corner point. Minimize L
𝑘 𝑝 1/𝑝
𝐿𝑝 (𝑊) = [(∑ 𝑊𝑗 𝑑𝑗 ) ]
𝑗=1

𝑊𝑗 : weights representing the relative importance of the 𝑗𝑡ℎ objective to the decision maker
𝑝: weight attached to the deviations according to their magnitudes
W1 = 1, W2 = 1, p = 1
min 𝐿1 = [(𝑊1 𝑑1 )𝑝 + (𝑊2 𝑑2 )𝑝 ]1/𝑝
|130 − 𝑍1 (𝑥)| |250 − 𝑍2 (𝑥)|
=1× +1×
|130 − 0| |250 − 0|
Corner points Z1 Z2 L
A (0,0) 0 0 2.00
B (0,400) 120 0 1.0
C (100, 300) 130 100 0.60
D (250, 0) 100 250 0.23

Since point D has the lowest L, so the best efficient solution according to Compromise Programming is
x1 = 250, x2 = 0 with Z1 = 100, Z2 = 250

5
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Formulate 𝐿∞ models
𝑚𝑖𝑛𝐿∞ (𝑊) = 𝑑

Subject to:
|𝑍1∗ − 𝑍1 (𝑥)|
𝑊1 × ≤𝑑
|𝑍1∗ − 𝑍1− |

|𝑍2∗ − 𝑍2 (𝑥)|
𝑊2 × ≤𝑑
|𝑍2∗ − 𝑍2− |
...
|𝑍𝑛∗ − 𝑍𝑛 (𝑥)|
𝑊𝑛 × ≤𝑑
|𝑍𝑛∗ − 𝑍𝑛− |

𝑥 ∈ Feasible region

Exercise: a. Find the ideal and anti-ideal solutions

Maximize 𝑍1 = 𝑥1 + 2𝑥2 b. Formulate the 𝐿1 and 𝐿∞ c by using Comprosmise
Minimize 𝑍2 = 5𝑥1 + 𝑥2 Programming method, given the weights of objective are
Subject to: 𝑊1 = 1, 𝑊2 = 1.
5𝑥1 − 3𝑥2 ≤ 14 c. Find the best efficient solution using 𝐿1 model
3𝑥1 + 2𝑥2 ≤ 20
𝑥1 + 3.5𝑥2 ≤ 15
2𝑥1 + 5𝑥2 ≥ 10
𝑥1 , 𝑥2 ≥ 0

a. Find the ideal and anti-ideal solutions

Corner points 𝑍1 = 𝑥1 + 2𝑥2 𝑍2 = 5𝑥1 + 𝑥2

(max) (min)
A (0, 2) 4 2
B (0, 4.29) 8.58 4.29
C (4.59, 2.98) 10.55 25.93
D (3.23, 0.71) 4.65 16.86

(Z1):
+ Ideal: Z1 = 10.55
+ Anti-ideal: Z1 = 4
(Z2):
+ Ideal: Z2 = 2
+ Anti-ideal: Z2 = 25.93

6
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

b. Formulate the 𝐿1 and 𝐿∞ c by using Comprosmise Programming method, given the weights of objective
are 𝑊1 = 1, 𝑊2 = 1.
|10.55 − 𝑍1 (𝑥)| |2 − 𝑍2 (𝑥)|
𝑚𝑖𝑛𝐿1 (𝑊) = 1 × +1×
|10.55 − 4| |2 − 25.93|

Subject to:
𝑥 ∈ Feasible region

𝑚𝑖𝑛𝐿∞ (𝑊) = 𝑑

Subject to:
|10.55 − 𝑍1 (𝑥)|
1× ≤𝑑
|10.55 − 4|

|2 − 𝑍2 (𝑥)|
1× ≤𝑑
|2 − 25.93|
𝑥 ∈ Feasible region

c. Find the best efficient solution using 𝐿1 model

Corner points 𝑍1 = 𝑥1 + 2𝑥2 𝑍2 = 5𝑥1 + 𝑥2 𝐿1

A (0, 2) 4 2 1
B (0, 4.29) 8.58 4.29 0.40
C (4.59, 2.98) 10.55 25.93 1
D (3.23, 0.71) 4.65 16.86 1.52

Since point B has the lowest L, so the best efficient solution according to Compromise Programming is
x1 = 0, x2 = 4.29 with Z1 = 8.58, Z2 = 4.29

7
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

METHOD 3. GOAL PROGRAMMING

1. Introduction
1.1. Concept
Simon’s argument: Human beings are more interested in reaching goals rather than in the abstract
concept of optimizing each outcome of the decision problem.

Example: possible ranking of goals for a hospital patient diet formation model
Priority level (rank) Goal
1 Satisfy nutritional requirements
2 Stay with daily budget
3 Meet patient’s food preferences

Goal 1 Goal 1 Goal 2 Goal 2

target achieved target achieved
Solution 1 100 90 100 90
Solution 2 100 100 100 80

1.2. Deviation
Deviations: the amount away from the desired standards or objectives:
- Overachievement/Overshoot (𝑑𝑖+ ≥ 0) and underachievement/undershoot (𝑑𝑖− ≥ 0)

- Desirable vs Undesireable deviation

Goal Example Expectation Objective in model
Max goals (≥) The profit should at least $700 M The more the better Min 𝑑𝑖−
𝑑𝑖+ desirable
𝑑𝑖− undesirable
Min goals (≤) The cost should at most $700 M The less the better Min 𝑑𝑖+
The labor hour is no more than 𝑑𝑖− desirable
120 hours 𝑑𝑖+ undesirable
Avoid
Exact goals (=) Equal Exactly equal Min 𝑑𝑖+ + 𝑑𝑖−
Both 𝑑𝑖+ and 𝑑𝑖− undesirable

- In GP, the objective is to minimize the (weighted) sum of undesirable deviation

(all undesirable 𝑑𝑖+ and 𝑑𝑖− →→ 0)
- For each goal, at least one of 𝑑𝑖+ and 𝑑𝑖− must be equal to “0”

8
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

1.3. Types of goal programming

Lexicographic (pre-emptive) Goals are assigned different priority levels
Weighted (non-preemptive) No goal is pre-determined to dominate any other goal
Chebyshev goal programming (Minmax) No priority, no weight, find a good balance for all objectives

2. Lexicographic (pre-emptive)
Example:
Suppose the weekly levels of two products to be manufactured are to be determined. Suppose our first
criterion relates to the amount of hours labor per week used to manufacture the two products A and B.
Assume our decision variables are
x1: The number of products of type A manufactured per week
x2: The number of products of type B manufactured per week
Suppose that the desired level is to use no more than 120h labor, and each type A product takes 4h to
manufacture and each type B product take 3h to manufacture.

The profit per item A is $100 and for product B is $150 per item. The company is somewhat unsure about
the level of profit to set as a target but give an initial estimate of $7000 per week.
The company has some strategic aims for their weekly production. They wish to maintain production of at
least 40 units of each of the products
The company has to purchase a minimum of 50L of this product weekly. Each item of type A uses 2L of the
product and each item of type B uses 1L of the product. Disposing of unused product is prohibitively
expensive. The second constraint relates to machine time, which caps the maximum combined production
of both products to 75 per week.

Suppose that the company has a clear order in which they wish to see the goals satisfied. An example of
such an order could be
Priority 1: Achieve the profit goal.
Priority 2: Achieve the strategic production goals.
Priority 3: Achieve the labor goal.

Step 0: (Draft) Determine the goal, priority of goal. Any hard constraint??
- Goal:
+ Goal 1: Achieve the lab.our goal (no more than 120h labour) (min) (priority 3)
+ Goal 2: Achieve the profit goal (at least $7000 per week) (max) (priority 1)
+ Goal 3: Achieve the strategic production goals (at least 40 units of each of the products) (max)
(priority 2)
- Hard constraint:
+ The company has to purchase a minimum of 50L of this product weekly.
+ Maximum combined production of both products to 75 per week.
9
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Step 1: Modelling
Minimize Z = P#(undesirable deviation) + P#(undesirable deviation) + ...
Subject to
Soft constraints:
Add: - overshoot + undershoot
Change sign (" ≥ ", " ≤ ") to “=”
Hard constraints
All variables ≥ 0,
Overshoot and undershoot ≥ 0.

Step 1.1. Leave the objective function blank. Fill from the “subject to” till the end.
For constraint of capacity (max bla bla), there is NO overshoot 𝑑𝑖+ since 𝑑𝑖+ = 0.
If the question says NORMAL capacity or there is goal of avoid overtime/overcapacity, there is still
both overshoot and undershoot.
Step 1.2. Based on the overshoot/undershoot notation and the type of goal (max goal/ min goal/
exact goal) or undesire, determine the objective in model

Minimize Z = 𝑃1(𝑑2− ) + 𝑃2(𝑑3− ) + 𝑃2(𝑑4− ) + 𝑃3(𝑑1+ )

Subject to
4𝑥1 + 3𝑥2 − 𝑑1+ + 𝑑1− = 120 (Labour)
100𝑥1 + 150𝑥2 − 𝑑2+ + 𝑑2− = 7000 (Profit)
𝑥1 − 𝑑3+ + 𝑑3− = 40 (Production)
𝑥2 − 𝑑4+ + 𝑑4− = 40 (Production)

2𝑥1 + 𝑥2 ≥ 50
𝑥1 + 𝑥2 ≤ 75

𝑥1 , 𝑥2 ≥ 0 , .
𝑑𝑖+ , 𝑑𝑖− ≥ 0, 𝑖 = 1, 2, 3, 4

Step 2: Draw graph

Step 2.1: Feasible region from hard constraint

10
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Step 2.2: Add soft constraint (the goals)

Step 2.3: Find feasible region from soft constraint follow the priority order
Minimize 𝑃1(𝑑2− )

Minimize 𝑃2(𝑑3− + 𝑑4− )

11
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Since goal 3 and goal 4 cannot be achieved at the same time, one of them has to be sacrifice

There are 2 feasible solutions (35, 40) and (40, 35)

Minimize 𝑃3(𝑑1+ ),
Cannot satisfied => choose point that give the lowest 𝑑1+ (current − priority goal) among feasible solutions
4𝑥1 + 3𝑥2 − 𝑑1+ + 𝑑1− = 120
(35, 40) => Labor = 4(35) + 3(40) = 260 (so với 120 thì dư (260 – 120 =) 140 => 𝑑1+ = 140)
(40, 35) => Labor = 4(40) + 3(35) = 265 (so với 120 thì dư (265 – 120 = ) 145 => 𝑑1+ = 145)
=> Choose (35, 40) since we want to minimize 𝑑1+

Tóm lại: Khi đang xét từng priority, đến priority nào mà mình không xét feasible region được nữa (do conflict
với các priority ở trên, thì mình sẽ chọn điểm trong feasible region nãy giờ mình làm, sao cho điểm đó có
minimum undesirable deviation (theo như objective). Trong bài này đang minimize 𝑃3(𝑑1+ ) nên mình chọn
điểm cho ra 𝑑1+ nhỏ nhất.
Nhắc lại kiến thức: overshoot 𝑑 + là phần bị thừa ra so với cái goal mình xét, undershoot 𝑑 − là phần bị hụt so
với goal mình xét.
Bài trên goal là 120, mà mình tính ra 260 với 265 lận => overshoot là 140 và 145.

Step 3: Solution
Goal Description Target level Satisfy? Achieved value
1 Labour 120 No 260
2 Profit 7000 Yes 9500
3 Type A produced 40 No 35
4 Type B produced 40 Yes 40

𝑥1 = 35, 𝑥2 = 40

12
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 1:
Product Labour (hour/unit) Material (lbs/unit) Profit ($/unit)
1 5 4 3
2 2 6 5
3 4 3 2

At present, the firm has a daily labor capacity of 240 available hours and a daily supply of 400 pounds of
material. The general linear programming formulation for this problem is as follows:
Management has developed the following set of goals, arranged in order of their importance to the firm:
(1) Because of recent labor relations difficulties, management wants to avoid underutilization of
normal production capacity.
(2) Management has established a satisfactory profit level of $500 per day.
(3) Overtime is to be minimized as much as possible.
(4) Management wants to minimize the purchase of additional materials to avoid handling and
storage problems.

Let 𝑥1 , 𝑥2 , 𝑥3 be the quantity of product 1, 2, and 3 produced, respectively

Minimize Z = 𝑃1(𝑑1− ) + 𝑃2(𝑑2− ) + 𝑃3(𝑑1+ ) + 𝑃4(𝑑3+ )

Subject to:
5𝑥1 + 2𝑥2 + 4𝑥3 − 𝑑1+ + 𝑑1− = 240 P1: Avoid (does not want) underutilization → minimize undershoot
3𝑥1 + 5𝑥2 + 2𝑥3 − 𝑑2+ + 𝑑2− = 500 P2: max goal of profit → minimize undershoot

P3: Minimize overtime as much as possible → minimize overshoot

No bound, no soft constraint
4𝑥1 + 6𝑥2 + 3𝑥3 − 𝑑3+ + 𝑑3− = 400 P4: minimize additional material → minimize overshoot

𝑥1 , 𝑥2 , 𝑥3 ≥ 0
𝑑𝑖+ , 𝑑𝑖− ≥ 0, 𝑖 = 1, 2, 3

13
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 2: The Growall Fertilizer Company produces three types of fertilizer—Supergro, Dynaplant, and
Soilsaver. The company has the capacity to produce a maximum of 2,000 tons of fertilizer in a week. It costs
$800 to produce a ton of Supergro, $1,500 for Dynaplant, and $500 for Soilsaver. The production process
requires 10 hours of labor for a ton of Supergro, 12 hours for a ton of Dynaplant, and 18 hours for a ton of
Soilsaver. The company has 800 hours of normal production labor available each week. Each week the
company can expect a demand for 800 tons of Supergro, 900 tons of Dynaplant, and 1,100 tons of Soilsaver.
The company has established the following goals, in order of their priority:
(1) The company does not want to spend over $20,000 per week on production, if possible.
(2) The company would like to limit overtime to 100 hours per week.
(3) The company wants to meet demand for all three fertilizers; however, it is twice as important
to meet the demand for Supergro (x1) as it is to meet the demand for Dynaplant (x2), and it is
twice as important to meet the demand for Dynaplant as it is to meet the demand for Soilsaver
(x3).
(4) It is desirable to avoid producing undercapacity, if possible.
(5) Because of union agreements, the company wants to avoid underutilization of labor.
Formulate a goal programming model to determine the number of tons of each brand of fertilizer to produce
to satisfy the goals.

Let 𝑥1 , 𝑥2 , 𝑥3 be the number of tons of fertilizer of brand Supergro, Dynaplant, and Soilsaver, respectively
Minimize Z = 𝑃1(𝑑1+ ) + 𝑃2(𝑑3+ ) + 𝟒𝑃3(𝑑4− ) + 𝟐𝑃3(𝑑5− ) + 𝑃3(𝑑6− ) + 𝑃4(𝑑7− ) + 𝑃5(𝑑2− )
Subject to:

800𝑥1 + 1500𝑥2 + 500𝑥3 − 𝑑1+ + 𝑑1− = 20000 P1: Does not want to spend more → minimize overshoot

10𝑥1 + 12𝑥2 + 18𝑥3 − 𝑑2+ + 𝑑2− = 800 Labor hour

𝑑2+ − 𝑑3+ + 𝑑3− = 100 P2: Limit (does not want) overtime → minimize overshoot

𝑥1 − 𝑑4+ + 𝑑4− = 800

P3: Does not want understock (production < demand)
𝑥2 − 𝑑5+ + 𝑑5− = 900 → minimize undershoot
𝑥3 − 𝑑6+ + 𝑑6− = 1,100
Since capacity is 2,000 => cannot have overshoot
𝑥1 + 𝑥2 + 𝑥3 + 𝑑7− = 2,000
Avoid (doest not want) undercapacity
→ minimize undershoot
Avoid (does not want) labor underutilization
→ minimize undershoot
Underutilization/undershoot chính là cái 𝑑2− trong soft
constraint 2 nên không cần viết lại nữa

𝑥1 , 𝑥2 , 𝑥3 ≥ 0
𝑑𝑖+ , 𝑑𝑖− ≥ 0, 𝑖 = 1, 2, 3, 4, 5, 6, 7

14
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 3: The East Midvale Textile Company produces denim and brushed-cotton cloth. The average pro-
duction rate for both types of cloth is 1,000 yards per hour, and the weekly production capacity (running two
shifts) is 80 hours. The marketing department estimates that the maxi- mum weekly demand is for 60,000
yards of denim and 35,000 yards of brushed cotton. The profit is $3.00 per yard for denim and $2.00 per yard
for brushed cotton. The company has es- tablished the following four goals, listed in order of importance:
(1) Eliminate underutilization of production capacity to maintain stable employment levels
(2) Limit overtime to 10 hours.
(3) Meet demand for denim and brushed cotton weighted according to profit for each.
(4) Minimize overtime as much as possible.
Formulate a goal programming model to determine the number of yards (in 1,000 yard lots) to
produce to satisfy the goals.

Let 𝑥1 , 𝑥2 be the number of yards (1,000 yard lots) of Denim and Brushed-cotton cloth, respectively

Minimize Z = 𝑃1(𝑑1− ) + 𝑃2(𝑑2+ ) + 𝟑𝑃3(𝑑3− ) + 𝟐𝑃3(𝑑4− ) + 𝑃4(𝑑1+ )

Subject to:

𝑥1 + 𝑥2 − 𝑑1+ + 𝑑1− = 80 P1: Eliminiate (does not want) underutilization → minimize undershoot

𝑑1+ − 𝑑2+ + 𝑑2− = 10 P2: Limit (does not want) overtime → minimize overshoot
𝑥1 − 𝑑3+ + 𝑑3− = 60 P3: Does not want understock (production < demand)
𝑥2 − 𝑑4+ + 𝑑4− = 35 → minimize undershoot

P4: Minimize overtime → minimize overshoot

(overtime chính là cái 𝑑1+ trong soft constraint 1 nên không cần viết lại
nữa)
𝑥1 , 𝑥2 ≥ 0
𝑑𝑖+ , 𝑑𝑖− ≥ 0, 𝑖 = 1, 2, 3, 4

15
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 4: The Bay City Parks and Recreation Department has received a federal grant of $600,000 to expand
its public recreation facilities. City council representatives have demanded four different types of facilities -
gymnasiums, athletic fields, tennis courts, and swimming pools. In fact, the demand by various communities
in the city has been for 7 gyms, 10 athletic fields, 8 tennis courts, and 12 swimming pools. Each facility costs a
certain amount, requires a certain number of acres, and is expected to be used a certain amount, as follows:
Facility Cost Required Acres Expected Usage (ppl/ week)
Gymnasium $80,000 4 1,500
Athletic field 24,000 8 3,000
Tennis court 15,000 3 500
Swimming pool 40,000 5 1,000

The Parks and Recreation Department has located 50 acres of land for construction (although more land could
be located, if necessary). The department has established the following goals, listed in order of their priority:
(1) The department wants to spend the total grant because any amount not spent must be returned to the
government.
(2) The department wants the facilities to be used by a total of at least 20,000 people each week.
(3) The department wants to avoid having to secure more than 50 acres of land already located.
(4) The department would like to meet the demands of the city council for new facilities. However, the goal
should be weighted according to the number of people expected to use each facility.

Let 𝑥1 , 𝑥2 , 𝑥3 , 𝑥4 be the number of gymnasiums, athletic field, tennis court, and swimming pool.

Minimize 𝑍 = 𝑃1(𝑑1− ) + 𝑃2(𝑑2− ) + 𝑃3(𝑑3+ ) + 𝟏. 𝟓𝑃4(𝑑4− ) + 𝟑𝑃4(𝑑5− ) + 𝟎. 𝟓𝑃4(𝑑6− ) + 𝟏𝑃4(𝑑7− )

Subject to:
P1: Do not want to spend less
80𝑥1 + 24𝑥2 + 15𝑥3 + 40𝑥4 − 𝑑1+ + 𝑑1− = 600
→ minimize undershoot
P2: Max goal (at least 20,000)
1.5𝑥1 + 3𝑥2 + 0.5𝑥3 + 1𝑥4 − 𝑑2+ + 𝑑2− = 20
→ minimize undershoot
P3: Min goal (no more than 50)
4𝑥1 + 8𝑥2 + 3𝑥3 + 5𝑥4 − 𝑑3+ + 𝑑3− = 50
→ minimize overshoot
𝑥1 − 𝑑4+ + 𝑑4− =7 P4: Meet demand = avoid understock
𝑥2 − 𝑑5+ + 𝑑5− = 10 → minimize undershoot
𝑥3 − 𝑑6+ + 𝑑6− =8
𝑥4 − 𝑑7+ + 𝑑7− = 12
.
𝑥1 , 𝑥2 , 𝑥3 , 𝑥4 ≥ 0
𝑑𝑖+ , 𝑑𝑖− ≥ 0, 𝑖 = 1, 2, 3, 4, 5, 6, 7

16
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 5: The Wearable Carpet Company manufactures two brands of carpet - shag and sculpture – in 100-
yard lots. It requires 8 hours to produce one lot of shag carpet and 6 hours to produce one lot of scuptured
carpet. The company has the following production goals, in prioritized order:
(1). Do not underutilize the production capacity, which is 480 hours.
(2). Achieve product demand of 40 (100-yard) lots for shag and 50 (100-yard) lots for sculptured carpet.
Meeting demand for shag is more important than meeting demand for scupture by a ratio of 5 to 2.
(3). Limit production overtime to 20 hours.

Formulate a goal programming model to determine the amount of shag and sculptured carpet to produce the
best meet the company’s goal.

Let 𝑥1 , 𝑥2 be the amount of shag and sculptured carpet (in 100-yard lots)

Minimize: 𝑍 = 𝑃1(𝑑1− ) + 𝟓𝑃2( 𝑑2− ) + 𝟐𝑃2( 𝑑3− ) + 𝑃3( 𝑑4+ )

Subject to:
P1: Avoid underutilization
8𝑥1 + 6𝑥2 − 𝑑1+ + 𝑑1− = 480
→ minimize undershoot
𝑥1 − 𝑑2+ + 𝑑2− = 40 P2: Meet demand = avoid understock
𝑥2 − 𝑑3+ + 𝑑3− = 50 → minimize undershoot
P3: Limit overtime
𝑑1+ − 𝑑4+ + 𝑑4− = 20
→ minimize overshoot

𝑥1 , 𝑥2 ≥ 0
𝑑𝑖+ , 𝑑𝑖− ≥ 0, 𝑖 = 1, 2, 3, 4

17
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 6: Computers Unlimited sells microcomputers and distributes them from three warehouses to four
universities. The available supply at each three warehouses, demand at four universities, and shipping costs
are shown in the following table:
University
Warehouse Supply
Tech A&M State Central
Richmond $22 17 30 18 420
Atlanta 15 35 20 25 610
Washington 28 21 16 14 340
Demand 520 250 400 380

Instead of original objective of cost minimization. Computers Unilimited has indicated the following goals,
arranged in order of their importance.
(1). A&M has been one of its better long-term customers, so Computers Unilimited wants to meet all of A&M’s
demands.
(2). Because of recent problems with the trucking union, it wants to ship at least 80 units from the Washington
warehouse to Central University.
(3). To maintain the best possible relations with all its customers. Computers Unlimited would like to meet no
less than 80% of each customer’s demands.
(4). It would like to keep total transportation costs to no more than 110% of the $22,470 total costs achieved
with the optimal allocation, using the transportation costs method.
(5). Because of the dissatisfaction with the trucking firrm it uses for the Atlanta-to-State deliveries. It would
like to minimize the number of units shipped over this route.

18
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Let 𝑥𝑖𝑗 be the quantity of microcomputers distributed by warehouse 𝒊 to serve university 𝒋 (i = 1, 2, 3, and j =
1, 2, 3, 4)

Minimize 𝑍 = 𝑃1(𝑑1− ) + 𝑃2(𝑑2− ) + 𝑃3(𝑑3− + 𝑑4− + 𝑑5− + 𝑑6− ) + 𝑃4(𝑑7+ ) + 𝑃5(𝑑8+ )

Subject to:
P1: meet demand = avoid understock
𝑥12 + 𝑥22 + 𝑥32 − 𝑑1+ + 𝑑1− = 250
→ minimize undershoot
P2: max goal (≥)
𝑥34 − 𝑑2+ + 𝑑2− = 80
→ minimize undershoot
𝑥11 + 𝑥21 + 𝑥31 − 𝑑3+ + 𝑑3− = 80% × 520
𝑥12 + 𝑥22 + 𝑥32 − 𝑑4+ + 𝑑4− = 80% × 250 P3: no less than (max goal (≥))
𝑥13 + 𝑥23 + 𝑥33 − 𝑑5+ + 𝑑5− = 80% × 400 → minimize undershoot
𝑥14 + 𝑥24 + 𝑥34 − 𝑑6+ + 𝑑6− = 80% × 380
22𝑥11 + 17𝑥12 + 30𝑥13 + 18𝑥14 + P4: no more than (min goal (≤))
15𝑥21 + 35𝑥22 + 20𝑥23 + 25𝑥24 + → minimize overshoot
28𝑥31 + 21𝑥32 + 16𝑥33 + 14𝑥34 − 𝑑7+ + 𝑑7− = 110% × 22,470
P5: minimize number of units shipped
𝑥23 − 𝑑8+ + 𝑑8− = 0 over this route = want 𝑥23 ≤ 0
(min goal) → minimize overshoot

𝑥11 + 𝑥12 + 𝑥13 + 𝑥14 ≤ 420

𝑥21 + 𝑥22 + 𝑥23 + 𝑥24 ≤ 610
𝑥31 + 𝑥32 + 𝑥33 + 𝑥34 ≤ 340

𝑥𝑖𝑗 ≥ 0, 𝑖 = 1, 2, 3, 𝑗 = 1, 2, 3, 4
𝑑𝑘+ , 𝑑𝑘− ≥ 0, 𝑘 = 1, 2, 3, 4, 5, 6, 7

19
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 7: The Oregon Atlantic Company produces two kinds of paper – newsprint and white wrapping
paper. It requires 5 minutes to produce a yard of newsprint and 8 minutes to produce a yard of wrapping
paper. The company has 4,800 minutes of normal production capacity available each week. The profit is $0.20
for a yard of newsprint and $0.25 for a yard of wrapping paper. The weekly demand is for 500 yards of
newsprint and 400 yards of wrapping paper.
The company has established the following goals, in order of priority:
(1) Limit overtime to 480 minutes.
(2) Achieve a profit of $300 each week.
(3) Fulfill the demand for the products in order of magnitude of their profits.
(4) Avoid underutilization of production capacity.

Let 𝑥1 , 𝑥2 be the number of yard of newsprint and yard of white wrapping paper.

Minimize 𝑍 = 𝑃1(𝑑2+ ) + 𝑃2(𝑑3− ) + 0.2𝑃3(𝑑4− ) + 0.25𝑃3(𝑑5− ) + 𝑃4(𝑑1− )

Subject to:
5𝑥1 + 8𝑥2 − 𝑑1+ + 𝑑1− = 4,800 Labor time
𝑑1+ − 𝑑2+ + 𝑑2− = 480 P1: limit overtime → minimize overshoot
0.2𝑥1 + 0.25𝑥2 − 𝑑3+ + 𝑑3− = 300 P2: max goal (≥)→ minimize undershoot
𝑥1 − 𝑑4+ + 𝑑4− = 500 P3: meet demand = avoid understock
𝑥2 − 𝑑5+ + 𝑑5− = 400 → minimize undershoot
P4: avoid underutilization

𝑥1 , 𝑥2 ≥ 0
𝑑𝑖+ , 𝑑𝑖− ≥ 0, 𝑖 = 1, 2, 3, 4, 5

20
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 8: Mac’s Warehouse is a large discount store that operates 7 days per week. The store needs the
following number of full-time employees working each day of the week:
Day Number of Employees Day Number of Employees
Sunday 47 Thursday 34
Monday 22 Friday 43
Tuesday 28 Saturday 53
Wednesday 35

Each employee must work 5 consecutive days each week and then have 2 days off. For example, any employee
who works Sunday through Thursday has Friday and Saturday off. The store currently has a total of 60
employees available to work. Mac’s has developed the following set of prioritized goals for employee
scheduling.
(1) The store would like to avoid hiring any additional employees.
(2) The most important days for the store to be fully staffed are Saturday and Sunday.
(3) The next most important day to be fully staffed is Friday.
(4) The store would like to be fully staffed the remaining 4 days in the week.
Formulate a goal programming model to determine the number of employees who should begin their 5-day
workweek each day of the week to achieve the store’s objectives.

21
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Let
𝑥1 be the number of employees begin their 5-day workweek on Sunday
𝑥2 be the number of employees begin their 5-day workweek on Monday
𝑥3 be the number of employees begin their 5-day workweek on Tuesday
𝑥4 be the number of employees begin their 5-day workweek on Wednesday
𝑥5 be the number of employees begin their 5-day workweek on Thursday
𝑥6 be the number of employees begin their 5-day workweek on Friday
𝑥7 be the number of employees begin their 5-day workweek on Saturday

Minimize 𝑍 = 𝑃1(𝑑1+ ) + 𝑃2(𝑑2− + 𝑑3− ) + 𝑃3(𝑑4− ) + 𝑃4(𝑑5− + 𝑑6− + 𝑑7− + 𝑑8− )

Subject to

𝑥1 + 𝑥2 + 𝑥3 + 𝑥4 + 𝑥5 + 𝑥6 + 𝑥7 − 𝑑1+ + 𝑑1− = 60 P1: Avoid hiring additional employees

→ minimize overshoot
Employees that work on Saturday include:
𝑥3 (who work on T3, T4, T5, T6, T7)
𝑥3 + 𝑥4 + 𝑥5 + 𝑥6 + 𝑥7 − 𝑑2+ + 𝑑2− = 53 (Sat)
𝑥4 (who work on T4, T5, T6, T7, CN)
𝑥5 (who work on T5, T6, T7, CN, T2)
𝑥6 (who work on T6, T7, CN, T2, T3)
𝑥7 (who work on T7, CN, T2, T3, T4)

𝑥4 + 𝑥5 + 𝑥6 + 𝑥7 + 𝑥1 − 𝑑3+ + 𝑑3− = 47 (Sun)

Same approach for Sunday
P2: Avoid lack of staff → minimize undershoot
𝑥2 + 𝑥3 + 𝑥4 + 𝑥5 + 𝑥6 − 𝑑4+ + 𝑑4− = 43 (Fri) P3: Avoid lack of staff → minimize undershoot
𝑥5 + 𝑥6 + 𝑥7 + 𝑥1 + 𝑥2 − 𝑑5+ + 𝑑5− = 22 (Mon)
𝑥6 + 𝑥7 + 𝑥1 + 𝑥2 + 𝑥3 − 𝑑6+ + 𝑑6− = 28 (Tue)
P4: Avoid lack of staff → minimize undershoot
𝑥7 + 𝑥1 + 𝑥2 + 𝑥3 + 𝑥4 − 𝑑7+ + 𝑑7− = 35 (Wed)
𝑥1 + 𝑥2 + 𝑥3 + 𝑥4 + 𝑥5 − 𝑑8+ + 𝑑8− = 34 (Thu)

𝑥𝑖 ≥ 0 , 𝑖 = 1, 2, … , 7

𝑑𝑗+ , 𝑑𝑗− ≥ 0, 𝑗 = 1, 2, 3, . . , 8

22
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 9: (Master exam)

The EMC company produces four parts of a machine. The manufacturing process consists of four operations
– A, B, C, D. The company has 20,000 square feet material delivered each month.
Part Material (ft2) Estimated Annual Demand Profits ($)
1 3 2000 80
2 1.5 1500 100
3 2 4100 85
4 3.2 1200 90
The processing times per unit for each operation and total available hours per year to produce these parts
are as follows:
Part (hr./unit) Total
Operation
1 2 3 4 hours/year
A 0.05 0.03 0.10 0.12 500
B 0.01 0.02 - - 680
C 0.04 0.17 0.08 0.14 570
D 0.07 0.12 0.05 0.13 900

The company has the following prioritized production goals:

(1) Avoid overtime. (2) Meet parts demand. (3) Achieve an annual profit of $800,000
(4) Avoid ordering more material because a surcharge is required by the supplier for changing the standard
monthly order.

Let 𝑥1 , 𝑥2 , 𝑥3 , 𝑥4 be the number of part 1, 2, 3, 4 that should be produced monthly

Minimize 𝑍 = 𝑃1(𝑑1+ + 𝑑2+ + 𝑑3+ + 𝑑4+ ) + 𝑃2( 𝑑5− + 𝑑6− + 𝑑7− + 𝑑8− ) + 𝑃3( 𝑑9− ) + 𝑃4( 𝑑10
+ )

Subject to:
12 × (0.05𝑥1 + 0.03𝑥2 + 0.10𝑥3 + 0.12𝑥4 ) − 𝑑1+ + 𝑑1− = 500 P1: Avoid overtime
12 × (0.01𝑥1 + 0.02𝑥2 ) − 𝑑2+ + 𝑑2− = 680 → minimize overshoot
12 × (0.04𝑥1 + 0.17𝑥2 + 0.08𝑥3 + 0.14𝑥4 ) − 𝑑3+ + 𝑑3− = 570 The capacity is in year so requiring
12 × (0.07𝑥1 + 0.12𝑥2 + 0.05𝑥3 + 0.13𝑥4 ) − 𝑑4+ + 𝑑4− = 900 multiply 𝑥𝑖 by 12 months.

12 × 𝑥1 − 𝑑5+ + 𝑑5− = 2000 P2: Meet demand = avoid understock

12 × 𝑥2 − 𝑑6+ + 𝑑6− = 1500 → minimize undershoot
12 × 𝑥3 − 𝑑7+ + 𝑑7− = 4100 The demand is in year so requiring
12 × 𝑥4 − 𝑑8+ + 𝑑8− = 1200 multiply 𝑥𝑖 by 12 months
P3: max goal → minimize undershoot
12 × (80𝑥1 + 100𝑥2 + 85𝑥3 + 90𝑥4 ) − 𝑑9+ + 𝑑9− = 800,000 The demand is in year so requiring
multiply 𝑥𝑖 by 12 months
P4: Avoid ordering more material
+ −
3𝑥1 + 1.5𝑥2 + 2𝑥3 + 3.2𝑥4 − 𝑑10 + 𝑑10 = 20,000
→ minimize overshoot

𝑥1 , 𝑥2 𝑥3 , 𝑥4 ≥ 0
𝑑𝑖+ , 𝑑𝑖− ≥ 0, 𝑖 = 1, 2, . . , 10

23
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

METHOD 4. DE NOVO PROGRAMMING

max 𝑧 = 𝐶𝑥 C Coefficient on objective function

𝑠𝑡: 𝐴𝑥 − 𝑏 ≤ 0 A Resource consumption for each decision variable
𝑝𝑏 ≤ 𝐵 b Resource available
𝑥≥0 x Decision variable
p Unit price of resource
B Total available budget

Step 1: Formulate model and reconstruct model

Model Reconstructed model
max 𝑧 = 𝐶𝑥 max 𝑧 = 𝐶𝑥
𝑠𝑡: 𝑠𝑡:
𝐴𝑥 ≤ 𝑏 𝑣𝑥 ≤ 𝐵
𝑥≥0 𝑥≥0
Where 𝑣 = 𝑝𝐴

Constraint phải luôn dưới dạng 𝐴𝑥 ≤ 𝑏, nếu có constraint 𝑎𝑥 ≥ 𝑏 thì phải xử lý. Xem ở trang 32

Model (from the question) Reconstruct model

Maximize: C Maximize: C
𝑍1 = 50𝑥1 + 100𝑥2 + 17.5𝑥3 𝑍1 = 50𝑥1 + 100𝑥2 + 17.5𝑥3
𝑍2 = 92𝑥1 + 75𝑥2 + 50𝑥3 𝑍2 = 92𝑥1 + 75𝑥2 + 50𝑥3
𝑍3 = 25𝑥1 + 100𝑥2 + 75𝑥3 𝑍3 = 25𝑥1 + 100𝑥2 + 75𝑥3
Subject to: Subject to:
12𝑥1 + 17𝑥2 + ≤ 1400 𝟎. 𝟕𝟓 (12𝑥1 + 17𝑥2 )
p A
A 3𝑥1 + 9𝑥2 + 8𝑥3 ≤ 1000 +𝟎. 𝟔 ( 3𝑥1 + 9𝑥2 + 8𝑥3 )
b(RHS)
10𝑥1 + 13𝑥2 + 15𝑥3 ≤ 1750 +𝟎. 𝟑𝟓 (10𝑥1 + 13𝑥2 + 15𝑥3 )
6𝑥1 + +16𝑥3 ≤ 1325 +𝟎. 𝟓 ( 6𝑥1 + +16𝑥3 )
12𝑥2 + 7𝑥3 ≤ 900 +𝟏. 𝟏𝟓 ( 12𝑥2 + 7𝑥3 )
9.5𝑥1 + 9.5𝑥2 + 4𝑥3 ≤ 1075 +𝟎. 𝟔𝟓 (9.5𝑥1 + 9.5𝑥2 + 4𝑥3 )
≤ 𝟒𝟔𝟓𝟖. 𝟕5
𝑥1 , 𝑥2 , 𝑥3 ≥ 0 𝑥1 , 𝑥2 , 𝑥3 ≥ 0
B
Given 𝑝1 = 0.75, 𝑝2 = 0.60, 𝑝3 = 0.35,
𝑝4 = 0.5, 𝑝5 = 1.15, 𝑝6 = 0.65

Total available budget B = sum (available resource (RHS) x unit price) = bp

= 1400 × 0.75 + 1000 × 0.60 + 1750 × 0.35 + 1325 × 0.50 + 900 × 1.15 + 1075 × 0.65
= 4658.75

24
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Step 2: Find ideal solution 𝑍 ∗

Reconstruct model “Thu gọn” version
Maximize: Maximize:
𝑍1 = 50𝑥1 + 100𝑥2 + 17.5𝑥3 𝑍1 = 50𝑥1 + 100𝑥2 + 17.5𝑥3
𝑍2 = 92𝑥1 + 75𝑥2 + 50𝑥3 𝑍2 = 92𝑥1 + 75𝑥2 + 50𝑥3
𝑍3 = 25𝑥1 + 100𝑥2 + 75𝑥3 𝑍3 = 25𝑥1 + 100𝑥2 + 75𝑥3
Subject to: Subject to:
𝟎. 𝟕𝟓 (12𝑥1 + 17𝑥2 ) 𝟐𝟑. 𝟒𝟕𝟓𝑥1 + 𝟒𝟐. 𝟔𝟕𝟓𝑥2 + 𝟐𝟖. 𝟕𝑥3 ≤ 𝟒𝟔𝟓𝟖. 𝟕5
+𝟎. 𝟔 ( 3𝑥1 + 9𝑥2 + 8𝑥3 )
+𝟎. 𝟑𝟓 (10𝑥1 + 13𝑥2 + 15𝑥3 ) 𝑥1 , 𝑥2 , 𝑥3 ≥ 0
+𝟎. 𝟓 ( 6𝑥1 + +16𝑥3 )
+𝟏. 𝟏𝟓 ( 12𝑥2 + 7𝑥3 )
+𝟎. 𝟔𝟓 (9.5𝑥1 + 9.5𝑥2 + 4𝑥3 )
≤ 𝟒𝟔𝟓𝟖. 𝟕5

𝑥1 , 𝑥2 , 𝑥3 ≥ 0

𝟐𝟑. 𝟒𝟕𝟓 = 𝟎. 𝟕𝟓 × 12 + 𝟎. 𝟔 × 3 + 𝟎. 𝟑𝟓 × 10 + 𝟎. 𝟓 × 6 + 𝟏. 𝟏𝟓 × 0 + 𝟎. 𝟔𝟓 × 9.5

𝟒𝟐. 𝟔𝟕𝟓 = 𝟎. 𝟕𝟓 × 17 + 𝟎. 𝟔 × 9 + 𝟎. 𝟑𝟓 × 13 + 𝟎. 𝟓 × 0 + 𝟏. 𝟏𝟓 × 12 + 𝟎. 𝟔𝟓 × 9.5
𝟐𝟖. 𝟕 = 𝟎. 𝟕𝟓 × 0 + 𝟎. 𝟔 × 8 + 𝟎. 𝟑𝟓 × 15 + 𝟎. 𝟓 × 16 + 𝟏. 𝟏𝟓 × 7 + 𝟎. 𝟔𝟓 × 4

Model Solution
Maximize:
𝒁𝟏 = 𝟓𝟎𝒙𝟏 + 𝟏𝟎𝟎𝒙𝟐 + 𝟏𝟕. 𝟓𝒙𝟑 𝑥1 = 0, 𝑥2 = 109, 𝑥3 = 0
Subject to:
23.475𝑥1 + 42.675𝑥2 + 28.7𝑥3 ≤ 4658.75 𝑍1∗ = 10915.696
𝑥1 , 𝑥2 , 𝑥3 ≥ 0

Maximize:
𝒁𝟐 = 𝟗𝟐𝒙𝟏 + 𝟕𝟓𝒙𝟐 + 𝟓𝟎𝒙𝟑 𝑥1 = 198.4537, 𝑥2 = 0, 𝑥3 = 0
Subject to:
23.475𝑥1 + 42.675𝑥2 + 28.7𝑥3 ≤ 4658.75 𝑍2∗ = 18257.738
𝑥1 , 𝑥2 , 𝑥3 ≥ 0

Maximize:
𝒁𝟑 = 𝟐𝟓𝒙𝟏 + 𝟏𝟎𝟎𝒙𝟐 + 𝟕𝟓𝒙𝟑 𝑥1 = 0, 𝑥2 = 0, 𝑥3 = 162.324
Subject to:
23.475𝑥1 + 42.675𝑥2 + 28.7𝑥3 ≤ 4658.75 𝑍3∗ = 12174.3031
𝑥1 , 𝑥2 , 𝑥3 ≥ 0

25
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Step 3: Construct a meta-optimum problem

𝑀𝑖𝑛 𝐵 = 𝑣𝑥
𝑠𝑡: 𝐶𝑥 ≥ 𝑧 ∗
𝑥≥0
Step 3.1: Find 𝐵 ∗ and 𝑥 ∗

𝑀𝑖𝑛 𝐵 = 23.475𝑥1 + 42.675𝑥2 + 28.7𝑥3

Subject to:
𝑍1 = 50𝑥1 + 100𝑥2 + 17.5𝑥3 ≥ 10915.696
𝑍2 = 92𝑥1 + 75𝑥2 + 50𝑥3 ≥ 18257.738
𝑍3 = 25𝑥1 + 100𝑥2 + 75𝑥3 ≥ 12174.3031

𝑥1 , 𝑥2 , 𝑥3 , ≥ 0

Solution
𝐵 ∗ = 6616.563
𝑥1∗ = 131.341, 𝑥2∗ = 20.683, 𝑥3∗ = 78.976

Step 3.2: Find 𝑏∗

𝑏∗ = 𝐴𝑥 ∗

𝑏1∗ = 12𝑥1∗ + 17𝑥2∗ + = 2080.703

𝑏2∗ = 3𝑥1∗ + 9𝑥2∗ + 8𝑥3∗ = 1292.978
𝑏3∗ = 10𝑥1∗ + 13𝑥2∗ + 15𝑥3∗ = 2883.929
𝑏4∗ = 6𝑥1∗ + +16𝑥3∗ = 2051.662
𝑏5∗ = 12𝑥2∗ + 7𝑥3∗ = 909.028
𝑏6∗ = 9.5𝑥1∗ + 9.5𝑥2∗ + 4𝑥3∗ = 1845.632

26
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Step 4: Use the optimum-path ration to obtain the final solution

Step 4.1: Find r
𝐵
𝑟=
𝐵∗
𝐵 4658.75
𝑟= = = 0.70385
𝐵 ∗ 6616.319

Step 4.2: Find 𝑥𝑖

𝑥𝑖 = 𝑟 × 𝑥𝑖∗

𝑥1 𝑥1∗ 92.48
[𝑥2 ] = 𝑟 × [𝑥2∗ ] = [20.90]
𝑥3 𝑥3∗ 55.61

Step 4.3: Find 𝑧𝑖

𝑧𝑖 = 𝑟 × 𝑧𝑖∗

𝑧1 𝑧1∗ 7686
[𝑧2 ] = 𝑟 × [𝑧2∗ ] = [12855.89]
𝑧3 𝑧3∗ 8572.4

Step 4.4: Find 𝑏𝑗

𝑏𝑗 = 𝑟 × 𝑏𝑗∗

𝑏1 𝑏1∗ 1560.2700
𝑏2 𝑏2∗ 775.7868
𝑏3 𝑏3∗ 1009.3750
=𝑟× =
𝑏4 𝑏4∗ 1025.8310
𝑏5 𝑏5∗ 1045.3820
[𝑏6 ] [𝑏6∗ ] [1199.6610]

27
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 1:
𝑀𝑎𝑥 𝑓1 = 400𝑥1 + 300𝑥2 Given 𝑝1 = 30, 𝑝2 , = 40, 𝑝3 = 9.5, 𝑝4 = 20, and 𝑝5 = 10 are
𝑀𝑎𝑥 𝑓2 = 6𝑥1 + 8𝑥2 market price ($ per unit)
Subject to:
4𝑥1 ≤ 10
a) Find the ideal solution
2𝑥1 + 6𝑥2 ≤ 12
b) Find min Budget to get the ideal solution
12𝑥1 + 4𝑥2 ≤ 30
c) If given B = $2,700, find f1, f2
3𝑥2 ≤ 5.25
d) Redesign the model with B = $2,700
4𝑥1 + 4𝑥2 ≤ 13
𝑥1 , 𝑥2 ≥ 0

Draft: Define p, A, v, b, B, C
4 0 10
2 6 12
400 300
𝒑 = [30 40 9.5 20 10] ; 𝑨 = 12 6 ; 𝒗 = 𝑝𝐴 ; 𝒃 = 30 ; 𝑩 = 𝑏𝑝 ; 𝑪 = [ ]
6 8
0 3 5.25
[4 4] [ 13 ]

a) Find the ideal solution

B = 10 × 30 + 12 × 40 + 30 × 9.5 + 5.25 × 20 + 13 × 10 = 1300

Reconstructed model:

𝑀𝑎𝑥 𝑓1 = 400𝑥1 + 300𝑥2

𝑀𝑎𝑥 𝑓2 = 6𝑥1 + 8𝑥2
Subject to:
30 × ( 4𝑥1 )
+ 40 × ( 2𝑥1 + 6𝑥2 )
+ 9.5 × (12𝑥1 + 4𝑥2 )
+ 20 × ( 3𝑥2 )
+ 10 × ( 4𝑥1 + 4𝑥2 ) ≤ 1300

⇔ 354𝑥1 + 378𝑥2 ≤ 1300

𝑥1 , 𝑥2 ≥ 0
Graph of feasible region

Corner points 𝑓1 = 400𝑥1 + 300𝑥2 𝑓2 = 6𝑥1 + 8𝑥2

A(0, 0) 0 0
B(3.67, 0) 1468 22.02
C(0, 3.44) 1032 27.52

Ideal solution 𝑓1∗ = 1468, 𝑥1 = 3.67, 𝑥2 = 0

Ideal solution 𝑓2∗ = 27.52, 𝑥1 = 0, 𝑥2 = 3.44

28
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

b) Meta-optimal model
𝑀𝑖𝑛 𝐵 = 354𝑥1 + 378𝑥2
Subject to:
𝑓1 = 400𝑥1 + 300𝑥2 ≥ 1468
𝑓2 = 6𝑥1 + 8𝑥2 ≥ 27.52
𝑥1 , 𝑥2 ≥ 0

Corner points 𝐵 = 354𝑥1 + 378𝑥2

A(0, 4.89) 1848.42
B(4.59, 0) 1624.86
C(2.49, 1.57) 1474.92

Minimum 𝐵 ∗ = 1474.92
With 𝑥1∗ = 2.49, 𝑥2∗ = 1.57
𝑏1∗ 4 0 9.96
𝑏2∗ 2 6 14.4
2.49
𝑏3∗ = 𝐴𝑥 ∗ = 12 4 . [ ] = 36.16
1.57
𝑏4∗ 0 3 4.71
[𝑏5∗ ] [4 4] [16.24]

c) If given B = $2,700, find f1, f2

𝐵 2,700
𝑟= = = 1.83
𝐵 ∗ 1,474.92
𝑓1 𝑓1∗ 1468 2686.44
[ ] = 𝑟 × [ ∗ ] = 1.83 × [ ]=[ ]
𝑓2 𝑓2 27.52 50.36
So f1 = 2686.44, f2 = 50.36

d) Redesign the model with B = $2,700

With B = $2,700, we have r and f1, f2 as calculated on question c
𝑥1 𝑥1∗ 2.49 4.56
[ ] = 𝑟 × [ ∗ ] = 1.83 × [ ]=[ ]
𝑥2 𝑥2 1.57 2.87
𝑏1 𝑏1∗ 9.96 18.23
𝑏2 𝑏2∗ 14.4 26.35
𝑏3 = 𝑟 × 𝑏3∗ = 1.83 × 36.16 = 36.16
𝑏4 𝑏4∗ 4.71 66.17
[𝑏5 ] [𝑏5∗ ] [16.24] [29.72]

29
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 2:

Max 𝑍1 = 𝑥1 + 𝑥2 Unit price: p = [0.5 0.75]

Max 𝑍2 = 𝑥1 + 4𝑥2 a) Find the ideal solution
Subject to: b) Find min Budget to get the ideal solution
3𝑥1 + 4𝑥2 ≤ 60 c) Find Z1, Z2 values with original value
𝑥1 + 3𝑥2 ≤ 30
𝑥1 , 𝑥2 ≥ 0

a) Find the ideal solution

Budget = 60 × 50 + 30 × 0.75 = 52.5
Reconstructed model

Max 𝑍1 = 𝑥1 + 𝑥2

Max 𝑍2 = 𝑥1 + 4𝑥2

Subject to:

0.5 × (3𝑥1 + 4𝑥2 )

+ 0.75 × (𝑥1 + 3𝑥2 ) ≤ 52.5

⇔ 2.25𝑥1 + 4.25𝑥2 ≤ 52.5

𝑥1 , 𝑥2 ≥ 0

Graphical method:

Corner points 𝑍1 = 𝑥1 + 𝑥2 𝑍2 = 𝑥1 + 4𝑥2

A(0, 0) 0 0
B(23.33, 0) 23.33 23.33
C(0, 12.35) 12.35 49.4

Ideal solution: 𝑍1∗ = 23.33, 𝑥1 = 23.33, 𝑥2 = 0

Ideal solution: 𝑍2∗ = 49.4, 𝑥1 = 0, 𝑥2 = 12.35

30
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

b) Find min Budget to get the ideal solution

𝑀𝑖𝑛 𝐵 = 2.25𝑥1 + 4.25𝑥2

Subject to:

𝑍1 = 𝑥1 + 𝑥2 ≥ 23.33

𝑍2 = 𝑥1 + 4𝑥2 ≥ 49.4

𝑥1 , 𝑥2 ≥ 0

Corner points 𝐵 = 2.25𝑥1 + 4.25𝑥2

A(0, 23.33) 99.15
B(14.64, 8.69) 69.87
C(49.4, 0) 111.15

Minimum 𝐵 ∗ = 69.87
With 𝑥1∗ = 14.64, 𝑥2∗ = 8.69

c) Find Z1, Z2 values with original value

𝐵 52.5
𝑟= = = 0.75
𝐵 ∗ 69.87
𝑍1 𝑍1∗ 23.33 17.5
[ ]=𝑟× [ ] = 0.75 × [ ]=[ ]
𝑍2 𝑍2∗ 49.4 37.05
So Z1 = 17.5 and Z2 = 37.05

31
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

Exercise 3: (Master exam)

Consider the following problem Given unit price of resource (1) is $3, resource (2) is $2, and
Max 𝑍1 = 3𝑥1 + 2𝑥2 resource (3) is $2, budget B = $53, aspiration level vector
Max 𝑍2 = 3𝑥2 𝑍1∗ (10, 0) = 𝟑𝟎, 𝑍2∗ (2.7, 1.8) = 𝟓. 𝟓
Subject to:
7𝑥1 − 𝑥2 ≥ 17 a. Identify the minimum budget 𝐵 ∗ and its corresponding resource
𝑥1 + 4𝑥2 ≤ 10 allocation with the aspiraction level.
𝑥1 + 2𝑥2 ≤ 13 b. Use the optimum-path ration to obtain the final solution.
𝑥1 , 𝑥2 ≥ 0

a. Identify the minimum budget 𝐵 ∗ and its corresponding resource allocation with the aspiraction level.
7𝑥1 − 𝑥2 ≥ 17 ⇔ −7𝑥1 + 𝑥2 ≤ −17
−7 1
𝑣 = 𝑝𝐴 = [3 2 2] . [ 1 4] = [−17 15]
1 2
Other way:
3 × (−7𝑥1 + 𝑥2 )
+2×( 𝑥1 + 4𝑥2 )
+2×( 𝑥1 + 2𝑥2 )
= −17𝑥1 + 15𝑥2

Min 𝐵 = −17𝑥1 + 15𝑥2

Subject to:
3𝑥1 + 2𝑥2 ≥ 30
3𝑥2 ≥ 5.5
𝑥1 , 𝑥2 ≥ 0
Graphical method
.
Corner points 𝐵 = −17𝑥1 + 15𝑥2
A(0, 15) 225
B(8.78, 1.83) -121.81

Minimum 𝐵 ∗ = 225
With 𝑥1∗ = 0, 𝑥2∗ = 15

32
 Multi-Criteria Decision Making
Noted by Mỹ Dung K20 Logistics

b. Use the optimum-path ration to obtain the final solution.

𝐵 53
𝑟= = = 0.24
𝐵 ∗ 225

𝑥1 𝑥1∗ 0 0
[ ] = 𝑟 × [ ∗ ] = 0.24 × [ ] = [ ]
𝑥2 𝑥2 15 53.6

𝑍1 𝑍1∗ 30 7.2
[ ]=𝑟× [ ] = 0.24 × [ ]=[ ]
𝑍2 𝑍2∗ 5.5 1.32

𝑏1∗ −7 1 15
0
[𝑏2∗ ] ∗
= 𝐴𝑥 = [ 1 4] . [ ] = [60]
15
𝑏3∗ 1 2 30

𝑏1 𝑏1∗ 15 3.6
[𝑏2 ] = 𝑟 × [𝑏2∗ ] = 0.24 × [60] = [14.4]
𝑏3 𝑏3∗ 30 7.2

33