Today there is a wide variety of tools that increase the probability of making

better decisions in any organization. It is worth mentioning that this range of

tools includes information systems, statistical methods, traditional industrial
engineering techniques, economic engineering, data processing, operations
research and others.

The activity called operations research developed during the Second World War,
but its origins can go back many years. Those who practiced it in wartime were
too busy with the problems that arose to devote themselves to a concise analysis
of its methodology or to write books that could help future generations. The
texts began to appear around the year 1950, which was when it was considered
that operations research was a subject worthy of inclusion in professional study
programs, such as in the fields of: economics, public administration, psychology,
social work, mathematics. , statistics and the numerous branches of engineering.

The objective of Operations Research is to provide a systematic and rational

method for the fundamental problems of system control, by making decisions,
which, in some way, produce the best results.

Definition
Operations Research is the work carried out by specialists in problem solving, in
order to produce solutions that best serve the objectives of any organization.

Organization
An organization can be interpreted as a system; every system has components
and interactions between them.
System
Every system is a structure that works. The form is the element that turns the
structure into a system. It can be concluded that every system is an information
system.

Operations research

1
Research is the application of scientific methodology through models, first to
represent the real problem that needs to be solved in a system and second to
solve it (objective).

Model formulation
It is the process of taking a real-world problem and describing it in mathematical
terms.

Solution techniques or algorithms

They are the mathematical techniques used to find answers from the models
created.

Computer solutions
It is the management of standardized computer programs to solve the models.

Philosophy
It is an overview of the relationships between the real world, models, managers
and solutions.

Operations research uses three types of problems: deterministic, with risks and
under uncertainty:

a) Deterministic problems are those in which each alternative to the problem

has one and only one solution.

b) Problems with risks are those in which each alternative of the problem has
several solutions.

c) Low uncertainty problems are those in which each alternative of the

problem has several solutions, however, it is unknown with what probability
or probabilistic distribution these solutions will occur.

Model construction
In operations research there are three classes of models: Iconic, Analog and
symbolic:

2
 a) Iconic models are scale images of the system whose problem needs to be
solved. For example, photographs, models, drawings and scale models of
ships, cars, airplanes, canals, etc.

b) Analog models are based on the representation of the properties of a

system whose problems are required to be solved using another system
whose properties are equivalent. For example, the properties of a
hydraulic system are equivalent to an electrical or even economic system.

c) Symbolic models are abstract conceptualizations of the real problem

based on the use of letters, numbers, variables and equations.

BASICS OF LINEAR PROGRAMMING

From a purely mathematical point of view, “Linear programming is a technique

used to maximize or minimize a function subject to linear constraints.” Now,
speaking in economic terms, "Linear programming is a technique used to
distribute a group of limited resources among a number of competing demanders,
who make decisions about the exchange of products", that is, the fundamental
role that What this mathematical tool plays within the economy is to economize
the available resources as much as possible, determining the most convenient
optimal program for its application in the unit (production center, company, etc.)
under study.

Basic concepts
The general problem of linear programming is represented by a series of linear
equations and inequalities formed by a finite number of variables.

Objective Function
It is the linear function that indicates the purpose pursued in the problem, in the
case of the economy, maximizing profits or minimizing costs, it is as follows: Y= a
+ bX.
Costs
It refers to the coefficients of the objective function variables (in the simplex
table we represent them with C i ), it implies the total monetary resources used
in the production of a given product.

3
Activity
It is a specific method that leads to the achievement of a specific goal. For
example, growing sesame, corn, beans, etc., on a farm, in such a way that the
desired production is obtained.

Alternatives
It involves the different forms of combination that can be done to achieve an
objective. For example, the forms of cultivation that must combine the three
previous products, in such a way that optimal utility is obtained.

Restrictions
It is the condition in which the scarce resources available to produce are
normally presented. If there are no restrictions there is practically no problem
(They can appear as equations or as inequalities).

Slack variable
It is the amount of resources not used in the program, whose main function from
a mathematical point of view is to break the inequality or inequality called
restriction, which occurs throughout the linear programming problem. The
coefficient, that is, the cost of this variable, will always be zero.

Artificial variable
It is an artifice to be able to work with the simplex table but it does not
intervene in the solution of the problem; This artificial variable when it is a
minimization problem, its value will be +M, where M is a value as large as desired.

optimal program
It is the basic feasible solution that leads us to obtain a greater utility or
benefit, that is, to maximize the profits or minimize the costs of a program.

Incoming variable
It is the variable or activity that enters or incorporates the solution to the
problem; it increases the function by a certain amount.
Outgoing variable
It is the activity that stops intervening in production because it is inefficient in
relation to other activities.

Minimize

4
It is the reduction of costs in the generation of a product in its optimal
expression.

Maximize
It is to raise the profits as much as possible in the solution of a linear
programming problem.

INTERSECTION OF LINES

Two non-parallel lines in the plane must intersect and the coordinates of the
point of intersection satisfy both equations since the point is on the two lines;
yes, for example, the equations of the lines are

2X - Y - 7 = 0 and 4X + 3Y - 9 = 0

then the solution of the system gives the values of X and Y that satisfy both
equations and these values must be the coordinates of the point of intersection.
A simple method to find the solution is elimination; We multiply the first
equation by three and adding it to the second, we obtain

6X - 3Y - 21 = 0
4X + 3Y - 9 = 0
10X - 30 = 0 : X = 3

Now, we replace X with 3 in either of the two initial equations and we get Y = -1.
The intersection point is (3, -1).

AND

3-

2-

5
 1-
x
-1 0 1 2 3 4

-1- (3, -1)

-2-

-3-

Two intersecting lines divide the plane into four regions. The following figure
shows how the lines:

2x – y = 4 and 3x + y = 11

They divide the plane into 4 regions that we will denote R 1 R 2 R 3 R 4. Each of the
regions is described by a pair of inequalities; that is, every point in R1 satisfies
simultaneously.

2x - y - 4 = 0 and 3x + y - 11 = 0

Similarly, the remaining regions satisfy the pairs of inequalities indicated in the
figure:

6
Considering the following pair of lines, answer the following questions:
a) What is the intersection?
b) Indicate the regions of the value domain.

Line 1: 2X - Y = 4 Line 2: 3X + Y = 11

Solution:

a) 5X = 15 X = 3 Y = 2 (3.2)

b) R 1 (3.4) R 2 (4.2) R 3 (3.0) R 4 (2.2)

2 < 4 6 > 4 6 > 4 2 < 4 Eq. 1
13 > 11 14 > 11 9 < 11 8 < 11 Eq. 2

7
(1) 2X – 3Y + 5 = 0 (2) X – 6Y + 7 = 0

R 1 (2.2) R 2 (-1.0) R 3 (-3.0) R 4 (-2.2)

1st. Eq. 3>0 3>0 -1 < 0 -5 < 0
2nd. Eq. –3 < 0 7>6 6>4 -7 < 0

MATHEMATICAL EXPRESSION OF LINEAR PROGRAMMING

The general problem of linear programming in mathematical form, the variable “X

i ” will be used in order to symbolize the activities involved in the problem. After

analyzing the concept of the general program of linear programming, it is

feasible to express it mathematically as follows: find X 1 , X 2 , X 3 ,...Xn. Whose
values optimize the linear function called objective function.
Max. Z x = U 1 X 1 + U 2 X 2 + U 3 X 3 + .......... + U n X n

Subject to conditions:

A 11 X 1 + A 12 X2+A 13 X 3 + .......... + A 1n Xn<B 1

8
A 21 X 1 + A 22 X 2 + A 23 X 3 + ........ + A 2n X n < B 2

. . . . .
. . . . .
. . . . .
A m1 X 1 + A m2 X 2 + A m3 X 3 + .....… + A mn

With X 1 , X 2 , X 3 , ........, X n > 0

Where A j , B j , U j are constants

It is also possible to express the general problem of linear programming in the

following way:

Max. Zx=UX
sa Ax < Bx

X>0

Both expressions have the same meaning, only the first is in algebraic form and
the second in matrix form.

STATEMENT OF LINEAR PROBLEMS

PROBLEM 1
A housewife interested in satisfying her husband's nutritional needs, using as
little expense as possible.
Decide to buy only milk, meat and eggs. When you arrive at the market you find
the following costs: a liter of milk costs $4.00, a kg of meat $40.00 and a dozen
eggs $4.00.
Consulting your diet guide you learn that the content of B vitamins in the above
amounts of food is equal to 20 mg; 10 mgs. And 10 mgs, respectively.
The vitamin C content is analogous: 10 mg, 20 mg, and 10 mg.
What quantities of each of the foods should the housewife buy if the minimum
daily requirements for an adult are: 60 mgs. of vitamin B and 50 mgs. of vitamin
C?

9
Problem Statement.
To reach the mathematical model more easily, it is convenient to prepare a table
presenting the interrelationships that exist between the elements that occur in
the problem.

VITAMINS LITER KILO DOZEN REQUIREMENTS

MILK MEAT MINIMUM EGGS

B 20 10 10 60
c 10 20 10 50
PRICES $4.00 $40.00 $4.00

Our decision variables will be:

X 1 = Number of liters of milk to buy
X 2 = Number of kgs. of meat to buy
X 3 = Number of dozens of eggs to buy .

It is known that the housewife wants to comply with the diet at the lowest
possible cost, that is, the $4.00 that a liter of milk costs (4X 1 ), plus the $40.00
that a kilogram of meat costs (40X 2 ), plus the $4.00 that a dozen eggs cost (4X
3 ),

They mean a minimum expense for her and at the same time she complies with
the diet. Then the objective function expressed mathematically will be:

Minimize G = 4X 1 + 40X 2 + 4X 3
This objective function will be subject to certain restrictions that correspond to
the minimum vitamin requirements that each of the purchased foods must have.
The first restriction indicates that the vitamin B content of the three products
must not be less than 60 mg, that is, the vitamin B content of the liter of milk
(20X 1 ) plus the vitamin B content of the kg. of meat (10X 2 ), plus the vitamin B
content of a dozen eggs (10X 3 ) purchased is greater than or equal to 60 mgs.
which is the minimum requirement for vitamin B to comply with the diet. In this
way our first restriction will be:

20X 1 + 10X 2 + 10X 3  60

10
Similarly, what is purchased of milk, meat and eggs should not be less than 50
mg. of vitamin C, which are required in the diet at a minimum. Then the second
constraint will be:

10X 1 + 20X 2 + 10X 3 > 50

Summarizing the model will be:

Minimize Z = 4X 1 + 40X 2 + 4X 3

sa 20X 1 + 10X 2 + 10X 3 > 60

10X 1 + 20X 2 + 10X 3 > 50

X1,X2,X3>0

PROBLEM 2
In an agricultural region, a modest farmer wants to know how much of each
product he should grow to obtain maximum profit under the following program.
The resources it has are: 12 hectares of land, 48 hours of labor and $360 of
capital.

Its real activities, that is, what it can grow given the nature of the soil, are:
corn, sorghum and oats.
The corresponding technical coefficients for growing corn are: 1 hectare of land,
6 man hours and $36 of capital, with which he expects to obtain a profit of $40.
The coefficients for sorghum similarly are: 1 hectare of land, 6 man hours and
$24 of capital, expecting to obtain $30 from the sale of the product. Finally, the
technical coefficients for growing oats are: 1 hectare of land, 2 man hours and
$18 of capital, with which you expect to obtain an income of $20 by selling the
product.
Model formulation:

Resources Corn Sorghum Oats Total availability

Land 1 ha. 1 ha. 1 ha. 12 hectares

I work 6 hrs 6 hrs. 2 hrs. 48 hrs.
Capital $36 $24 $18 $360
Profit $40 $30 $20 ----

11
Because what is desired is to know the number of has. that should be planted for
each crop, the decision variables are:

X 1 = Number of has. to plant corn

X 2 = Number of hectares to plant sorghum
X 3 = Number of has. to sow oats

Once the table is prepared and knowing that the farmer's desire is to maximize
his utility, the objective function will be: Maximize the $40 that he expects to
receive from the sale of corn (40X 1 ) plus the $30 that he expects to obtain
from the sale of sorghum ( 30X 2 ), plus the $20 he expects to get from the sale
of oats (20X 3 ); mathematically.

Maximize Z = 40X 1 + 30X 2 + 20X 3

sa X 1 + X 2 + X 3 < 12

6X 1 + 6X 2 + 2X 3 < 48

36X 1 + 24X 2 + 18X 3 < 360

X 1, X 2, X 3 > 0

PROBLEM 3
Project to control floods of a river.
The government of a country wants to build a system of dams to control the
floods of an extensive river basin.
For topographical reasons, technicians estimate that 4 dams should be built along
the river in different places; In addition, for the flood system and for it to be
effective, it is estimated that the minimum capacities of the dams should be:
5000, 10,000, 12,000 and 7,000 hectares/m 3 for dams P 1 , P 2 , P 3 , P 4 .

The government considers that the completion of the project will help the rapid
agricultural and industrial development of the region; To encourage such
development and at the same time minimize part of the project cost, it is

12
planned to distribute the capacity of the dams with the purpose of providing
water for irrigation and water to generate electricity.

Furthermore, taking into consideration the possibilities of economic development

of the different localities, it has been established that the total capacity of the
irrigation water project and the total capacity to generate electrical energy has
been distributed as follows:

Dam %For irrigation %For e. Electric

P 1 0 40
P 2 20 30
P 3 30 20
P4 50 10

The cost on ha. irrigation is $200 per hectare. of electric energy is $500. You
want to know how much you have. should be allocated for irrigation and how much
for electricity.

Minimize Z = 200X 1 + 500X 2

sa 0.4X 2 > 5000

0.2X 1 + 0.3X 2 > 10000

0.3X 1 + 0.2X 2 > 12000

0.5X 1 + 0.1X 2 > 7000

X1,X2>0

From Eq. 1 : (0.40X 2  5000 )(10) 4X 2 = 50000 X 2 = 12500

From Eq. 2: (0.20X 1 + 0.30X 2 > 10000)(10) 2X 1 + 3X 2 = 100000

X1=0X2=0
X 2 = 33.33 X 1 = 50000

From Eq. 3: (0.30X 1 + 0.20X 2  12000)(10) 3X 1 + 2X 2 > 120000

13
 x1=0 X2 = 0

X 2 = 60000 X 1 = 40000

From Eq. 4: (0.50X 1 + 0.10X 2 > 7000)(10) 5X 1 + X 2 = 70000

x1=0 X2 = 0

X 2 = 70000 X 1 = 1

X2
(thousands)

70

60
SOLUTION PLAN
50 FEASIBLE
40

30

20 1
10

X 1 (thousands)
10 20 30 40 50 60 70 80 90 100
4 3 2

Eq. 1 and 2 Eq. 1 and 3 Eq. 2 and 3

X 2 = 12500 X 2 = 12500 (2X 1 + 3X 2 = 100000)(3)
2X 1 + 3X 2 = 100000 3X 1 + 2X 2 = 120000 (3X 1 + 2X 2 = 120000)(2)
X 1 = 31250 X 1 = 31666.66 X 1 = 32000
X 2 = 12500 X 2 = 12500 X 2 = 12000
Z = 12500000 Z = 12583332 Z = 12400000

14
 SIMPLEX METHOD

BASIC CONCEPTS

Slack Variables
The simplex method requires that the constraints be equations rather than
inequalities. Any inequality can be converted into an equation by adding a non-
negative quantity to the lowest value side of the inequality.
Consider the following constraint, adding a non-negative variable, called the slack
variable.

6X 1 + 12X 2 + X 3 = 120 (to*)

In ( a * )

X 3 = amount of time not used in construction.

(Time used in construction) + (Time not used in construction) =120

(6X 1 + 12X 2 )+( X 3 ) =120

Similarly, let X 4 be the slack variable representing the number of hours not used
in paint for the given paint capacity constraint.

8X 1 + 4X 2 + X 4 = 64 (b*)

It is important to note that slack variables can never be negative.

BASIC VARIABLES AND BASIC FEASIBLE SOLUTIONS

The set of basic solutions in the new problem given in (a*) and (b*) cannot be
graphed since we now have four variables. Therefore, we must look for another
alternative, to find the feasible solutions (a*) and (b*) have two equations and
four variables. If we have more variables than equations, we can have an extra
set of variables equal to 0, thus obtaining a system with the same number of

15
variables and restrictions. Such a solution is called “basic solution” and
therefore, we are interested in finding basic solutions in such a way that all
variables have non-negative values. The variables that have positive values in the
basic feasible solution are called basic variables, the other variables that have
zero value are called non-basic variables. This is the basic feasible solution.

X 3 and X 4 are the basic variables.

X 1 and X 2 are the non-basic variables.

SUMMARY OF THE SIMPLEX METHOD

1. The simplex method finds an optimal solution (or a feasible basic optimal
solution).
2. The simplex method is a method of changing bases. A variable enters the
base, the incoming basic variable, and a variable leaves the base, the outgoing
basic variable.
3. The change method involves replacing a system of constraints – inequalities
with an equivalent system of constraints – equations.
4. In a system of constraints – equations, an equation can be replaced by an
equivalent equation by applying the following operations:
a) Replace an equation by itself, as many times as a non-zero constant.
b) Replace an equation by itself, added to as many times a non-zero
constant as another restriction equation.
5. The simplex method requires that the objective function be expressed in
such a way that each basic variable has a coefficient of 0.
6. The simplex method requires that each basic variable appear in one and only
one constraint equation.

Pivot column. It is the column of coefficients that are associated with the non-
basic variable that has been chosen to become the incoming basic variable.

Pivot row. It is the row of coefficients that contains the current basic variable
and that contains coefficient + 1, it is the one that has been chosen as the
outgoing basic variable.

16
Pivot number . It is the coefficient that is at the intersection between the
column and the pivot row.

Considering the problem of the Tennessee Furniture Factory (FMT).

Maximize Z = 200X 1 + 240X 2

sa 6X 1 + 12X 2 < 120 (1)

8X 1 + 4X 2 < 64 (2)
X1,X2>0

Solution:

Z - 200X 1 - 240X 2 = 0 (0)

6X 1 + 12X 2 + X 3 = 120 (1)
8X 1 + 4X 2 + X 4 = 64 (2)

Variable No. of Side

basic equation Z x1 x2 x3 x4 right
Z 0 1 -200 -240 0 0 0
x3 1 0 6 12 1 0 120
x4 2 0 8 4 0 1 64

Variable No. of Side

basic Equation Z x1 x2 x3 x4 right
Z 0 1 -80 0 20 0 2400
x2 1 0 ½ 1 1/12 0 10
x4 2 0 6 0 -1/3 1 24

Variable No. of Right side

basic equation Z x1 x2 x3 x4
Z 0 1 0 0 280/18 40/3 2720
x2 1 0 0 1 1/9 -1/12 8
x1 2 0 1 0 -1/18 1/6 4

Z = 2720
x1=4
X2 = 8

17
GRAPHIC CHECK

6X 1 + 12X 2 = 120 (1)

8X 1 + 4x 2 = 64 (2)

From Eq. (1) From Eq. (2)

X1=0X2=0 X1=0X2=0
X 2 = 10 X 1 = 20 x2 X2 = 16 x1=8

18

16

Solution: 14

12 2
x1=4
X2 = 8 10

8
Z = $2720 Flat
6 of
Solutions 1
x 1

6 8 10 12 14 16 18 20

Maximize Z = 12X 1 + 4X 2 (Notebook Page. 63-A)

sa X 1 + 2X 2 < 800
X 1 + 3X 2 < 600
2X 1 + 3X 2 < 2000
X 1, X 2 > 0

Variable No. of Z x1 x2 x3 x4 x5 Side

basic equation right
Z 0 1 -12 -4 0 0 0 0
x3 1 0 1 2 1 0 0 800

18
 x4 2 0 1 3 0 1 0 600
x5 3 0 2 3 0 0 1 2000

Variable No. of Z x1 x2 x3 x4 x5 Side

basic equation right
Z 0 1 0 32 0 12 0 7200
x3 1 0 0 -1 1 -1 0 200
x1 2 0 1 3 0 1 0 600
x5 3 0 0 -3 0 -2 1 800

x 1 = 600
X2 = 0

Z = 7200

Verification:

0 12 0 800 7200
1 -1 0 600 = 200
0 1 0 2000 600
0 -2 1 800

Max. Z = 2X 1 + X 2 + 3X 3 (Polytechnic Book Page. 91

sa X 1 + X 2 + X 3 < 6 notebook Page. 78-A)
2X 1 + 3X 2 + x 3  9
4X 1 + 2X 2 + x 3 = 10
X1,X2,X30

X1+X2+X3+X4=6 (1)

2X 1 + 3X 2 + X 3 + X 5 = 9 (2)

4X 1 + 2X 2 + X 3 + X* 6 = 10 (3)

Max Z = 2X 1 + X 2 + 3X 3 - MX* 6

Z - 2X 1 - X 2 - 3X 3 + MX* 6 = 0 (0)
-M(4X 1 + 2X 2 + X 3 + X* 6 = 10)

19
Z + X 1 (-2 –4M) + X 2 (-1 –2M) + X 3 (-3 –M) = -10M

Variable No. of Z x1 x2 x3 x4 x5 X* 6 Side

basic equation Right
Z 0 1 -2-4M -1-2M -3-M 0 0 0 -10M
x4 1 0 1 1 1 1 0 0 6
x5 2 0 2 3 1 0 1 0 9
X* 6 3 0 4 2 1 0 0 1 10

Variable No. of Z x1 x2 x3 x4 x5 X* 6 Side

basic equation Right
Z 0 1 0 0 -5/2 0 0 1/2+M 5
x4 1 0 0 ½ ¾ 1 0 -1/4 7/2
x5 2 0 0 2 ½ 0 1 -1/2 4
x1 3 0 1 1/2 1/4 0 0 1/4 5/2

Variable No. of Z x1 x2 x3 x4 x5 X* 6 Right side

basic equation
z 0 1 0 5/3 0 10/3 0 -1/3+M 50/3
x3 1 0 0 2/3 1 4/3 0 -1/3 14/3
x5 2 0 0 5/3 0 -2/3 1 -1/3 5/3
x1 3 0 1 1/3 0 -1/3 0 1/3 4/3

Z = 50/3
x 1 = 4/3
X2 = 0

X 3 = 14/3

Minimize Z = 3X 1 + 5X 2
sa 2X 1 – 5X 2 > 8 (1)
X1+ X2<5 (2)
X 1 - 3X 2 = 7 (3)
X 1, X 2 > 0

Of Inequality (1)
(2X 1 – 5X 2 > 8) (-1)

20
-2X 1 + 5X 2 < -8
(-2X 1 + 5X 2 + X 3 = -8) (-1)
2X 1 –5X 2 – X 3 + X* 4 = 8 (1)

Of inequality (2)
X1+X2<5
X1+X2+X5=5 (2)

Of inequality (3)
X 1 – 3X 2 = 7
X 1 – 3X 2 + X* 6 = 7 (3)

Of inequality (0)
Min Z = 3X 1 + 5X 2
(Min Z = 3X 1 + 5X 2 ) (-1)
Max -Z = -3X 1 - 5X 2 + 0X 3 + 0X 5 – MX* 4 – MX* 6

Max -Z + 3X 1 + 5X 2 + MX* 4 + MX* 6 = 0

Max -Z + 3X 1 + 5X 2 + MX* 4 + MX* 6 = 0
(2X 1 – 5X 2 – X 3 + X* 4 = 8) (-M)

-Z + (3 –2M)X 1 + (5 +5M)X 2 + MX 3 + MX* 6 = -8M

(X 1 – 3X 2 + X* 6 = 7) (-M)

-Z + (3 –3M)X 1 + (5 + 8M)X 2 + MX 3 = -15M (0)

Variable No. of Z x1 x2 x3 X* 4 x5 X* 6 Side

basic equation right
Z 0 -1 3-3M 5+8M M 0 0 0 -15
X* 4 1 0 2 -5 -1 1 0 0 8
x5 2 0 1 1 0 0 1 0 5
X* 6 3 0 1 -3 0 0 0 1 7

NOTE: This problem has no solution, it was checked with the Tora program.

21
Minimize Z = 0.4X 1 + 0.5X 2 Z = 4X 1 + 5X 2
sa 0.3X 1 + 0.1X 2 < 2.7 3X 1 + X 2 < 27
0.5X 1 + 0.5X 2 = 65X 1 + 5X 2 = 60
0.6X 1 + 0.4X 2 > 6 6X 1 + 4X 2 > 60
X1,X2>0 X1,X2>0
Of inequality (1)
3X 1 + X 2 + X 3 = 27 (1)

Of inequality (2)
5X 1 + 5X 2 + X* 4 = 60 (2)

Of inequality (3)
6X 1 + 4X 2 – X 5 + X* 6 = 60 (3)
(Min Z = 4X 1 + 5X 2 ) (-1)
Max -Z = - 4X 1 – 5X 2 – MX* 4 – MX* 6

-Z + 4X 1 + 5X 2 + MX* 4 + MX* 6 = 0
-M(5X 1 + 5X 2 + X* 4 = 60)

-Z + X 1 (4 – 5M) + X 2 (5 – 5M) + MX* 6 = - 60M

-M(6X 1 + 4X 2 – X 5 + X* 6 = 60)

-Z + X 1 (4 – 11M) + X 2 (5 – 9M) + MX 5 = -120M (0)

Variable No. of x2 x5 x3 X* 6 Side

basic equation Z x1 X* 4 right
Z 0 -1 4-11M 5–9M M 0 0 0 -120M
x3 1 0 3 1 0 1 0 0 27
X* 4 2 0 5 5 0 0 1 0 60
X* 6 3 0 6 4 -1 0 0 1 60

Variable No. of x5 Side

basic equation Z x1 x2 x3 X* 4 X* 6 right

Z 0 -1 0 11/3 – M -4/3 + 0 0 -36-21M

22
 16/3M 11/3M
x1 1 0 1 1/3 0 1/3 0 0 9
X* 4 2 0 0 10/3 0 -5/3 1 0 15
X* 6 3 0 0 2 -1 -2 0 1 6

Variable No. of Side

basic equation Z x1 x2 x5 x3 X* 4 X* 6 right
Z 0 -1 0 0 11/6 – 7/3 – 0 -11/6+ - 47 – 5M
10/6M 5/3M 16/6M
x1 1 0 1 0 1/6 2/3 0 -1/6 8

X* 4 2 0 0 0 10/6 5/3 1 -5/3 5

x2 3 0 0 1 -1/2 -1 0 1/2 3

Variable No. of Side

basic equation Z x1 x2 x5 x3 X* 4 X* 6 right
Z 0 -1 0 0 0 1/2 -11/10+M M 315/6
x1 1 0 1 0 0 1/2 -1/10 0 15/2

x5 2 0 0 0 1 1 3/5 -1 3
x2 3 0 0 1 0 -1/2 3/10 0 9/2

X 1 = 15/2
X2 = 9/2
Z = 5.25
Min. Z = 32X 1 + 8X 2

sa X 1 > 200
2X 1 + X 2 > 300
X 1, X 2 > 0

From inequality (1):

X 1 – X 3 + X* 4 = 200 (1)

From inequality (2):

2X 1 + X 2 – X 5 + X* 6 = 300 (2)

23
(Min. Z = 32X 1 + 8X 2 ) (-1)
Max. –Z = -32X 1 – 8X 2 – X* 4 – X* 6

-Z + 32X 1 + 8X 2 + MX* 4 + MX* 6 = 0

-M(X 1 – X 3 + X* 4 = 200)

-Z + X 1 (32 – M) + 8X 2 + MX 3 + MX* 6 = -200M

-M(2X 1 + X 2 –X 5 + X* 6 = 300)

-Z + X 1 (32 – 3M) + X 2 (8 – M) + MX 3 + MX 5 = -500M (0)

Variable No. of Z x1 x2 x 3 x 5  x*  x* Side

basic equation 4 6 right
Z 0 -1 32-3M 8-M M M 0 0 -500M
 X* 4 1 0 1 0 -1 0 1 0 200
 X* 6 2 0 2 1 0 -1 0 1 300

Variable No. of Z x1 x2 x3 x5  x*  x* 6 Side

basic equation 4 right
Z 0 -1 0 - M 16 – 0 -16+3/2M -4800 – 50M
8+1/2 1/2M
M
 X* 4 1 0 0 -1/2 -1 1/2 1 -1/2 50
x1 2 0 1 1/2 0 -1/2 0 1/2 150

Variable No. of Z x1 x2 x3 x5  X* 4  X* Side

basic equation 6 right
Z 0 -1 0 8 32 0 -32+M M -6400
x5 1 0 0 -1 -2 1 2 -1 100
x1 2 0 1 0 -1 0 1 0 200

Verification:

-32+MM 200 -6400 + 500M

2 -1 300 = 100
1 0 200

24
 x 1 = 200
X2 = 0

Z = 6400

DUALITY

Given any set of data for a linear programming model, we can use the same data
to form a different linear programming model. The resulting problem will be
called Dual Origin. The Dual has theoretical, economic and computational
importance that we are going to analyze. First, let's see how exactly the dual
problem is formed.

Transformation rules.

To examine the theory of duality in a satisfactory way we have to drop the

restriction that the variables of a linear programming model be non-negative.
For example, consider the following problem.

Problem (E1)

Max 3X 1 + 4X 2 – 2X 3
sa 4X 1 –12X 2 + 3X 3  12
- 2X 1 + 3X 2 + X 3  6
- 5X 1 + X 2 – 6X 3  -40
3X 1 + 4X 2 – 2X 3  10
X 1  0, X 2  0, X 3 not restricted in sign.
We see in this problem the presence of all types of restrictions ( < , > and =). In
addition, we have discarded the requirement that all variables must be non-
negative. In this example, we have only required that X 1 be non-negative. The
variable X 2 has been required to be non-positive and X 3 lacks sign restriction.
(That is, the optimal value of X 3 may be positive, negative or zero). The dual of
this problem (El) is formulated by applying the following rules.

RULE 1.- The number of Variables of the dual problem is equal to the number of
restrictions of the original problem.
The number of constraints in the dual problem is equal to the number of
variables in the original problem.

25
RULE 2.- The coefficients of the objective function in the dual problem will be
the resource vector of the original problem. The objective function of the dual
problem will be:

Z = 12Y 1 + 6Y 2 – 40Y 3 + 10Y 4

RULE 3.- If the original problem is a maximization model, the dual will be a
minimization model. If the original problem is one of minimization, the dual will be
one of maximization.

Min Z = 12Y 1 + 6Y 2 – 40Y 3 + 10Y 4

RULE 4.- The coefficients of the first restriction function of the dual problems
are the coefficients of the first variable in the restrictions of the original
problem, and in an analogous way for the other restrictions.

sa 4Y 1 – 2Y 2 – 5Y 3 + 3Y 4
-12Y 1 + 3Y 2 + Y 3 + 4Y 4
3Y 1 + Y 2 – 6Y 3 – 2Y 4

RULE 5.- The right sides of the dual restrictions are the coefficients of the
objective function of the original problem.

Restriction function L.D.

4Y 1 – 2Y 2 – 5Y 3 + 3Y 4 3
-12Y 1 + 3Y 2 + Y 3 + 4Y 4 4
3Y 1 + Y 2 – 6Y 3 – 2Y 4 -2

RULE 6.- The meaning of the ith dual restriction is = if and only if the ith
variable of the original problem does not have a sign restriction.

3Y 1 + Y 2 – 6Y 3 – 2Y 4 = -2

RULE 7.- If the original problem is a maximization (minimization) model, then,

after applying rule 6, assign the dual constraints the same (opposite) meaning as
the corresponding variable of the original problem.

26
 Y 1 – 2Y 2 – 5Y 3 + 3Y 4 > 3
-12Y 1 + 3Y 2 + Y 3 + 4Y 4 < 4

RULE 8.- The ith variable of the dual problem will not have a sign restriction if
and only if the ith restriction of the original problem is an equality.

RULE 9.- If the original problem is a maximization problem

(minimization), then, after applying rule 8, assign the other dual variables the
opposite sign (the same sign) as the corresponding constraints in the original
problem.

Y 1  0 Y 2  0 Y 3  0 Y 4 not restricted in sign

Examples:

PRIMAL:
Min –2X 1 – X 2 – 4X 3 – 5X 4
sa X 1 + 3X 2 + 2X 3 + 5X 4  20
2X 1 + 16X 2 + X 3 + X 4  4
3X 1 – X 2 – 5X 3 + 10X 4  -10
X1,X2,X3,X40

DUAL:
Max 20y1 + 4y2 – 10y3
sa y1 + 2y2 + 3y3  -2
3y1 + 16y2 – y3  -1
2y1 + y2 – 5y3  -4
5y1 + y2 + 10y3  -5
y1  0, y2  0, y3  0

PRIMAL:
Min. Z = X 1 + 12X 2 – 2X 3
sa 4X 1 + 2X 2 + 12X 2  10
2X 1 – X 2 + 11X 3  -2
X 1  0, X 2 not restricted in sign, X 3  0

27
DUAL:
Max Z = 10Y 1 – 2Y 2
sa 4Y 1 + 2Y 2  1
2Y 1 – Y 2 = 12
12Y 1 + 11Y 2  -2
Y 1  0, Y 2  0

ECONOMIC INTERPRETATION OF THE DUAL PROBLEM

For example, the primal problem often has the interpretation of finding utility-
maximizing levels of production subject to constraints on resource scarcity.
How would the dual of this problem be interpreted? Since the dual would be a
minimization problem, we would say that it is a cost minimization model. But
minimize the cost of doing what? The following overview will give the answer to
this question.

Suppose a firm owns two factories in two different business districts. For
simplicity, assume that each factory uses the same three scarce raw materials.
Factory 1 produces two products, such as grass mowers and spreaders, in
quantities X 1 and X 2 . Factory 2 produces three different products, door
knobs, refrigerator handles, and cattle bells, in quantities Z 1 , Z 2 , and Z 3 , and
factory 2 uses precisely the same raw materials as factory 1. The data of
factory 1 is given in the figure below.

Factory data 1
Input per grass Input by Availability
Raw material mower sprinkler total

1 6 4 38
2 1 3 34
3 10 7 44
Profitability 4 3
per unit

28
Factory production model 1

Max 4X 1 + 3X 2
sa 6X 1 + 4X 2 < 38
X 1 + 3X 2 < 34
10X 1 + 7X 2 < 44
X1,X2>0

Factory production model 2

Max 6Z 1 + 2Z 2 + Z 3
sa 4Z 1 + 2Z 2 + 7Z 3 < 54
3Z 1 + 9Z 2 + 8Z 3 < 126
6Z 1 + 5Z 2 + 2Z 3 < 33
Z1,Z2,Z3>0

Factory data 2
Raw material Input per Input per handle Input per Availability
knob refrigerator cowbell total

1 4 2 7 54
2 3 9 8 126
3 6 5 2 33
Profitability
per unit 6 2 1

WHAT IS A FAIR PRICE?

29
Let us now recall our assumptions that the same company owns both factories
and that they are located in different commercial districts. Let us also assume
that the three raw materials are scarce because long periods of time elapse for
their distribution. Now suppose that the manager obtains information that for
various economic reasons, prices (i.e., profits) in factory district 2 are going to
rise dramatically. It is not known exactly how much they will increase, but the
manager is confident that the increase will be so large that factory 2 should
take over all production. Thus, all raw materials stocks from factory 1 must be
transferred to factory 2. However, the administrator decides that factory 2
must pay a “fair price” to factory 1 for the transfers of those raw materials.
What will that price be?

Intuitively it seems clear that, at least from the point of view of factory 1, the
fair price would be the one that equals the maximum possible profit that it would
obtain if it retained the use of its resources, which are now going to be
transferred to factory 2. This is the VO from factory 1 and would be a fair
“package payment” for the three raw materials. The firm, however, needs to
know more than the package payment. You must have unit prices for each
material. These prices are needed for financial reporting (i.e. for tax).

All accounting for individual products is formulated on a per item basis. The firm
must find “fair” unit prices that can accept the scrutiny of careful review. To
obtain “fair” prices per unit we will make the following observations. If factory
2 pays per unit prices of Y 1 , Y 2 , and Y 3 for the raw materials, since factory 1
owns 38, 34 and 44 units, respectively, of each raw material.

Amount paid by factory 2 = 38y 1 + 34Y 2 + 44Y 3

Factory 2 wants to obtain the maximum possible profit, so its goal is

Min 38Y 1 + 34Y 2 + 44Y 3

However, Factory 1 wants to ensure that it would make as much profit as if it
remained in business on its own. In the data shown in the factory 1 data table, we
see that if factory 1 had 6 units of raw material 1, 1 unit of ka raw material 2,
and 10 units of ka raw material 3, it could produce one grass mower to obtain a
profit of $4. Remember that Y 1 is the selling price of commodity i. Thus,
factory 1 will insist that

30
 6Y 1 + Y 2 + 10Y 3 > 4

If this condition does not hold, factory 1 will prefer not to sell its raw materials
to factory 2. Using the raw materials to produce grass mowers will produce more
profits. Similarly, factory 1 will insist that

4Y 1 + 3Y 2 + 7Y 3 > 3

Otherwise, it is more profitable to manufacture sprinklers than to sell the raw

materials to factory 2. In summary, then, the problem of determining fair prices
consists of

Min 38Y 1 + 34Y 2 + 44Y 3

sa 6y 1 + y 2 + 10y 3 > 4
4Y 1 + 3Y 2 + 7Y 3 > 3
Y1,Y2>0

And this problem is the dual of (F1). With this we have shown that the dual of
the production problem has the following interpretation.

It provides “fair prices” in the sense that they produce the minimum
acceptable settlement payment.

COMPUTATIONAL INTERPRETATION OF THE DUAL PROBLEM

When the simple algorithm is used to solve a linear programming problem, it

turns out that the optimal solutions are obtained both for the original problem,
which can be maximization or minimization, and for its dual. We have already
seen this fact about the list of results by computer (subject to a possible change
of sign) and will be demonstrated mathematically in Chapter 6. So if you need to
solve a particular problem you can continue with the direct resolution of the
problem. Alternatively, you can take the dual problem from the original and solve
it on the computer. This will produce a solution of the dual of the dual, which is
the original problem. Since each of these possible routes leads to the same
result, it will be interesting, from a computational point of view, to inquire which
is the most efficient procedure.

31
To illustrate this point, note should be taken of the empirical fact that the time
required to solve a linear program depends critically more on the number of
constraints than on the number of variables. If the original problem has m
constraints and n variables, the dual problem will have n constraints and m
variables. It is then evident that, all else being equal, one should choose to solve
the problem that has the fewest restrictions.

Although the above rule of thumb is a reasonably good general prescription, when
dealing with fairly large and structured models it may fail, because in such cases
the other factors may not be equal. Possible reasons for departing from this rule
of thumb tend to be overly technical in nature. In some cases, regardless of the
number of restrictions, one of the two problems, due to its form, can be solved
by a special code, such as what we call network code, as opposed to the general
PL code. However, the other problem may not have the special structure required
and therefore may need to be solved with general code. Since special structure
codes tend to be more computationally efficient than general ones, this is an
important consideration.
Other technical considerations refer to the fact that, even with a general linear
programming code, it may be easier to “get started” with one problem than with
the other.

The choice between solving the original problem or its dual does not have much
importance for computing when it comes to small problems, say when both models
have a few hundred constraints, since such problems can be handled at high
speed by computing equipment. modern. When problems become complicated, in
the range of several thousand constraints, the choice between the original
problem and its dual becomes very important. In these cases, technical
consultation with a linear programming professional may well be valuable.

SENSITIVITY

Sensitivity analysis constitutes a very essential part in almost all linear

programming studies. Since some or all of the parameter values used in the
original model are only estimates of future conditions, it is necessary to
investigate the effect that would be had on the optimal solution if other

32
conditions prevailed. Furthermore, certain values of these parameters (such as
the amount of resources) may represent management decisions, in which case
their choice should be the most important point of the investigation and, of
course, can be studied through the analysis of sensitivity.

Sensitivity analysis is based on the fact that all the data except one part of the
problem is kept fixed and that we ask for information about the effect of the
change on this part of the data that is allowed to vary. The information we
might be interested in includes: (1) the effect on VO (i.e., the maximum possible
utility) and (2) the effect on the optimal policy E*, F*

PROBLEM

PROTRAC, Inc, produces two lines of heavy equipment. One of these product
lines (called debris removal equipment) is essentially for construction
applications. The other line (called forestry equipment) is intended for the
logging industry. The largest member of the debris removal equipment line (the
E-9) and the largest member of the forestry equipment line (the F-9) are
produced in the same department and with the same equipment. Using economic
predictions for the next month, PROTRAC's marketing manager judges that
during that period it will be possible to sell all the E-9s and F-9s the company can
produce. Management must now recommend a production target for next month.
That is, how many E-9s and F-9s should be produced?

When making the decision, the main factors to consider are the following:
PROTRAC will make a profit of $5,000 for each E-9 sold and $4,000 for each F-
9. Each product goes through mechanical operations in both department A and
department B.

For next month's production, these two departments have 150 and 160 hours
available, respectively. Each E-9 consumes 10 hours of mechanical operation in
department A and 20 hours in department B, while each F-9 consumes 15 hours in
department A and 10 hours in department B. These data are summarized in the
following figure.

33
 Hours
Department E-9 F-9 Total available

TO 10 15 150
b 20 10 160

In order to fulfill a commitment to the union, the total work hours that will be
dedicated to verifying the next month's finished products cannot be less than
10% of an established goal of 150 hours. This verification is carried out in a
third department that has no relation to the activities of departments A and B.
Each E-9 requires 30 hours of checking and each F-9, 10. Since 10% of 150 is 15,
the total work hours devoted to verification cannot be less than 135. These data
are concentrated in the figure.

Requirements in
1 E-9 1 F-9 total hours
Hours
for verification 30 10 135

Complete formulation of the PROTRAC, Inc. problem.

Max 5000E + 4000F (objective function)

sa 10E + 15F < 150 (hours in department A) (1)

F 20E + 10F < 160 (hours in department B) (2)

30E + 10F > 135 (checking hours) (3)

E - 3F < 0 (mixing restriction) (4)
11 E + F > 5 (total units required) (5)
E,F > 0 (non-negativity conditions)

34
 9
Feasible solutions plan
7 3
Objective function line
5 Flat
5
3 4
1
1 2
AND
1 3 5 7 9 11 13 15 17

SENSITIVITY OF COEFFICIENTS OF THE OBJECTIVE

FUNCTION

Let us consider increases in the coefficient of F in the objective function, that

is, in its profitability per unit, while the coefficient of E remains fixed. We have
seen in the figure that the contours of the objective function will tend to be
more horizontal (has a less negative slope) when this coefficient increases.
Referring to the previous figure we see that the optimal solution remains at the
vertex E * = 4.5, F * = 7.0 until the coefficient of F increases enough so that the
contours of the objective function are parallel to constraint (3). When this
occurs, there will be alternative optimal solutions, the current vertex (E* = 4.5,
F* = 7.0) and the vertex determined by the intersection of constraints (3 and 5).

If the coefficient of F continues to increase, the current solution (E* = 4.5, F* =

7.0) will no longer be optimal and the point determined by the intersection of
constraints (3 and 5) will be the only optimal one. The allowable increase for the
coefficient of F is thus determined by the increase in the coefficient that
makes the contours of the objective function parallel to constraint (3). We can
ask, When does this happen?

The contours of the objective function will be parallel to constraint (3) when
both lines have the same slope, which means that the coefficients must satisfy
the following equality.

Coefficient of E in (3) = coefficient of E in objective

35
 F coefficient in (3) = F coefficient in target

Which implies that

10 = 5000
15 F ratio on target

from which the coefficient of F in the objective is obtained = (5000)(15/10) =

7500

In the objective function, the current coefficient of F in the objective function

is 4000. This becomes parallel to (3), that is, alternative optima occur if this
value increases to 7500. Then the current optimal solution remains valid as long
as the increment of F is < 3500. This is called the allowable increase in the F
coefficient. It is the value shown in figure 5.6 under the “ALLOWABLE
INCREASE” heading for F. This combination of algebra and geometry explains
both the meaning and value of this data in the computational result.
Furthermore, by observing how changing coefficients affect the slope of the
objective function, we can make the following important generalizations.

EFFECT ON THE CONTOURS OF THE OBJECTIVE FUNCTION

Changing the coefficients of the objective function alters the slope of its
contours. This may or may not affect the optimal solution and the optimal value
of the objective function.

We now remember that when the coefficient of F increases (keeping the

coefficient of E fixed) we will finally obtain a new solution in which the optimal
value of F will have increased. This is intuitive, since an increase in the
profitability of F will not motivate us to produce F at a lower level. This
illustrates the general concept that

EFFECT ON ACTIVITY IN A MAXIMIZATION MODEL

36
A maximization model is the increase in the profitability of an activity, keeping
the other data unchanged, it cannot reduce the optimal level of that activity.

The situation for a cost minimization model is precisely the opposite. Since we
want to minimize the total cost, it will not be expected that when the costs of an
activity are increased, while keeping other data constant, a higher optimal level
of that activity can be reached. This illustrates the general concept that

EFFECTS ON ACTIVITY IN A MINIMIZATION MODEL

In a minimization model, the increase in the cost of an activity, keeping the

other data unchanged, cannot increase the optimal level of said activity.

Consider the following PL.

Maximize 20X 1 + 10X 2

sa X 1 + 2X 2 < 120
X 1 + X 2 < 90
x1 < 70
X 2 < 50
X 1 > 0; X2 > 0

Final table

Basic No. of x 2 x 5 x 6 Right side

variable equation Z x 1 x 3 x 4
Z 0 1 0 0 0 10 10 0 1600
x3 1 0 0 0 1 -2 1 0 10
x2 2 0 0 1 0 1 -1 0 20
x1 3 0 1 0 0 0 1 0 70

37
 x6 4 0 0 0 0 -1 1 1 30

Resolve the issue by making the following changes.

Changes b i : (b 1 , b 2 , b 3 , b 4 ) = (100, 70, 70, 40)
Changes C i : (C 1 , C 2 ) = (15, 20)

Verification:

0 10 10 0 120 1600
1 -2 1 0 90 10
0 1 -1 0 70 = 20
0 0 1 0 50 70
0 -1 1 1 30

0 10 10 0 100 1400
1 -2 1 0 70 30
0 1 -1 0 70 = 0
0 0 1 0 40 70
0 -1 1 1 40

Basic No. of x 2 x 5 x 6 Right side

variable equation Z x 1 x 3 x 4
Z 0 1 5 0 0 10 10 0 1400
x3 1 0 0 0 1 -2 1 0 30
x2 2 0 0 1 0 1 -1 0 0
x1 3 0 1 0 0 0 1 0 70
x6 4 0 0 0 0 -1 1 1 40

38
 0 10 10 0 1 20 20 – 15 = 5
1 -2 1 0 1 0
0 1 -1 0 1 = 0
0 0 1 0 0 1
0 -1 1 1 0

Basic No. of x 2 x 5 x 6 Right side

variable equation Z x 1 x 3 x 4
Z 0 1 0 0 0 10 5 0 1050
x3 1 0 0 0 1 -2 1 0 30
x2 2 0 0 1 0 1 -1 0 0
x1 3 0 1 0 0 0 1 0 70
x6 4 0 0 0 0 -1 1 1 40

0 10 10 0 2 10 10 – 20 = -10
1 -2 1 0 1 0
0 1 -1 0 0 = 1
0 0 1 0 1 0
0 -1 1 1 0

Basic No. of x 2 x 5 x 6 Right side

variable equation Z x 1 x 3 x 4
Z 0 1 0 -10 0 10 5 0 1050
x3 1 0 0 0 1 -2 1 0 30
x2 2 0 0 1 0 1 -1 0 0
x1 3 0 1 0 0 0 1 0 70
x6 4 0 0 0 0 -1 1 1 40

Basic No. of x 2 x 5 x 6 Right side

39
variable equation z x 1 x 3 x 4
Z 0 1 0 0 0 20 -5 0 1050
x3 1 0 0 0 1 -2 1 0 30
x2 2 0 0 1 0 1 -1 0 0
x1 3 0 1 0 0 0 1 0 70
x6 4 0 0 0 0 -1 1 1 40

Basic No. of x 2 x 5 x 6 Right side

variable equation Z x 1 x 3 x 4
Z 0 1 0 0 5 10 0 0 1200
x5 1 0 0 0 1 -2 1 0 30
x2 2 0 0 1 1 -1 0 0 30
x1 3 0 1 0 -1 2 0 0 40
x6 4 0 0 0 -1 1 0 1 10

For this problem, the solution is:

x 1 = 40
X2 = 30

Z = 1200

If we check all the changes, solving the model as an original one, the result would
be the same.

TRANSPORTATION PROBLEMS

Linear programming is the workhorse of quantitative models. The ability to

handle hundreds of constraints, thousands of decision variables, and the
incredible number of iterations that those numbers imply, make linear
programming (LP) a tool for a wide variety of problems.

In transportation problems, the manager must determine how to get products

from his various warehouses to his consumers in order to satisfy demand at a
minimum cost. This model is important because of its successful applications and
because it can be solved quickly and efficiently using special algorithms.

40
EXAMPLE 1
PROTRAC has four assembly plants in Europe. They are located in:
(1) Leipzig East Germany
(2) Nancy France
(3) Liège Belgium
(4) Tilburg Holland
The assembly machines used in these plants are produced in the US and they
embark to Europe, arriving at the ports of Amsterdam (A), Antwerp (B), Le
Havre (C).
Production plans for the third quarter (July-September) have already been
formulated. The requirements (Demand in destinations) for E-4 diesel engines
are as follows:

PLANT NUMBER OF ENGINES

(1) LEIPZIG 400

(2) NANCY 900

(3) LIEGE 200

(4) TILBURG 500

2000

The available quantity of E-4 machines at ports (the supply at origins) in time for
use in the third quarter is shown below:

PORT NUMBER OF ENGINES

(A) AMSTERDAM 500

(B) ANTWERP 700

(C) HAVRE 800

2000

41
PROTRAC must decide how many machines to send from each port to each plant.
Machines are shipped via common carriers and a charge per machine is payable.
The relevant costs are shown below.

Cost of transporting an engine from an origin to a destination.

FROM THE TO THE TARGET

ORIGIN 1 2 3 4

TO 12 13 4 6

b6 4 10 11

c 10 9 12 4

Amsterdam (A) . . Tilburg (4)

. Leipzig (1)
Antwerp (B) .

. Liege (3)

.
Le Havre (C)

. Nancy (2)

The decision variables are:

X ij = Number of engines shipped from port to plant j.

42
i = A, B, C
j = 1, 2, 3, 4

The objective. 12X A1 + 13X A2 + ........ + 4X C4

The problem has two types of restrictions:

1.- The number of items shipped from a port cannot exceed the number
available. For example,

X A1 + X A2 +X A3 +X A4

It is the total number of engines shipped from A. Since there are only 500
motors available in A, the constraint is

X A1 + X A2 +X A3 +X A4 < 500

A similar restriction is required for each origin

2.- The demand of each plant must be satisfied. For example,

X A1 + X B1 +X C1

It is the total number of engines sent to plant 1. Since the demand

of plant 1 is 400 engines, the restriction will be

X A1 + X B1 +X C1 = 400

A similar restriction is required for each plant.

MIN 12X + 13X A2 + 4X A3 + 6X

A1 A4 + 6X B1 + 4X B2 + 10X B3 + 11X B4 + 10X C1 +
9X C2 + 12X C3 + 4X C4

Subject to 1) X A1 + X A2 + X A3 + X A4 < 500

2) X B1 + X B2 + X B3 + X B4 < 700
3) X C1 + X C2 + X C3 + X C4 < 800
4) X A1 + X B1 + X C1 = 400
5) X A2 + X B2 + X C2 = 900
6) X A3 + X B3 + X C3 = 200

43
 7) X A4 + X B4 +X C4 = 500

Objective function value = 12,000

SOLUTION OF THE TRANSPORTATION PROBLEM

NORTHWEST CORNER RULE

1. Start in the upper left corner (source A, destination 1) and assign as many
units to that cell as possible. That is, use as much supply from source A as
possible to satisfy the demand from destination 1. This means that the
allocated quantity will be the minimum between the supply at A and the
demand at 1.
2. Reduce the source's current available supply and the destination's current
unmet demand by the allocated amount.
3. Identify the first origin with available offer. This is either the current
origin or the one directly below.
4. Identify the first destination with unmet demand. This is either the
current destination or the one immediately to the right of it.
5. Assign, as in step 1, as many items as possible to the route associated with
the origin-destination combination identified in steps 3 and 4.
6. Return to step 2.

ORIGIN 1 2 3 4
OFFER

12 13 4 6

TO
400 100 500 100
0

6 4 10 11
b
700 700 0

10 9 12 4
800 700

c 100 200 500 500 0

DEMAND 400 900 200 500

0 0 0 0 44
Objective function Z = 12X A1 + 13X A2 + 4X B2 + 9X C2 + 12X C3 + 4X C4
Z = 12(400) + 13(100) + 4(700) + 9(100) + 12(200) + 4(500)
Z = 14200.00

VOGEL APPROXIMATION METHOD

1. For each row with an available supply and each column with an unsatisfied
demand, calculate a penalty cost by subtracting the lowest data from the
next in value.
2. Identify the row or column that has the highest penalty cost.
(Ties are resolved arbitrarily.)
3. Allocate the maximum possible amount to the available route that has the
lowest cost in the row or column chosen in step 2.
4. Reduce the appropriate supply and demand by the amount assigned in step
three.
5. Discard any rows with zero available supply and columns with zero unmet
demand for further consideration.
6. Return to step one.

Penalties of rows and columns.

Minimum of the Second minimum of

row A row A
Penalties
Origin 1 2 3 4 0offer of line

12 13 4 6 500 2
TO

6 4 10 11 700 2
b

10 9 12 4 800 5
c

Demand 400 900 200 500

45
 Penalties 4 5 6 2
column
Maximum penalty Calculated as 6-4

Origin 1 2 3 4 0offer

12 13 4 6
TO 200 300 500 300
0

6 4 10 11
b 700 700
0

10 9 12 4
c 800 600
400 200 200 0

Demand 400 900 200 200 500 200

0 0 0 0

Total cost
according to the Vogel approximation method.

Amount of Cost ($)

Route engines by engine Cost ($)
At 3 200 4 800
At 4 300 6 1800
B2 700 4 2800
C1 400 10 4000
C2 200 9 1800
C4 200 4 800
Total cost $12,000

PROBLEM 1
A company builds a master plant for the production of an item in a period of four
months. The demands in the four months are 100, 200, 180 and 300 units
respectively. The variable production cost per unit in any month is four pesos. A
unit produced for later consumption will incur a storage cost at the rate of 50
cents per unit per month. On the other hand, items ordered in previous months
incur a penalty cost of 2 pesos per unit per month.

46
The production capacity to produce the product varies each month. The
calculations for the next four months are 50, 180, 280 and 270 units
respectively.
The objective is to formulate the production inventory plan at minimum cost.

Destinations
Origin 1 2 3 4 Ability
1 4 4.5 5 5.5 50
2 6 4 4.5 5 180
3 8 6 4 4.5 280
4 10 8 6 4 270
Demand 100 200 180 300

Total cost according to the Vogel approximation method.

Origin 1 2 3 4 Ability
4 4.5 5 5.5 .5
TO 50
50 0

6 4 4.5 5 .5 .5 .5
180 130
b 50 130 0

8 6 4 4.5 .5 .5 .5 .5 .5
280 210
c 70 180 30 30 0

10 8 6 4 2 2 2 2 2
270
d 270 0

Demand 100 50 200 70 180 300

0 0 0 0

2 .5 .5 .5
2 2 .5 .5
2 ,5 .5
2 2 .5
2 .5

47
Total cost according to the Vogel approximation method.

Amount of Cost
Route Units per unit Cost ($)
At 1 50 4 200
B1 50 6 300
B2 130 4 520
C2 70 6 420
C3 180 4 720
C4 30 4.5 135
D4 270 4 1080
Total cost $3,375

PROJECT MANAGEMENT (PERT AND CPM)

Modern projects start from the construction of a suburban shopping center to

putting a man on the moon; They are impressively large, complex and expensive.
Completing such projects on time, and within budget, is no easy task. In
particular, we will see that complicated problems arise when scheduling such
projects, often said to be structured through the interdependence of activities.
Normally, some of the activities cannot be completed before others have been
completed.

Some of the questions that will be answered with these methods are the
following:

1. What is the project completion date?

2. What is the probable variability of that data?
3. What are the scheduled start and end dates for each specific activity?
4. Which activities are critical in the sense that they must finish exactly as
scheduled to reach the goal of completion of the total project?
5. How long can non-critical activities be delayed before causing a delay in
the completion date of the total project?

48
 6. How can I effectively concentrate resources and activities to accelerate
project completion?
7. What controls must be exercised on the flow of financial resources for
the various activities during the project, in order to adhere to the overall
project?

PERT : Program Evaluation Review Technique.

CPM : Critical Path Method.

PERT

It was developed in the late 1950s by the Navy Special Projects Office. In
collaboration with the management consulting firm of Booz, Allen and Hamilton.
The technique received considerable publicity favorable to its use in the Polaris
missile development and engineering program, a complicated project that had
250 prime and 9,000 sub contractors. contractors. Since that date, it has been
widely received in other branches of government and industry and has been
applied in projects as different as the construction of factories, buildings and
roads, management research, product development, installation of new computer
systems, etc. .

CPM

It was developed in 1957, by JE Kelly, by Remington Rand, and M.R. Walker, from
Dupont. It differs from PERT in principle because of the details of how time and
cost are managed. In reality, the differences between PERT and CPM in
effective implementation have been blurred as companies have integrated the
best features of both systems into their own efforts to effectively manage
their projects.

CPM is generally used. in projects, where there has been more experience and
the times for each activity are already quite defined.

49
The PERT method is used in those projects where the duration of each activity
is not known with certainty and therefore an estimate must be made with a
certain level of security.

CRITICAL PATH METHOD (CPM)

It is necessary to define some terms before developing a network.

Activity. It is the individual part of work that must be carried out in a project.
It is a unique job with a specific duration. The activity is represented by an
arrow.

Event. It is the starting point of an activity and occurs only when all the
activities that precede it have come to an end.

Grid. It is the set of activities and events that faithfully reflect the project.

Virtual activity. It is an activity that lasts a time equal to zero. Since each
activity must be preceded by an event and must conclude with another event, it
is sometimes necessary to use a virtual activity to satisfy the above rule. This
activity is indicated by a dotted arrow.

Free time or slack. It is the time that exists between the end of one activity
and the beginning of the next.

Critical route. It is the sequence of activities and events where free time is
minimal. The critical path duration is the minimum time required to complete the
project.

50
EXERCISE 1

The activities involved in an evening choral service are given in the table below.
Build the network model and perform critical path calculations.

No. Activity Predecessor Duration

activity (days)
TO Music selection -- 21
b music learning TO 14
c Making copies and purchasing books TO 14
d Evidence B,C 3
AN essays d 70
D
F Individual trials d 70
g chandelier rental d 14
h Purchase of candles g 1
Yo Installation and decoration of chandeliers h 1
J. Purchase of decorative items d 1
K Installation of decorative items J. 1
l Request for stoles for the choir d 7
M Stole ironing l 7
N Sound system review d 7
EI Selection of music tracks N 14
TH
ER
Q Sound system installation EITHER 1
Q final essay E,FP 1

51
 R choir group Q,I,K 1
Yes Final program M,R 1

PREPARATION OF THE CRITICAL PATH

94 59 108 60
N 14 EITHER 1 Q
45 94 49 59 108 49 60 109

45 109
49 0
49 7 108 1 0 109 60
b 0 0 EQ 109 1
38 108 0
35 38 3 108 108 1 0 108 109 109
35 70
14 3 8
24 35 7 38 109 110
Or 0 21 TO d 39 108F Yo 110 54R
21 70 0
0 21 21 38 38 1 108 109 54 110 56 110 110
21 38 94 54 40 110
9 1
14 0 0 7 0 56 1 56
8
35 109
C 35 3 14 52 g 108 1 53 H 70
0
35 35 59 70 52 108 56 53 109
7

1 39 J 109 1 40 K 110

70 39 109 70 40 110
45

L 104 7 52 M 111 0 52 S 111 1

45 104 52 111 111 111
59 59 0

(A, B, D, E, Q, R, S) = 111 days

(A, B, D, F, Q, R, S) = 111 days
(A, C, D, E, Q, R, S) = 111 days
(A, C, D, F, Q, R, S) = 111 days

52
 BIBLIOGRAPHY

1. INTRODUCTION TO OPERATIONS RESEARCH

Frederick S. Hillier, Gerald J. Lieberman
McGraw Hill

2. OPERATIONS RESEARCH.
Hamdy A. Taha
Alpha Omega

3. OPERATIONS RESEARCH IN MANAGEMENT SCIENCE

Eppen, Gould and Schmidt
Prentice Hall

4. OPERATIONS RESEARCH
Moskowitz and Wright
Prentice-hall

53
54