Module – 3

Linear Programming

Dr. K.K. Ramachandran

Professor
Dept. of Mechanical Engineering
Gov. Engineering College Thrissur
Thrissur – 680 009
E-Mail: chandrankkr535@gmail.com
 Formulation of Linear Programming Problems
 A LPP has two parts – the objective function and the constraints
 The objective will be either to maximize or minimize the objective
function
 In the formulation LPP, the following elements are used.
 Decision variables (e.g. for product mix problem – No. of product A,
B, etc.)
 Coefficients in the objective function (e.g. cost per unit or profit per
unit, etc.)
 Technological coefficients (e.g. number of hours of machining per
unit, etc.)
 Resource availability (e.g. raw material availability, etc.)
 Non negative constraints (the variable should be non negative)

2 9/14/2021
KKR, GEC Tcr
 Generalized mathematical Expression of LPP
Maximize or Minimize Z(x) = F(x) = c1x1+c2x2+…+cnxn
Subject to linear constraints;
a11x1 + a12x2 + ................+ a1nxn = b2
a21x1 + a22x2 + ................+ a2nxn ≤ b2
...................................................
am1x1 + am2x2 + ................+ amnxn ≤ bm
And non-negativity constraints;
x1 ≥ 0, x2 ≥ 0, .................., xn ≥ 0
Where cj are cost coefficients, bi - capacities, Xj- decision variables
and aij are technological coefficients

3 KKR, GEC Tcr 9/14/2021

 Generalized mathematical Expression of LPP
 In matrix form;
Maximize or Minimize Z = C’X

Subject to constraints: AX = B and X ≥ 0

 c1   x1   a11 a12 ... a1n   b1 

c  x  a a ... a  b 
Where C =  2  , X =  2  , A =  21 22 2n 
and B =  2 
.  .   . . . .   . 
       
 n
c  n
x  m1
a a m2 ... a mn  b m 

4 KKR, GEC Tcr 9/14/2021

 Formulation of LPP…
E.g. A company manufactures two types of products; P and Q and both
uses milling and turning for its production. The process time required
for product P is 10 hours milling and 8 hours turning per unit whereas
for product Q it is 20 hours and 8 hours for milling and turning
respectively. The maximum hours available per week on milling machine
and lathe (for turning) are 120 hours and 80 hours respectively. The
profit per unit on sale of product P and Q is Rs.12 and Rs.16,
respectively. Formulate a LP model to find the production volume of P
and Q such the total profit is maximized.
Solution:
Data given: No. of products 2; Name of products P and Q
Operation time/unit: Product P – milling:10 hr & turning: 8 hr
Product Q – milling:20 hr & turning: 8 hr
5 KKR, GEC Tcr 9/14/2021
 Formulation…
Resource constraints:
Milling hours available – 120 hr /week
Turning hours available – 80 hr /week
Profit (objective function coefficients):
Product P – Rs.12/unit
Product Q – Rs.16/unit
Let x1 and x2 be the volume of production of P and Q
Thus, the profit z = 12 x1 + 16 x2 and the problem is to
maximize profit z
Technological and resource constraints:
Milling: 10x1 + 20x2 ≤ 120
Turning: 8x1 + 8x2 ≤ 80
6 KKR, GEC Tcr 9/14/2021
 Formulation…
Therefore the LPP model is;
Maximize z = 12 x1 + 16 x2
Subject to constraints;
10x1 + 20x2 ≤ 120
8x1 + 8x2 ≤ 80
x1, x2 ≥ 0

7 KKR, GEC Tcr 9/14/2021

 Solution of LPP
1. Graphical Method
E.g. Maximize z = 12 x1 + 16 x2
Subject to constraints;
10x1 + 20x2 ≤ 120
8x1 + 8x2 ≤ 80
x1, x2 ≥ 0
Solution:
Rewrite the constraints as equalities
i.e. 10x1 + 20x2 = 120 (1)
8x1 + 8x2 = 80 (2)
From equation(1), find value of x2 when x1 = 0 and value x1 when x2 = 0
x1 x2
0 6
8 KKR, GEC Tcr 9/14/2021
12 0
1. Graphical Method…
Similarly from equation (2), find value of x2 when x1 = 0 and
value x1 when x2 = 0 x1 x2
x2 0 10
12
10 0
10

Equation (2)
8

D Equation (1)
6

4
(8,2)
C
2 Feasible region - ABCD

0A
9 0 2 4 6 8 B 10 12 9/14/2021
KKR, GEC Tcr
x1 14
 1. Graphical Method…
It can be shown that the optimal z will be on any one of the corner
points of the feasible region.
Thus, evaluate z = 12 x1 + 16 x2 at the corner points A, B, C and D
Point A (0,0): z = 12 x 0 + 16 x 0 = 0
Point B (10,0): z = 12 x 10 + 16 x 0 = 120
Point C (8,2): z = 12 x 8 + 16 x 2 = 128
Point D (0,6): z = 12 x 0 + 16 x 6 = 96
The maximum value of z is at point C (8,2).
Therefore, the optimum value x1 = 8, x2 = 2 and the
corresponding profit z = Rs.128

10
KKR, GEC Tcr 9/14/2021
 1. Graphical Method…
E.g. 2 Find the maximum of z = 100 x1 + 60 x2
Subject to constraints;
5x1 + 10x2 ≤ 50
Use graphical method
8x1 + 2x2 ≥ 16
3x1 – 2x2 ≥ 6
E.g. 3 Find the minimum of z = 4x1 + 6x2
Subject to constraints;
x1 + x2 ≥ 8
6x1 + x2 ≥ 12
x1, x2 ≥ 0

11 KKR, GEC Tcr 9/14/2021

 Solution of LPP – Simplex Method
 Solve the given LPP using Simplex method.
Maximize z = 12 x1 + 16 x2
Subject to constraints;
10x1 + 20x2 ≤ 120
8x1 + 8x2 ≤ 80
x1, x2 ≥ 0
Solution:
Step 1: Convert the inequalities of constraints into equalities by
adding slack variables.
10x1 + 20x2 +s1 = 120 s1 and s2 are slack variables
8x1 + 8x2 +s2 = 80
12 KKR, GEC Tcr 9/14/2021
 Simplex Method…
Step 2: Modify the objective function by adding the slack variables
and hence the problem is:
Max z = 12 x1 + 16 x2 +0 s1 + 0 s2
Subject to: 10x1 + 20x2 +s1 = 120 (1)
8x1 + 8x2 +s2 = 80 (2)
x1, x2 , s1, s2 ≥ 0
Step 3: Create the initial Simplex table
Cj 12 16 0 0 Cj - coefficients of variables in the
CBi Basic Solution Ratio objective function
x1 x2 s1 s2
Variable CBi - Coefficient of basic variable in the
0 s1 10 20 1 0 120 objective
0 s2 8 8 0 1 80 2
Z j =  (CBi a ij )
Zj 0 0 0 0 0 i =1
Cj - Zj 12 16 0 0
Z1 = (0 x 10)+((0 x 8) = 0
9/14/2021
13
KKR, GEC Tcr
 Simplex Method…
Step 4: Now check for optimality
 Condition for optimality;
1. For maximization problems: all (Cj – Zj) ≤ 0
2. For minimization problems: all (Cj – Zj) ≥ 0
• Since, in the present case, all the (Cj – Zj) values are greater than or
equal to zero, the solution is not optimal (problem is maximization)
• Therefore, we need to proceed further to reach the optimal solution
• Select the maximum value in the (Cj – Zj) row and mark the column
containing that value as key column
• Find the ratio of solution to key column values
• Identify the least value of ratio and mark the row containing the least
ratio as key row
• Mark the intersecting element of the key row and key column as key
element 9/14/2021
14
KKR, GEC Tcr
 Simplex Method…
Step 5: Initial Simplex table
Cj 12 16 0 0
CBi Basic Solution Ratio Key element
x1 x2 s1 s2
Key row Variable
Lowest ratio
0 s1 10 20 1 0 120 120/20= 6
0 s2 8 8 0 1 80 80/8= 10
Zj 0 0 0 0 0
Cj - Zj 12 16 0 0

Key column
Highest

# Key element = 20
# x2 is the entering variable – variable in the key column
# s2 is the leaving variable – variable in the key row
15 KKR, GEC Tcr 9/14/2021
 Simplex Method…
Step 6 : Iteration 1
Cj 12 16 0 0
CBi Basic Solution Ratio
x1 x2 s1 s2
Variable

16 x2 10/20 20/20 1/20 0/20 120/20 Divide the coefficients

by the key element
0 s2 4 0 -2/5 1 32

Zj 8 16 4/5 0 96
Cj - Zj 4 0 -4/5 0

Values in the second row

Corresponding key column value  corresponding key row value
= Old value -
Key element
E.g. for first cell : Value = 8 – (8 x 10)/20 = 8 – 4 = 4

16 KKR, GEC Tcr 9/14/2021

 Simplex Method…
 Now, check for optimality. Since all the (Cj – Zj) values are not less
than or equal to zero, optimal condition is not reached.
 So, find the largest value in the (Cj – Zj) row and mark the key
column, key row and key element as before.
Iteration 1
Cj 12 16 0 0
CBi Basic Solution Ratio
x1 x2 s1 s2
Variable Key element = 4

16 x2 1/2 1 1/20 0 6 12
Outgoing variable – s2
0 s2 4 0 -2/5 1 32 8
Entering variable – x1
Zj 8 16 4/5 0 96
Cj - Zj 4 0 -4/5 0

17 KKR, GEC Tcr 9/14/2021

 Simplex Method…
Step 7 : Iteration 2
Cj 12 16 0 0
CBi Basic Solution Ratio
x1 x2 s1 s2
Variable

16 x2 0 1 1/10 -1/8 2

12 x1 1 0 -1/10 1/4 32/4 = 8

Zj 12 16 2/5 1 128
Cj - Zj 0 0 -2/5 -1

Now check for optimality: Since all the (Cj – Zj) values are less
than or equal to zero, the solution is optimal.
The optimal values of x1 = 8, x2 = 2 and Profit Z = 128 (ANS)

18 KKR, GEC Tcr 9/14/2021

 Simplex Method – Minimization Problem
 Solve the LPP given below using Simplex method
Minimize Z = 2x1 - 3x2 + 6x3
Subject to: 3x1 – x2 +2x3 ≤ 7
2x1 + 4x2 ≥ -12
-4x1 + 3x2 + 8x3 ≤ 10
x1, x2, x3 ≥ 0
Solution:
Note: For Simplex method to apply, the inequalities of all the
constraints must be ‘≤’
• In the given problem, inequality of the second constraint is ≥.
Therefore, it should be converted to ≤ by multiplying both sides
by (-1)

19 KKR, GEC Tcr 9/14/2021

 Simplex Method – Minimization Problem
i.e. 2x1 + 4x2 ≥ -12 → -2x1 - 4x2 ≤ 12
Modify the objective function and constraints by adding the slack
variables:
Minimize Z = 2x1 - 3x2 + 6x3 +0 s1 +0 s2+0 s3
Subject to: 3x1 – x2 +2x3 +s1 = 7 (1)
-2x1 - 4x2 + s2 = 12 (2)
-4x1 + 3x2 + 8x3 +s3 = 10 (3)
x1, x2, x3, s1, s2, s3 ≥ 0

20 KKR, GEC Tcr 9/14/2021

 Simplex Method – Minimization Problem
Initial Simplex Table
Cj 2 -3 6 0 0 0
CBi Basic Solution Ratio
x1 x2 x3 s1 s2 s3
Variable

0 S1 3 -1 2 1 0 0 7 -7

0 S2 -2 -4 0 0 1 0 12 -3

0 s3 -4 3 8 0 0 1 10 10/3

Zj 0 0 0 0 0 0 0
Cj - Zj 2 -3 6 0 0 0

▪ Check for optimality: Since all the (Cj-Zj) values are not greater than
or equal to zero, the solution is not optimal (minimization problem)

21 KKR, GEC Tcr 9/14/2021

 Simplex Method – Minimization Problem
 Identify the most negative element in the (Cj – Zj) row and
mark the column containing that element as key column.
Note: If there exist more than one most negative elements in the
(Cj-Zj) row, select any one arbitrarily.
 Find the ratios and identify the lowest positive ratio and mark
the corresponding row as key row.
Note: (i) If all the ratios are negative or infinity, the problem is
unbounded and we cannot proceed further
(ii) For minimization problems, also, the key row is the row
corresponding to the lowest positive ratio
 The leaving variable is s3 and the entering variable is x2

22 KKR, GEC Tcr 9/14/2021

 Simplex Method – Minimization Problem
Iteration 1
Cj 2 -3 6 0 0 0
CBi Basic Solution Ratio
Variabl x1 x2 x3 s1 s2 s3
e
0 S1 5/3 0 14/3 1 0 1/3 31/3 31/5

0 S2 -22/3 0 32/3 0 1 4/3 76/3 -38/11

-3 x2 -4/3 1 8/3 0 0 1/3 10/3 -5/2

Zj 4 -3 -8 0 0 -1 -10
Cj - Zj -2 0 14 0 0 1

▪ Since all the Cj-Zj values are not greater than or equal to zero, the
solution is not optimal
▪ Entering variable – x1 and leaving variable s1
23
KKR, GEC Tcr 9/14/2021
 Simplex Method – Minimization Problem
Iteration 2
Cj 2 -3 6 0 0 0
CBi Basic Solution Ratio
Variabl x1 x2 x3 s1 s2 s3
e
2 x1 1 0 14/5 3/5 0 1/5 31/5

0 S2 0 0 156/5 22/5 1 14/5 354/5

-3 x2 0 1 32/5 4/5 0 3/5 58/5

Zj 2 -3 -68/5 -6/5 0 -7/5 -112/5

Cj - Zj 0 0 98/5 6/5 0 7/5

Since all the Cj-Zj values are greater than or equal to zero, the
solution is optimal
Optimal value of x1 = 31/5, x2=58/5, x3=0 & Z=-112/5
[Check the optimal value of Z by substituting the values of variables in the objective function]
24
KKR, GEC Tcr 9/14/2021
 Special Cases in LPP
1. Infeasible Solution (Infeasibility)
 Infeasible solution happens when the constraints have contradictory
nature.
 Therefore, it is not possible to find a solution which can satisfy all
constraints.
E.g. Max. Z = 5 x1 + 8 x2
Subject to constraints
4 x1 + 8 x2 ≥ 40
4 x1 + 6 x2 ≤ 24
4 x1 + 8 x2 ≥ 40
x1, x2 ≥ 0
4 x1 + 6 x2 ≤ 24
In Simplex method, after a particular iteration, all the Cj-Zj values are negative or zero for a
max. problem, but, the solution are negative or infinite is indication of infeasibility
25 KKR, GEC Tcr 9/14/2021
 Special Cass in LPP…

2. Unbounded Solution (Unboundedness)

 A problem which has infinity solutions is called unbounded solution.
 A maximization problem with all greater than or equal to constraints
 A minimization problem with all lesser than or equal to constraints

 In Simplex method, the existence

of a non positive values in the
Ratio column (even after a key
column or pivotal column found
to exists) is indication of
Unboundedness

26 KKR, GEC Tcr 9/14/2021

 Special Cass in LPP…
3. Redundant Constraint (Redundancy)
 A constraint is called redundant, when it does not affect the
solution.
 If a constraint is redundant, the feasible region does not depend
on that constraint.
 Therefore, even if we remove the constraint from the solution,
the optimal solution is not affected.
E.g. Max. Z = 5 X1 + 8 X2
Subject to Constraints x1 ≤ 16
3 x1 + 2 x2 ≤ 24
x1 + 3 x2 ≤ 12
x1 ≤ 16
x1, x2 ≥ 0
27 KKR, GEC Tcr 9/14/2021
 Special Cass in LPP…
4. Alternate Optimal Solution: (Multiple Optimal Solution)
 Alternate or multiple optimal solution means, a problem has more
than one optimal solution
 A situation is that when the iso-z function (iso-profit or iso-cost)
line is parallel to any one or more of the constraint lines.

28 KKR, GEC Tcr 9/14/2021

 Solution of LPP – Big M Method
 Solve the following LPP using big M method
Min Z = 7x1 + 15x2 + 20x3
Subject to: 2x1 + 4x2 + 6x3 ≥ 24
3x1 + 9x2 + 6x3 ≥ 30
x1, x2, x3 ≥ 0
Note:
 For Simplex method to apply, the constraints must be lesser than
or equal so that slack variable s is added to establish equality.
 If all the constraints are with greater than or equal to (≥) inequality
sign, the usual Simplex method cannot be used to solve the LPP
 If the all the constraints are greater than or equal to (≥) inequality,
convert them to equality by subtracting surplus variable ‘s’ and
adding artificial variable ‘A’
29 KKR, GEC Tcr 9/14/2021
 Solution of LPP – Big M Method…
Modify the objective function and constraints by adding surplus
variables and artificial variables with coefficient M (a large
quantity)
Min Z = 7x1+15x2+20x3+0s1+0s2+MA1+MA2
Subject to: 2x1 + 4x2 + 6x3 – s1 + A1 = 24
3x1 + 9x2 + 6x3 –s2 + A2 = 30
x1, x2, x3,s1,s2,A1,A2 ≥ 0
Construct the initial Simplex table as before

30 KKR, GEC Tcr 9/14/2021

 Solution of LPP – Big M Method…
Initial Simplex Table
Cj 7 15 20 0 0 M M

Least positive value

Basic Solution Ratio
CBi x1 x2 x3 s1 s2 A1 A2
Variable

R1 M A1 2 4 6 -1 0 1 0 24 6

R2 M A2 3 9 6 0 -1 0 1 30 10/3

Zj 5M 13M 12M -M -M M M 54M

15- 20-
Δj= Cj - Zj 7-5M M M 0 0
13M 12M
Most negative value
For optimality, all the Cj-Zj values must be greater than or equal to
zero (min problem). Hence not optimal
31 KKR, GEC Tcr 9/14/2021
 Solution of LPP – Big M Method…
Iteration 1
Cj 7 15 20 0 0 M M

Basic Solution Ratio

CBi x1 x2 x3 s1 s2 A1 A2
Variable

R3 M A1 2/3 0 10/3 -1 4/9 1 - 32/3 16/5

R4 15 x2 1/3 1 2/3 0 -1/9 0 - 10/3 5

2/3M 10/3 4/9M 32/3M+
Zj 15 -M M
+5 M+10 -5/3 50
-10/3
M -4/9
-2/3
Δj= Cj - Zj 0 0
M+2 M+10 M+5/3
Put blank as A2 left
• Not optimal.
• The leaving variable is A1 and the entering variable is x3
32 KKR, GEC Tcr 9/14/2021
 Solution of LPP – Big M Method…
Iteration – 2
Cj 7 15 20 0 0 M M

Basic Solution Ratio

CBi x1 x2 x3 s1 s2 A1 A2
Variable
-
R5 20 x3 1/5 0 1 2/15 - - 16/5
3/10

R6 15 x2 1/5 1 0 1/5 -1/5 - - 6/5

Zj 7 15 20 -3 -1/3 - - 82
Δj= Cj - Zj 0 0 0 3 1/3 - -

Since all the Cj-Zj values are greater than or equal to zero, the
solution is optimal.
Solution: x1 = 0, x2 = 6/5, x3 = 16/5 and Z = 82
33 KKR, GEC Tcr [Z=7x1+15x2+20x3 = 7 (0)+15(6/5)+20(16/5) = 82] 9/14/2021
 Duality - Concept
 Every linear programming problem has a mirror image associated
with it.
 If the original problem is maximization, the mirror image will be
minimization and vice versa.
 The original problem is called 'primal' and the mirror image is
called 'dual'.
 Features of dual;
 Dual of the dual is primal
 If either the primal or dual has a solution, then the other also has a
solution.
 If any of the primal or dual is infeasible, then the other has an
unbounded solution.
 The optimal solution of both the primal and dual is equal.

34 KKR, GEC Tcr 9/14/2021

 Duality – Concept…
 For primal to dual conversion, the sign of the constraints must be as
below:
 For maximization – all constraints should be lesser than or equal to (≤)
 For minimization – all constraints should be greater than or equal to (≥)
 If the primal problem contains a large number of rows and a smaller
number of columns, we can reduce the computational procedure by
converting into dual. (Can use Simplex method)
 If the primal is a minimization problem with ≥ constraints, the dual is
Max problem with ≤ constraints and vice versa.
E.g. 1. Convert the following primal problem to its dual
Min Z = 4 x1 + 3 x2
Subject to:
200 x1 + 100 x2 ≥ 4000 ; x1 + 2 X2 ≥ 50
40 x1 + 40 x2 ≥ 1400
x1, x2 ≥ 0
35 KKR, GEC Tcr 9/14/2021
 Duality – Concept…
 Solution
Primal Dual
Minimisation Problem (Min. Z) Maximisation Problem (Max. Z*)
Constraints are of ' ≥′ type Constraints are of '≤' type.
Decision variables are x1, x2 etc. Decisión variables are y1, y2 etc.
Coefficients of objective function of primal RHS of constraints in Dual
(4 & 3)
RHS of Constraints of Primal (4000, 50, 1400) Coefficients of objective function of the Dual
In the LHS (left side) of constraints – all Horizontal coeficiente in constriñas of the
vertical coefficient values Dual: 200 y1 + 1 y2 + 40 y3 ≤ 4
200 x1 + 100 x2 ≥ 4000
1 x1 + 2 X2 ≥ 50
40 x1 + 40 x2 ≥ 1400
No of constraints in Primal No. of variables in Dual

36 KKR, GEC Tcr 9/14/2021

 Duality – Concept…
Primal Dual
Min Z = 4 x1 + 3 x2 Max Z* = 4000 y1 + 50 y2 + 1400 y3
Subject to: Subject to:
200 x1 + 100 x2 ≥ 4000 200 y1 +1 y2 + 40 y3 ≤ 4
1 x1 + 2 x2 ≥ 50 100 y1 + 2 y2 + 40 y3 ≤ 3
40 x1 + 40 x2 ≥1400 y1, y2, y3 ≥ 0
x1, x2 ≥ 0

E.g. 1. Get the dual of the following primal problem

Max Z = 40x1 + 35 x2 Solution:
Min Z*=60y1 + 96 y2
Subject to : 2x1 + 3x2 ≤ 60
Subject to: 2y1 + 4y2 ≥ 40
4x1 + 3x2 ≤ 96 ; x1, x2 ≥ 0 3y1 + 3y2 ≥ 35; y1, y2 ≥ 0
37 KKR, GEC Tcr 9/14/2021
 Duality – Concept…
E.g. 3. Min Z = 2x1 + 24 x2
Subject to: 20x1 + 50x2 ≥ 4800
80x1 + 500x2 ≥ 7200; x1, x2 ≥ 0
E.g. 4. Min Z =10x1 + 20x2
`Subject to: 3x1 + 2x2 ≥ 18
x1 + 3x2 ≥ 8
2x1 – x2 ≤ 6; x1, x2 ≥ 0
E.g. 5. Max Z =22x + 30y
`Subject to: 2x + 2y ≤ 100
2x + y + 22 ≤ 100
2x + 2y + 22 ≥ 100 ; x, y ≥ 0
38 KKR, GEC Tcr 9/14/2021
 Duality – Concept…
E.g. 6. Min Z = x1 + 2x2
Subject to: 2x1 + 4x2 ≤ 160
x1 – x2 = 30
x1 ≥ 10 ; x1, x2 ≥ 0
Solution:
First, convert the equality constraint into to inequality constraints
Convert all ‘≤’ constraint into ‘≥’
Min Z = x1 + 2x2 Equality constraint should be considered
as two constraints; one ≤ constraint and
Subject to: -2x1 - 4x2 ≥ 160 another ≥ constraint
x1 – x2 ≥ 30
-x1 + x2 ≥ -30 (from x1 – x2 ≤ 30)
x1 ≥ 10 ; x1, x2 ≥ 0
39 KKR, GEC Tcr 9/14/2021
 Dual Simplex Method
E.g. Solve the following LPP using Dual Simplex method
Max Z = 6 x1 + 8x2 The simplex algorithm is such that when we
Subject to: 5x1 + 2x2 ≤ 20 obtain optimal solution of ‘dual’, we
automatically get the optimal solution of the
x1 + 2 x2 ≤ 10 ‘primal’ and vice versa.
x1, x2 ≥ 0
Solution: First get the dual of the problem and solve it using the
Simplex method
Dual of the problem: Min Z* = 20y1 + 10 y2
Subject to: 5y1 +1y2 ≥ 6
2y1 + 2y2 ≥ 8
y1, y2 ≥ 0

40 KKR, GEC Tcr 9/14/2021

 Dual Simplex Method…
Modify the problem by incorporating surplus variable ‘s’
Min Z* = 20y1 + 10 y2 +0s1 + 0s2 +MA1 + MA2
Subject to: 5y1 +y2 - s1+A1 = 6
2y1 + 2y2 - s2 + A2 = 8
y1, y2, s1, s2, A1, A2 ≥ 0
Initial Simplex Table:

Cj 20 10 0 0 M M
CBi Basic Solution Ratio
y1 y2 s1 s2 A1 A2
Variable

M A1 5 1 -1 0 1 0 6 6/5

M A2 2 2 0 -1 0 1 8 4

Zj 7M 3M -M -M M M 14M
20- 10-
M M 0 0
41 Cj - Zj 7M 3M
KKR, GEC Tcr 9/14/2021
 Dual Simplex Method…
Iteration 1
Cj 20 10 0 0 M M
CBi Basic Solution Ratio
y1 y2 s1 s2 A1 A2
Variable

20 y1 1 1/5 -1/5 0 - 0 6/5 6

M A2 0 8/5 2/5 -1 - 1 28/5 14/4

4+ -4 + 24+
20 -M M
Zj 8/5M 2/5M 28/5M
6- 4-
0 M 0
Cj - Zj 8/5M 2/5M

Solution not optimal as all Cj-Zj are not ≥ 0 (minimization problem)

42 KKR, GEC Tcr 9/14/2021

 Dual Simplex Method…
Iteration 2

Cj 20 10 0 0 M M
CBi Basic Solution Ratio
y1 y2 s1 s2 A1 A2
Variable

20 y1 1 0 -1/4 1/8 - - 1/2

10 y2 0 1 1/4 -5/8 - - 7/2

Zj 20 10 -5/2 -30/8 45
Cj - Zj 0 0 5/2 30/8

Solution optimal as all Cj-Zj are ≥ 0 (minimization problem)

Optimal Solution: The (Cj-Zj) values corresponding to s1 and s2
i.e. x1 = (Cj-Zj)s1 = 5/2 and x2 = (Cj-Zj)s2 = 30/8 = 15/4
Optimal Z = 45 (Substitute in the primal obj. fn. and check)
43 KKR, GEC Tcr 9/14/2021
 Dual Simplex Method…
E.g. 2. Solve using dual simplex method
Min Z = 2x1 + 24 x2
Subject to: 20x1 + 50x2 ≥ 4800
80x1 + 500x2 ≥ 7200; x1, x2 ≥ 0
Solution:
Dual: Max Z*=4800 y1 + 7200 y2
Subject to: 20y1+80y2 ≤ 2
50y1 + 500y2 ≤ 24; y1, y2 ≥ 0
Modify by adding slack variables;
Max 4800 y1 + 7200 y2+0s1+0s2
20y1+80y2+s1=2; 50y1+500y2+s2=24; y1, y2,s1,s2 ≥ 0
44 KKR, GEC Tcr 9/14/2021
 Dual Simplex Method…
Initial simplex table

Cj 4800 7200 0 0
CBi Basic Solution Ratio
y1 y2 s1 s2
Variable

0 s1 20 80 1 0 2 1/40

0 s2 50 500 0 1 24 6/125

Zj 0 0 0 0 0
Cj - Zj 4800 7200 0 0

45 KKR, GEC Tcr 9/14/2021

 Dual Simplex Method…
Iteration 1
Cj 4800 7200 0 0
CBi Basic Solution Ratio
y1 y2 s1 s2
Variable

7200 y2 1/4 1 1/80 0 1/40 1/10

0 s2 -75 0 -5/8 1 23/2 -23/150

Zj 1800 7200 90 0 195

Cj - Zj 3000 0 -90 0

Yet again, the lowest positive ratio is for the first row. So, the
solution is unbounded for primal.

46 KKR, GEC Tcr 9/14/2021

 Dual Simplex Method…
E.g. 3. Min Z =6x1 + 4x2+3x3
`Subject to: 4x1 + 5x2+3x3 ≥ 40
2x1 +x2+6x3 ≥ 50
3x1 +4x2 +2x3 ≥ 60; x1, x2,x3 ≥ 0
Solution:
Dual of the LPP
Max Z* = 40y1+50y2+60y3 Max Z*=40y1+50y2+60y3+0s1+0s2+0s3
Subject to: 4y1+2y2+3y3 ≤6 S.T: 4y1+2y2+3y3+s1 = 6
5y1+y2+6y3 ≤ 4 5y1+y2+6y3 + s2 = 4
3y1+6y2+2y3 ≤ 3 3y1+6y2+2y3+s3 = 3
y1, y2,y3 ≥ 0 y1, y2,y3,s1,s2,s3 ≥ 0
47 KKR, GEC Tcr 9/14/2021
 Dual Simplex Method…
Initial simplex tableau
Cj 40 50 60 0 0 0
CBi Basic Solution Ratio
y1 y2 y3 s1 s2 s3
Variable

0 S1 4 2 3 1 0 0 6 2

0 S2 5 1 4 0 1 0 4 1

0 s3 3 6 2 0 0 1 3 3/2

Zj 0 0 0 0 0 0 0

40 50 60 0 0 0
Cj - Zj

48 KKR, GEC Tcr 9/14/2021

 Dual Simplex Method…
Iteration 1
Cj 40 50 60 0 0 0
CBi Basic Solution Ratio
y1 y2 y3 s1 s2 s3
Variable

0 S1 3/2 3/2 0 1 -1/2 0 4 8/3

60 y3 5/6 1/6 1 0 1/6 0 2/3 4

0 s3 4/3 17/3 0 0 -1/3 1 5/3 5/17

Zj 50 10 60 0 10 0 40

-10 40 0 0 -10 0
Cj - Zj

Optimal Z = 730/11, x1 = 130/11, x2 = 70/11 & x3=420/11 (ANS)

49 KKR, GEC Tcr 9/14/2021

 Sensitivity Analysis
 In the development of mathematical models, we always make
approximations and therefore replication of exact reality is impossible.
 Linearity assumption is a significant approximation in LPP
 Another important approximation comes because of the uncertainty of
the data that put into the model
 i.e. the values in A, B or C may have uncertainties
 Moreover, information may change as time progresses resulting in change
in C values and sometimes in A and B values, also.
 Sensitivity analysis is a systematic study of how sensitive the solutions are
to (small) changes in the data.
 The basic idea is to be able to give answers to questions of the form:
1. If the objective function (C) changes, what happened to the solution ?
2. If resources available changes (A&B), what happened to the solution?
3. If a new constraint is added to the problem, what happened to the
solution

50 KKR, GEC Tcr 9/14/2021

 Sensitivity Analysis…
 So, the objective of sensitivity analysis is to determine the new
values of solution (if possible, from the given simplex solution).
 It is established that the solution will be unaltered, only if the
changes in the objective function or constraint capacities is within
certain limits.
 These will be two limits, lower limit (max. possible decrease) and
upper limit (max. possible increase).
 The sensitivity analysis provides these two limits within which the
simplex table (the solution) remains optimal.
 So, as long as the variation in C, A and B is within the limits, the
same simplex tableau can be used to find the optimal solution.

51 KKR, GEC Tcr 9/14/2021

 Sensitivity Analysis…
 The final simplex tableau of a Max. problem is shown
 Since all the Cj-Zj values are less than or equal to zero, the simplex
table shows optimal solution.
Cj 100 60 40 0 0 0
CBi Basic Solution Ratio
x1 x2 x3 s1 s2 s3
Variable

60 x2 0 1 5/6 5/3 -1/6 0 200/3

100 x1 1 0 1/6 -2/3 1/6 0 100/3

0 s3 0 0 4 0 0 1 100

Zj 100 60 200/3 100/3 20/3 0 22000/3

Cj - Zj 0 0 -80/3 -100/3 -20/3 0

52 KKR, GEC Tcr X1 = 100/3, x2 = 200/3, x3 = 0 and Z = 22000/3 9/14/2021

 Sensitivity Analysis…
I. Range (limits) of profit coefficients (Cj) for the solution applies:
j
Range = of the final simplex tableau
xn
where Δj is the values of Δj row in the simplex tableau and xn is the
values in xi row corresponding to Δj in the simplex tableau
j
1. Range of coefficients of x1 =
x1
− 80 − 100 − 20
i.e. 3 = −160 3 = 50 3 = −40
1 −2 1
6 3 6
This indicates that the (+) side limit is 50 and the (-) side limit is the
least of 160 or 40 , i.e. 40.
Since the original value of x1 = 100, the limits are;
100+50 = 150 and 100-40 = 60 (even if the cost coefficient of x1
varies from 60 to 150, the final simplex solution is valid)
53 KKR, GEC Tcr 9/14/2021
 Sensitivity Analysis…
j
2. Range of coefficients of x2 =
x2
− 80 − 100 − 20
3 = −32 3 = −20 3 = 40
i.e.
5 5 −1
6 3 6
This indicates that the (+) side limit is 40 and the (-) side limit is
the least of 32 or 20 , i.e. 20.
Since the original value of x2 = 60, the limits are;
60+40 = 100 and 60 -20 = 40 (i.e. even if the cost coefficient of
x2 varies from 40 to 100, the final simplex solution is valid)

54 KKR, GEC Tcr 9/14/2021

 Sensitivity Analysis…
3. Range of coefficients of x3
x3 = 0 in the solution
From the simplex table, Δ of x3 = -80/3
 It means, x3 will have influence on the profit Z, only if the
present cost coefficient is increased by 80/3.
 i.e. x3 will be included, if its coefficient becomes [40 + 80/3] =
Rs. 200/3 or more.
 Hence, the present simplex solution remains optimal till the
coefficient of x3 is 200/3.

55 KKR, GEC Tcr 9/14/2021

 Sensitivity Analysis
II. Range (limits) of Capacity of Resources (RHS of constraints B):
Range of capacity of resources B will be;
B Value in the Bi or Solution column
=− =
s n Value in the corresonding s column
▪ In the given problem, there are three constraints (as there are three
slack variables)
1. Range for capacity represented by first constraint:
▪ Capacity of first constraint is represented in simplex table by slack
variable s1:
Thus, the Range will be the ratio of ‘B' column and ‘s1' column.
200 100
i.e. − 3 = −40, − 3 = 50
5 −2
56 KKR, GEC Tcr 3 3 9/14/2021
 Sensitivity Analysis…
▪ This indicates that the (+) side limit is 50 and the (-) side limit is 40.
▪ It Is given in the problem that the RHS of constraint 1 is 100,
constraint 2 is 600 and constraint 3 is 300
▪ Since the original value RHS of constraint 1 = 100, the limits are;
100+50 = 150 and 100 - 40 = 60 (i.e. even if the RHS of
constraint 1 varies from 60 to 150, the final simplex solution is
valid)
2. Range for capacity represented by the second constraint:
▪ Capacity of second constraint is represented in simplex table by
slack variable s2:
Thus, the Range will be the ratio of ‘B' column and ‘s2' column.
200 100
i.e. − 3 = 400, − 3 = −200
KKR, GEC Tcr
−1 1
57 6 6 9/14/2021
 Sensitivity Analysis…
▪ This indicates that the (+) side limit is 400 and the (-) side limit is
200.
▪ Since the original value RHS of constraint 2 = 600, the limits are;
600+400 = 1000 and 600 - 200 = 400 (i.e. even if the RHS of
constraint 2 varies from 400 to 1000, the final simplex solution is
valid)
3. Range for capacity represented by the third constraint:
▪ Capacity of third constraint is represented in simplex table by slack
variable s3:
Thus, the Range will be the ratio of ‘B' column and ‘s3' column.
200 100
i.e. − 3 = , − 3 = , − 100 = −100
0 0 1
58 KKR, GEC Tcr 9/14/2021
 Sensitivity Analysis…
▪ This indicates that the (+) side limit is ∞ and the (-) side limit is
100.
▪ Since the original value RHS of constraint 3 = 300, the limits are;
300+ ∞ = ∞ and 300 - 100 = 200 (i.e. even if the RHS of
constraint 3 varies from 200 to ∞, the final simplex solution is
valid)
E.g. What will be the effect on solution, if capacity of the first
constraint (RHS of constraint 1) is increased by 30%?
Solution:
Original capacity = 100 hrs. (given)
Increase = 30%
New capacity = 100 + (0.3 x 100) = 130 hrs.

59 KKR, GEC Tcr 9/14/2021

 Sensitivity Analysis…
Note:
▪ To find the effect on solution, we need to find the range (limit) of
resource capacity for which the solution remains applicable.
▪ Also, we could find the effect on solution, only if the new capacity is
within the range or limit so determined.
▪ If the new capacity is outside the range, then we have to solve the
problem afresh.
• We have seen that the limit of first resource capacity is (100 + 50) = 150
and (100 – 40) = 60.
• Therefore, the new capacity, 130 is within the range.
The change in capacity = 130 - 100 = +30 hrs.
As the first resource capacity of first constraint is represented by the s1
column of final simplex table, multiply column S1 by +30 to get the
resulting change in ‘B' column.

60 KKR, GEC Tcr 9/14/2021

 Sensitivity Analysis…
S1 x 30 Solution
60 x2 5/3 x 30 = 50 200/3 + 50
100 x1 -2/3 x 30 = -20 100/3 - 20
Z 25000/3

(Revised Z = [60 x (200/3 + 50)] + [100 x (100/3-20)]= 25000/3)

Increase in profit = revised profit – original profit
= 25000/3 – 22000/3 = 1000
Revised x1 = 40/3; Revised x2 = 350/3

61 KKR, GEC Tcr 9/14/2021

62 KKR, GEC Tcr 9/14/2021