Linear Programming I
Solution Methods
Linear Programming
Is a mathematical technique for finding the best uses of an
organizations resources.
LINEAR is used to describe a relationship between two or
more variables, a relationship which is directly and precisely
proportional
PROGRAMMING refers to the use of certain mathematical
techniques to get the best possible solution to a problem
involving limited resources
Four Major Characteristic
1. Objective
The firm must have an objective to achieve which can be
expressed as a function.
Total
selling price _ variable cost
sales volume
=
x
Contribution
per unit
per unit
in units
2. Alternative courses of action
- Should our firm allocate its manufacturing capacity to tables
& chair in the ratio of 50:50? 25:75? 70:30? And some other
ratio.
3. Resources must be in limited supply.
- Our furniture plant has limited number of machine hours
available per week; consequently, the more hrs it schedules
for tables, the fewer chairs it can make.
4. We must be able to express the firms objective and its
limitation as mathematical equations or inequalities, and
these must be linear equations or inequalities.
Profit per table
P = 8 (number of tables +
produced per week)
Profit per chair
6 ( number of chairs
produced per week)
EQUATION
Profit per table
P = 8 (number of tables +
produced per week)
Profit per chair
6 ( number of chairs
produced per week)
INEQUALITIES
Example :
The statement that the total cost of T tables (at a unit cost of $5 per
table) & C chairs (at a unit cost of $4 per chair) must not exceed
$120 is
5T + 4C 120
Graphic method to solve linear
programs
It is possible to solve linear programming
problems graphically as long as the number
of variables is no more than two.
Example:
Dimensions, Ltd., makes two products, tables and chairs,
which must be processed through assembly & finishing
departments. Assembly has 60 hrs available per week;
finishing can handle up to 48 hrs of work a week.
Manufacturing one table requires 4 hrs in assembly and 2
hrs in finishing. Each chair requires 2 hrs in assembly & 4
hrs in finishing.
If profit is $8 per table and $6 per chair the problem is to
determine the best possible combination of tables & chairs
to produce & sell in order to realize the maximum profit.
There are two limitations (also called constraints) in the
problem: the weekly time available in assembly & the
weekly time available in finishing.
FIRST STEP
Let us use T to represent the number of table & C to represent the
number of chairs
8T = total weekly profit from sale of tables
6C = total weekly profit from sale of chairs
Objective function = 8T + 6C
Department Time Constraints:
Assembly:
Finishing:
4T + 2C 60
2T + 4C 48
In order to obtain meaningful answers, the values calculated for T
and C cant be negative; thus all element of the solution to a linear
programming problem must be greater than or equal to 0. (T 0,C
0).
Maximize:
Profit = 8T + 6C
Subject to the constraints:
4T + 2C 60
2T + 4C 48
T0
C0
Mathematical Summary of the Problem
SECOND STEP
The inequality 4T + 2C 60 may be located on the graph by first
locating its two terminal points & joining these points by a straight
line. The two terminal points for inequality can be found in the ff:
manner.
1. if we assume that none of the time available in assembly is used in
making tables (the production of table is 0), then up to 30 chairs could be
made. Thus if we let T = 0 , then C 30. If we make the maximum
numbers of chairs, then C = 30. Our first point, thus, is (0, 30);
2. In order to find the second point, we assume that none of the time
available in assembly is used in making chairs (the production of chairs is
0), under this assumption we could produce up to 15 tables. Thus if we
let C = 0, then T 15. If we make the maximum numbers of tables, then
T = 15. Our second point, thus, is (15, 0);
4T + 2C 60
C=0
4T
60
___ = ___
4
4
T=0
2C
60
___ = ___
2
2
0 15
30 0
C
5 10 15 20 30
4T + 2C = 60
. .
5 10 15 20 30
Table of Values Sample
Assembly (4T + 2C 60)
1st point (0,30) this denotes the production of 0 tables & 30 chairs per week.
2nd point is (15,0) this denotes the production of 15 tables & 0 chairs per week .
C
Numbers of chairs
30 b (0, 30)
25
20
15
10
5
0
a0
c (15, 0)
5
10
15 20
25
Numbers of tables
30
Capacity constraint in
assembly department.
Finishing (2T + 4C 48)
1st point (0,12) this denotes the production of 0 tables & 12 chairs per week.
2nd point is (24,0) this denotes the production of 24 tables & 0 chairs per week .
C
Numbers of chairs
24
20
16
12
e (0,12)
8
4
0
a0
(24,0)
f
4
8
12
16
20
Numbers of tables
24
Capacity constraint in
finishing department.
Plot combination of Assembly and Finishing department.
Numbers of chairs
32
b (0,30)
28
24
20
Assembly department
16
12 e (0,12)
Finishing department
c (15,0)
4
12
16
20
f (24,0)
24 28
32
Numbers of tables
Graphic representation of
problem constraints.
The combinations of tables and chairs that fall within aedc is the feasible
region . Combinations outside aedc are called infeasible.
Example 1. For 10 tables and 5 chairs per week.
Assembly:
4T + 2C 60 hr available
4(10)+ 2(5) = 50 hr required
Finishing:
2T + 4C 48 hr available
2(10)+ 4(5) = 40 hr required
The time required to make 10 tables and 5 chairs per week falls within the
time available in both departments (see example figure 3) and so 10 tables
and 5 chairs is a feasible region solution.
Example 1.
2. For 8
3.
11
10tables
tablesand
and12
10
5 chairs
chairsper
perweek.
week.
Assembly:
4T + 2C 60
60hr
hravailable
available
4(11)+
4(8)+
4(10)+2(12)
2(10)
2(5) ==56
50
64hrhrrequired
required
[Infeasible]
[Feasible]
Numbers of chairs
Finishing:
2T + 4C 48
48hr
hravailable
available
2(11)+
2(8)+
2(10)+4(12)
4(10)
4(5) ==40
64
62hrhrrequired
required
C
32
28
24
20
16
12 e (8,12)
8
4
..
.
(10,5)
(11,10)
d
c
12
16
f
20
24 28
32
Numbers of tables
Example3
THIRD STEP
Locate point D, because once that point is known, all the points
defining the shaded area aedc will have been delineated precisely.
This is because we already have three points. a (0,0) , e (0,12) and c
(15,0)
How can d be located? One possibility is to read its location from an accurately
drawn graph below.
C
32
Numbers of chairs
b (0,30)
28
24
20
Assembly department
16
12 e (0,12)
Finishing department
c (15,0)
4
12
16
20
f (24,0)
24 28
Numbers of tables
32
Another method is to solve simultaneously the equation of the two
lines which intersect to from point d, the only point common to both
equations. The equations to be solved are:
4T + 2C = 60
2T + 4C = 48
First, multiply the first equation by -2:
-2 (4T + 2C=60) = -8T 4C = -120
+ 2T 4C =
48
Add the second equation.
-6T
= -72 (divide both side by -6)
T
=
12
And now substitute 12 for T in the second equation:
2T + 4C = 48
2(12) + 4C = 48
24 + 4C = 48 = 24 + 4C = 48
4C = 24
C=6
so 4C = (48-24)
(divide both side by 4)
So point d = (12,6)
Solving d algebraically
FOURTH STEP
Test the four corners of the shaded area to see which yields the greatest
weekly dollar profit;
Objective function/Profit = 8(T ) + 6(C)
Point
Point
Point
Point
a (0, 0) :
e (0,12) :
c (15,0) :
d (12,6) :
8
8
8
8
(0) + 6 (0)
(0) + 6 (12)
(15) + 6 (0)
(12) + 6 (6)
=0
= 72
= 120
= 132
The point which yields the greatest weekly profit is point d ($132).
The concept that the most profitable combination of tables and chairs is found at
point d (12,6) can be seen more clearly by first plotting the objective function 8T +
6C (given in the first step) directly on a graph of the feasible region.
To accomplish this, we first let profits equal some minimum dollar figure we know we can
attain without violating a constraint. In this case we have elected to let weekly profits equal
to $48, a profit easily attainable.
Objective function 48 = 8T + 6C
First locate two terminal points and join them with a straight line. (when T=0, C=8 and when
C=0, T=6)
This line represent all the possible
combinations of tables & chairs which
would yield a total profit of $48. You
might want to check one such
combination.
For example,
point X represents the
manufacture of 4 tables
and 2 chairs per week.
4(8) + 2 (6) = 48
Numbers of chairs
20
16
12
8
4
48
(X)
8T
d
+
6C
c
8
12
16
Numbers of tables
20
Suppose we now graph another isoprofit line representing all combinations of
tables and chairs which would produce a $96 weekly profit.
When T = 0, C = 16 and when C = 0, T = 12.
96 = 8T + 6C
Both profit equations (48 = 8T + 6C and 96 = 8T + 6C) are illustrated on the graph.
C
the isoprofit line which can be located
farthest from the origin (a) will contain
all the combination of tables & chairs
which will generate the greatest
possible profit.
20
16
12
6C
6C
8T
8T
48
96
Numbers of chairs
It is obvious, then, that
c
8
12
16
Numbers of tables
20
It is also true that one parallel isoprofit line will pass through point d. This
particular profit line (line 3), together with the first two profit lines. Although most of
the combinations of tables and chairs on profit line 3 do not fall within feasible
region (aedc), one point does, point d.
That point represent the most profitable combinations of products. Point d lies on
isoprofit line 3 and is still within the feasible region;
thus it represent the most profitable
combination of tables(12) and
chairs (6) for Dimensions, Ltd.,
to manufacture each week.
20
16
12
6C
6C
8T
8T
48
96
Numbers of chairs
c
8
12
16
20
Numbers of tables
Checking
Let say T=12, and C=6.
Assembly: 4T + 2C 60
4(12) + 2(6) 60
48
+ 12 = 60
Finishing: 2T
+ 4C 48
2(12) + 4(6) 48
24
+ 16 = 48
18
8T
6C
48
6C
18 = T
24
144 = 8T + 6C
144 = 8T +
8
8
12
8T
24 = C
16
20
96
Isoprofit:
144 = 8T + 6C
144 =
+ 6C
Numbers of chairs
c
8
12
16
20
Numbers of tables
Significance of an Integer Solution
( Chapter 13)
Graphic procedure for a
minimization problem
The objective in our Dimension, Ltd., problem was to maximize profits. However, we
can also consider linear programming problems in which the objective is to be
minimized.
Example: (Consider a modification of the Dimension, Ltd., problem)
Jeff Smith, the Director of Marketing at Dimensions, has promised customers the
firm will make at least 2 tables (T 2) and at least 4 chairs (C 4) per week.
Jeff has determined that it costs $20 per unit to manufacture a table & $8 per
unit to manufacture a chair. In order to see a simple example where the objective
function is to be minimized, lets suppose that Jeff wants to minimize the total
manufacturing cost per chairs and tables.
When we add these two new constraints to the problem, the feasible region is now the shaded
area ghid .
Objective Quantity:
Manufacturing Cost = 20T + 8C
Constraints:
T 2 C 4
Numbers of chairs
20
16
12
8
4
d
i
h
4
12
16
Numbers of tables
20
Example (in order to do this, we have plotted three isocost lines)
These 3 lines are;
20T + 8C = 288 (passing through point d)
20T + 8C = 180
20T + 8C = 72 (passing through point h)
In this case, point h represents the least costly combination of tables (2) and chairs (4) for
Dimensions to manufacture per week.
20
0
18
8
28
C=
C=
+8
+8
16
12
T
20
T
20
Numbers of chairs
8
4
d
i
Line 3 passes through point h
4
12
16
Numbers of tables
20
Infeasibility
Means there is no solution which satisfies all the constraints
The problem of Dimentions, Ltd., from earlier in this chapter with two additional
constraints: (1) the marketing manager must have at least 16 tables a week, and
the marketing manager must have at least 12 chairs a week.
C
32
Numbers of chairs
Assembly
constraint
28
Marketing constraint(1)
T 16
24
20
Marketing constraint(2)
C 12
16
12 e
8
Finishing
constraint
c
4
12
16
20
24 28
Numbers of tables
32
(2)
Unboundedness
If the objective can be made infinitely large without violating any of the constraints.
C
op
Is
e5
lin
fit
ro
e4
lin
fit
ro
12
e1
lin
fit
ro
16
Marketing
Constraint (1)
T 16
op
Is
20
op
Is
24
e2
lin
fit
ro
28
op
Is
32
Numbers of chairs
Marketing
Constraint (2)
C 12
op
Is
e3
lin
fit
ro
4
4
12
16
20
24 28
32
36
Numbers of tables
40
44 48
Redundancy
Additional Constraint:
he cannot sell more than 20 chairs a week.
C
32
Numbers of chairs
Assembly
constraint
28
24
Marketing constraint 20
20
16
12 e
8
Finishing
constraint
c
4
12
16
20
24 28
Numbers of tables
32
Alternative Optima
In this situation below is the example of Alternative Optima; here the isoprofit line
farthest from the origin coincides with one of the constraint lines, specially line ed
Numbers of chairs
12
e
Line 4 concludes with
Constraint line ed
Iso
pro
fit
Iso
p
Iso
rofi
t
lin
e
lin
e
pro
fit l
ine
1
c
4
12
Numbers of tables
16
Linear Programming II
The simplex method
Chapter Objectives
Introduce the simplex method of linear programming
Examine in detail all the steps of the simplex method
Define and discuss all the elements in the simplex
tableau in terms of their significance to the problem
statement and problem solution
Demonstrate the use of the simplex method on
minimization problems and how to handle all three
kinds of constraints
Show how the simplex method deals with infeasibility,
unboundedness, alternative optima and degeneracy
Linear Programming is a three-stage process
1.
2.
3.
Problem formulation: gathering the
relevant information, learning what
questions need to be answered, and setting
the problem up as a linear program.
Problem solution: finding the optimal
solution to this linear programming.
Solution interpretation and
implementation: checking that the
solution to the linear program is indeed a
solution to the original real problem
Recall that the variables in the problem were T and C,
the numbers of tables and chairs that Dimensions
Ltd., should manufacture per week.
The algebraic statement of the problem is:
Maximize:
Profit = 8T + 6C
Subject to:
Assembly:
4T + 2C 60
Finishing:
2T + 4C 48
All variables 0
Using slack variables to generate equations
S = slack variable (unused weekly time) in assembly
S = slack variable (unused weekly time) in finishing
S = 60hrs (assembly)
S = 48hrs (finishing)
We can express these two statements in mathematical form by
writing equations for slack variables S and S as follows:
Assembly =
Finishing =
S = 60 4T 2T
S = 48 2T 4T
example
Assume that in assembly we process 5 tables and 3 chairs
per week:
S = 60 4(5) 2(3)
= 34hr unused time in assembly
Assume that in assembly we process 4 tables and 6 chairs
per week
S = 48 2T 4T
= 16hr unused time in finishing
Constraint Equation
By adding a slack variable to each inequality constraint, we
convert them into these equations:
4T + 2C + S = 60
2T + 4C + S = 48
(final form)
Maximize:
Profit = 8T + 6C + 0S + 0S
Subject to:
4T + 2C + S + 0S = 60
2T + 4C + 0S + S = 48
All variables 0
Parts of the simplex tableau and their functions
Cj column (profits per unit)
Product-mix column
Constant column (quantities of product in the mix)
Variable columns
Cj
Product
mix
0
0
S
S
Quantity
60
48
8 6
T C
4 2
2 4
0
0
S S
1
0
0
1
Real
products
Slack
time
Cj row
Variable row
2 rows illustrating
Constraint equations
(coefficients only)
