1

Multi-objective linear programming practical cases

Julio Cesar Ramírez Pisco

Luis Alfredo Montoya Tafur 20192377039

Brayan Camilo Poveda Fajardo 20192377022

Francisco José de Caldas District University

Production Engineering
Deterministic Production Models
Bogotá D.C.
2020
 2

Multi-objective linear programming practical activities (Case 1)

Problem: According to the summary of the case study presented in (Morales, 2014); The company wants to maximize
the income obtained from the sale of tickets sold per passenger transported by class in two types of
Airbus (a)–Boeing (b) planes and at the same time minimize fuel consumption per passenger expressed
in liters per 100 km of the two planes. The benefits for (a) are represented as €25354 and for
€22,215 after buying costs and benefits; regarding fuel consumption for (a) it is
9.715 L/Km and for (b) 6.21 L/Km after the calculations regarding the number of passengers and consumption.
individual/100; additionally, it is said that for maintenance at least 2 units must be purchased of
Airbus (a); the acquisition price of (a) is 225,000,000 € and that of (b) is 180,000,000, which should not
exceed a budget of 1,350,000,000; it is also said that the number of daily trips that can be made
Model A is 2 while for model B it is 3 to cover a daily demand of trips.
less 10 that the company must do.

Then modeling the objectives individually and evaluating the results, it is found that the company is
ready to
Consume a maximum of 47L/KM of fuel.
Obtain at least €140,000 as profit.

Primal Model

Objective Minimize: Z = 9.715*a + 6.21*b

S.A.R

-Buy Air bus: a ≥ 2 Budget: 225000000*a + 180000000*b ≤ 1350000000;

Acceptable Benefit: 25354*a + 22215*b ≤ 140000 Requested trips: 2*a + 3*b ≤ 10;
-No negativity: a, b ≥ 0
Case 1: AIR BUS - BOEING

GAMS Code
 3

Results Analysis

By using the method of constraints and under the

last goal scheduled below the acceptance level of
the company, prioritizing the minimization of
consumed fuel and taking the maximization of
benefits as a restriction in which the benefits
acceptable are at least 140,000 €, obtained in the
solution a consumption of 44.391 L/Km; therefore it
fully meets the proposed goals of the
company; a profit of €140,000 and a consumption that
does not exceed 47L/km with 44.391L/km. From this
result part of a second analysis in which one can
take as a priority the benefit taking the consumption
as a restriction at the edge of 47L/Km taking the
with the aim of maximizing profit by obtaining
149334 €; according to the analysis made, the best alternative is
take as a restriction the minimization of
combustible and leave the acceptable benefits (A
approximate result obtained with the model in
NICE for goals developed in the text) of
(Morales, 2014).

In the context of multi-objective linear programming, a great variety of

alternatives or generating techniques that show efficient results for a problem
with multiple requirements, these problems resemble reality more as they
they must consider the optimization of these requirements according to the arrangement of the
company or resources of the proposing person or entity, these methods are: weighting,
restrictions and goals; for the first contextual activity, the method of restrictions is used
since it fits the allocation of resources that the company is willing to undertake, trying
one individual objective and the other taken as a constraint as shown above
table.

-Practical case 1 modeled in the G.A.M.S. software summarized and adapted from:
(Morales, 2014)
 4

Multi-objective linear programming practical activities (Case 2)

Problem: According to the case study summary presented in (Ramírez, Arias, & Ledesma, 2015); The company
Yogures YOYO has three different flavors; Strawberry, Blackberry, and Peach, and two machines.
empacadoras EMP1, EMP2 para el empaque y saborizacion de estas referencias.
According to a marketing study conducted, there will be an increase in demand that exceeds quarterly capacity.
from the packers in the following way; Strawberry 830 units, Blackberry 1100 units, and peach 520 units. For
to meet the demand there must be extra hours for the quarter with a value of $15410 for EMP1 and $550 for EMP2, this
value takes into account water, energy, and raw materials consumed.
La capacidad de las maquinas es: Fresa 9 EMP1; 3 EMP2, Mora 9 EMP1; 6 EMP2, y Melocotón 6 EMP1;
3 EMP2.
There is a percentage of waste associated with each machine and product is associated by Kg/h to: Strawberry
[{"EMP":"EMP1","value":"3.10"},{"EMP":"EMP2","value":"2.2"},{"fruit":"Mora","EMP":"EMP1","value":"2.5"},{"EMP":"EMP2","value":"3.2"},{"fruit":"Melocotón","EMP":"EMP1","value":"3.5"},{"EMP":"EMP2","value":"1.8"}]
The production manager wants to know the ideal mix of overtime to schedule his packers next.
quarter minimizing waste and costs associated with machine Set-Up starts.

Modeling the two objectives of minimizing costs and minimizing waste individually, the company...
propose two goals:
Minimize waste to approximately 899.4 Kg
Minimize costs to approximately $3,392,220.

Primal Model
Objective Minimize: Z = Q
S.A.R
-Strawberry Demand: 9X11 + 6X21 ≥ 830 Demand Constraint: 9X12 + 6X22 ≥ 1100;
Peach Demand: 6X13 + 3X23 ≥ 520;
- Acceptable Costs: ((15410X1j + 5500X2j) - 3392220) ≤ 3392220(Q/3)
-Acceptable Wastes: (3.1X11 + 2.2X12 + 2.5X21 + 3.2X22 + 3.5X13 + 1.8X23) - 899.4) ≤ 899.4(Q)

Case 2: Yogurt YOYO

GAMS code
 5

Results Analysis

Taking the two objective functions as constraints

under the criteria accepted by the company as a goal
and using the MINIMAX method proposed by
(Ramírez, Arias, & Ledesma, 2015) it is obtained that for
minimize waste by approximately 899.4
With costs of $3,392,220, 92 hours must be used.
extra in the packing line 1 for strawberry yogurt, 36 for
blackberry yogurt and 0 for peach; while in the
Packing line 2 requires 1 additional hour to
strawberry, 130 for blackberry and 174 for peach; complying
with the proposed goals with a deviation of 0.2304;
23.04% slightly surpassing the solver presented in the
text from (Ramírez, Arias, & Ledesma, 2015) when adding
in the code an optimization of 0.000000000000001.

This demonstrates that there are different alternatives for

address a multi-objective problem.

-Practical case 2 modeled in G.A.M.S. software summarized and adapted from:

(Ramirez, Arias, & Ledesma, 2015)
 6

Multi-objective linear programming practical activities (Case 3)

Problem: According to the practical case presented in (URIBE); Suppose a bank that has $20 million
capital, $150 million in deposits available in checking accounts (cc) and $80 million in certificates of
fixed-term deposits (CD).

The following table shows the categories among which the bank can make investments, along with
relevant information

The bank wants to maximize its utility by using the rates of return.

Se tienen las siguientes restricciones: [Link] invertirse todo el dinero disponible [Link] reservas de efectivo
They must be at least 14% of cc plus 4% of cd 3. The fraction of liquid investments must be at least
47% of cc plus 36% of cd. At least 5% of the funds must be invested in each category.
30% of the funds must be invested in commercial credits

Primal Model

Objective Maximize: Max Z = (0.040x2 + 0.045x3 + 0.055x4 + 0.070x5 + 0.105x6 + 0.085x7 +

0.092x8) + ((-0.005*x2 - 0.04*x3 - 0.05*x4
- (0.075*x5 - 0.1*x6 - 0.1*x7 - 0.1*x8 )/20) + ((-x6-x7-x8)/20)
S.A.R

Investment restriction: x1 + x2 + x3 + x4 + x5 + x6 + x7 + x8 ≤ 20 + 150 + 80

Cash reserves: x1 ≥ 0.14 * (150) + 0.04 * (80)
Fraction of liquid investments: 1.00x1 + 0.995x2 + 0.960x3 + 0.900x4 +
0.850x5 ≥ 0.47*(150) + 0.36*(80)
Investment by category: xj ≥ 0.05 * (20 + 150 + 80)
Investment of commercial credits: x8 ≥ g0.30 * (20 + 150 + 80)
-No negativity: x1, x2, x3, x4, x5, x6, x7, x8 ≥ 0

Case 3: Investment Portfolio

 7

GAMS code

Results Analysis
Taking into account information provided by the
exercise, and when implementing the proposed model in GAMS, it
It can be analyzed that when validating the first restriction, it
meets the requirement which is to spend the total amount of money
available that the bank has which when added gives us a
a total of $250 million that includes capital, accounts
current accounts and fixed-term deposit certificates
refers, in addition to this when analyzing the variables that
they intervene in the exercise which in this case are the data
of return rates of each investment category
it is obtained that to maximize utility by minimizing the
solvency coefficient and illiquid assets to invest
24.2$ in cash, 12.5$ in short-term, 12.5$ in bonds of
5-year government, 12.5 in 10-year government bonds,
88.3$ in government bonds for more than 10 years, 12.5$ in
installment credits, $12.5 in mortgage credits and $75
in commercial credits thus complying with the
investment restriction on the same; it is worth noting
that the investment that represents more benefits in this
the case is that of government bonds for more than 10 years (x5)
which does not appear to be a risk asset and having
calculate the solvency ratio as well as that of assets
without liquidity, a profit of $11,315 is achieved

Practical case 3 modeled in the G.A.M.S. software summarized and adapted from: (URIBE)
 8

Multi-objective linear programming practical activities (Case 4)

Problem: Activity taken from: (Crissey, 1992)

The decision to be made using the three different operational research approaches is 'How to choose'
a balanced diet consisting of a group of six foods:
Milk meat, eggs, bread, lettuce salad and ~ juice range? "There is a limit on the amount of
cada alimento que puede ser incluido en la dieta, y la dieta debe cumplir con ciertos estándares mínimos
for the amount of calories, milligrams (mg) of iron, units of vitamin A, and grams (g) of proteins
In this diet problem, three
Separate objectives are considered:
Maximize carbohydrates in the diet.
2. Minimize cholesterol content
3. Minimize the cost
The following table shows the exemplified data of a diet program and the minimum standards that
must take into account Vitamin A, Iron, Calories, and Proteins.

Primal Model

Objective Maximize: Max z: (24*x1 + 27*x2 + 15*x4 + 1.1*x5 + 52*x6) +

-10*x1 - 20*x2 - 120*x3 - 0.22*x1 - 2.2*x2
0.8*x3 - 0.1*x4 - 0.05*x5 - 0.26*x6
Required Vitamin A restriction: 702*x1 + 107*x2 + 7080*x3 + 134*x5 + 1000*x6 ≥ 5000
Iron requirement constraint: 0.2*x1 + 10.1*x2 + 13.2*x3 + 0.75*x4 + 0.15*x5 + 1.2*x6 ≥ 12.5
Restriccion de Calorías:344*x1 + 1460*x2 + 1040*x3 + 75*x4 + 17.4*x5 + 240*x6≥2500
Protein Restriction: 18*x1 + 151*x2 + 78*x3 + 2.5*x4 + 0.2*x5 + 4*x6 ≥ 63
 9

Case 4: Balanced Diet

GAMS Code

Results Analysis

Taking the objective functions proposed in the

exercise by the author is carried out the
programming of the proposed exercise in GAMS
obtaining the following results:

It is essential to include in a diet to maximize the

available carbohydrates, minimizing consumption of
cholesterol at an affordable price (minimizing the
price) and complying with the requirements of
Vitamin A, Iron, Calories and Protein: Milk (6 in
availability), Meat (1 available), do not include
huevos, Incluir pan (10 en disponibilidad), Ensalada
of lettuce (10 available) and finally Juice of
orange that provides (4 in availability); of this
a total of 453.940 carbohydrates is obtained with
the minimum cholesterol consumption and the minimum cost of
acquisition; the difference regarding the diet
proposal by (Crissey, 1992) is in the consumption of 1
1 unit of meat, 3 units of milk and 2 units of
lettuce; demonstrating that there are different ways
to model multi-objective problems according to the criterion
of the executor; in this case priority was given to the
maximization of carbohydrates in a diet for
adults with 453,940.
 10

-Activity 4 modeling in G.A.M.S. software summarized and adapted from:

(Crissey, 1992)

References

Crissey, M. G. (1992). MULTICRITERIA DECISION MAKING WITH SAS SOFTWARE.

San Francisco: USA FORCE.
Morales, S. S. (2014). Multi-objective Programming: Practical case applied to a
airline. Murcia: University of Murcia.
Ramírez, C. A., Arias, D. M., & Ledesma, J. D. (2015). Fifth Research Meeting
Formative Industrial Engineering Medellín (Minutes). Medellin: University
Pontifical Bolivarian.
URIBE, C. J. (n.d.). MULTI-OBJECTIVE LINEAR PROGRAMMING. Barranquilla:
University of the Coast.