1

DJM3A - OPTIMIZATION TECHNIQUES

Unit I

Linear Programming Problem(LPP): Mathematical Formulation - Graphical

Method of Solution – Simplex Method – Big „M‟ Method – Two Phase Simplex
Method – Duality – Dual Simplex Method.

Unit II

Transportation Problems: Mathematical Formulation – Balanced and

unbalanced TP – North-West Corner Rule – Least Cost Method – Vogel‟s
Approximation Method – Test for Optimality – Maximization problems in TP –
Assignment Problems(AP): Mathematical Formulation – Method of Solution –
Maximization in AP.

Unit III

Sequencing Problems: Introduction – Basic Assumptions – Sequencing n Jobs

on 2 Machines – Sequencing n Jobs on 3 machines – Sequencing 2 Jobs on n
Machines. Queuing Theory : General Concepts and Definitions – Classification
of queues – Poisson Process, Properties of Poisson Process – Queuing Models:
1. (M/M/1):(∞/FCFS), 2. (M/M/1):(N/FCFS), 3. (M/M/c):( ∞/FCFS).

Unit IV

Inventory Control: Basics – Types of Inventory Models: Deterministic Models:

Model I Purchase Model without Shortages – Model II Production Model
without Shortages – Model III Purchase Model with Shortages – Model IV
Production Model with Shortages – Model V Price Break Model. Probabilistic
Models: Newspaper boy problem – Discrete and Continous type cases – ABC
Analysis.

Unit V

Network Analysis: Drawing Network Diagrams – Critical Path Method –

Concept of Slack and Floats on network – PERT – Three types of time
estimates – Algorithm for PERT – Differences in PERT and CPM.
 2

UNIT-I
Linear Programming Problem
Linear programming can be used in a variety of situations.
In most of the business or economical situation, resources will be limited. The
problem there will be to make use of the available resources in such a way that
to maximise the production or to minimize the expenditure. These datas can be
formulated as linear programming models.

The objective of the linear programming problem is to

maximize the profit and minimize the total cost.

The LPP is to determine the values of the decision variables

such that all the constraints are satisfied and gives the maximum or minimum
value for the objective function. The maximum or minimum value of the
objective function is called an Optimum value.

Formulation of LPP:

The formulation of any situation to a LPP is based on the

following guidelines:

1. Identification of decision variables

2. Formation of objective function which is to be either maximize or
minimize.
3. The various constraints involved due to the limited availability of
resources.

Mathematical Formulation of LPP:

The general form of LPP is as follows:

𝑀𝑎𝑥 𝑜𝑟 𝑀𝑖𝑛 𝑧 = 𝑐1 𝑥1 + 𝑐2 𝑥2 + ⋯ + 𝑐𝑛 𝑥𝑛
≤
Subject to, 𝑎11 𝑥1 + 𝑎12 𝑥2 + ⋯ + 𝑎1𝑛 𝑥𝑛 ≥ 𝑏1
=

≤
𝑎21 𝑥1 + 𝑎22 𝑥2 + ⋯ + 𝑎2𝑛 𝑥𝑛 ≥ 𝑏2
=

………….
≤
𝑎𝑚 1 𝑥1 + 𝑎𝑚 2 𝑥2 + ⋯ + 𝑎𝑚𝑛 𝑥𝑛 ≥ 𝑏𝑚 , 𝑥𝑖 ≥ 0, ∀𝑖.
=
 3

Here the variables 𝑥𝑖 are called decision variables. The function z is called the
objective function and 𝑐1 , 𝑐2 … 𝑐𝑛 are called the cost variables.

The linear constraints of the general LPP involved one of the

relations,” = or ≤ 𝑜𝑟 ≥”. The addition of some non-negative variables on the
left hand side of the inequations may convert inequations into equations. For
that purpose we use the variables called slack or surplus variables.The variable
which we add to constraint equation is called slack variable. The variable which
we subtract to constraint equation is called surplus variable.

An n-tuple(𝑥1 , 𝑥2 … 𝑥𝑛 ) of real numbers which satisfies the

linear constraints and the non-negative restriction of the given LPP is called
feasible solution.

An n-tuple(𝑥1 , 𝑥2 … 𝑥𝑛 ) of real numbers which satisfies the

linear constraints and the non-negative restriction of the given LPP and
optimize the objective function is called optimum feasible solution.

A basic solution to the system AX=B is called degenerate if

one or more of the basic variables are zero.

Graphical Method
If the objective function z is a function of two variables then
the problem can be solved by graphical method. The procedure is as follows.

Step 1: First of all we consider the constraints as equalities or equations.

Step 2: Then we draw the lines in the plane corresponding to each equation
obtained in step 1 and non-negative restrictions.

Step 3: Then we find the permissible region(region which is common to all the
equations) for the values of the variables which is the region bounded by the
lines drawn in step 2.

Step 4: Finally we find a point in the permissible region which gives the
optimum value of the objective function.

Note:

1. If there is no permissible region in a problem then we say that the

problem has no solution using graphical method.
 4

2. If the maximum value of z appears only at ∞ then the problem has

unbounded solution.
3. If the maximum value of z appears at 2 points there exists infinite
number of solutions to the LPP.

Problem:

Find the maximum value of z to the following LPP using graphical method,
𝑀𝑎𝑥 𝑧 = 5𝑥1 + 7𝑥2 , subject to 𝑥1 + 𝑥2 ≤ 4, 3𝑥1 + 8𝑥2 ≤ 24,

10𝑥1 + 7𝑥2 ≤ 35 𝑎𝑛𝑑 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2.

Solution:

The given problem contains only two variables 𝑥1 𝑎𝑛𝑑 𝑥2 . So graphical

method is possible. Consider the equations, we get

𝑥1 + 𝑥2 = 4,

3𝑥1 + 8𝑥2 = 24,

10𝑥1 + 7𝑥2 = 35

Now to find the points,

Consider 𝑥1 + 𝑥2 = 4,

Put 𝑥1 = 0 => 𝑥2 = 4

and 𝑥2 = 0 => 𝑥1 = 4

Therefore (0,4) and (4.0) are the pts.

Similarly consider 3𝑥1 + 8𝑥2 = 24.

We get (0,3) and (8,0) are the pts.

Consider 10𝑥1 + 7𝑥2 = 35.

We get (0,5) and (3.5,0) are the pts.

The graphical representation is

 5

Solution is 𝑥1 = 1.6, 𝑥2 = 2.4 and 𝑀𝑎𝑥 𝑧 = 24.8

Exercise:

1. Solve graphically the following LPP 𝑀𝑎𝑥 𝑧 = 𝑥1 + 𝑥2 , subject to

5𝑥1 + 3𝑥2 ≤ 15, 𝑥1 + 𝑥2 ≥ 6, 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2.
2. Solve graphically the following LPP 𝑀𝑖𝑛 𝑧 = 2𝑥1 + 𝑥2 , subject to
3𝑥1 + 𝑥2 ≥ 3, 4𝑥1 + 3𝑥2 ≥ 6, 𝑥1 + 2𝑥2 ≥ 2, 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2.

Simplex Method
The procedure is as follows.

Step 1: The objective function of the given LPP is to be maximised or convert

into a maximisation problem.

Step 2: All the 𝑏𝑖 ‟s should be non-negative. If some 𝑏𝑖 is negative multiply the

corresponding equation by -1.

Step 3: The inequations of the constraint must be converted into equations by

introducing slack or surplus variables in the constraints. The cost of these
variables are taken to be zero.
 6

Step 4: Obtain the initial basic feasible solution 𝑋𝐵 = 𝐵−1 𝑏 and form the
starting simplex table.
𝑛
Step 5: Compute the net evaluation 𝑧𝑗 − 𝑐𝑗 (j=1,2,…n), where 𝑧𝑗 = 𝑗 =1 𝐶𝐵𝑖 𝑌𝑖𝑗
(i=1,2..m). If all 𝑧𝑗 − 𝑐𝑗 ≥ 0, then the initial basic feasible soln is an optimum
basic feasible solution.

Step 6: Choose the most negative 𝑧𝑗 − 𝑐𝑗 . Let it be 𝑧𝑟 − 𝑐𝑟 . If all 𝑌𝑖𝑟 < 0 then
there is an indication of unbounded solution.
𝑋 𝐵𝑖
Step 7: Compute the ratio . The (k,r)th element is called leading element or
𝑌𝑖𝑟
pivot element.

Step 8: Divide each element of the kth row by the pivot element and make all
other elements in the rth column to zero by elementary row operation.

Step 9: Repeat step 5 to 8 until an optimum solution is obtained.

Problem:

Solve the following LPP using simplex method,

𝑀𝑎𝑥 𝑧 = 5𝑥1 + 3𝑥2 , subject to 𝑥1 + 𝑥2 ≤ 2, 5𝑥1 + 2𝑥2 ≤ 10,

3𝑥1 + 8𝑥2 ≤ 12 𝑎𝑛𝑑 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2.

Solution:

The objective function is maximization so the resulting LPP becomes

𝑀𝑎𝑥 𝑧 = 5𝑥1 + 3𝑥2 + 0𝑥3 + 0𝑥4 + 0𝑥5 subject to

𝑥1 + 𝑥2 + 𝑥3 = 2

5𝑥1 + 2𝑥2 + 𝑥4 = 10

3𝑥1 + 8𝑥2 +𝑥5 = 12, 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2. . .5

The initial basic feasible solution is obtained by putting 𝑥1 = 𝑥2 = 0 in the

reformulated form of LPP and we get 𝑥3 = 2, 𝑥4 = 10, 𝑥5 = 12.

Starting Simplex Table:

 7

𝐶𝑗 5 3 0 0 0 Ratio

𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4 𝑌5 𝑋𝐵𝑖
𝑌𝑖𝑟
0 𝑋3 =2 𝑌3 1 1 1 0 0 2
0 𝑋4 =10 𝑌4 5 2 0 1 0 2
0 𝑋5 =12 𝑌5 2 8 0 0 1 4
𝑛
0 0 0 0 0
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗
𝑗 =1
𝑧𝑗 − 𝑐𝑗 -5 -3 0 0 0

First Iteration:
𝑜𝑙𝑑 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛
New Pivot equation=
𝑝𝑖𝑣𝑜𝑡 𝑒𝑙𝑒𝑚𝑒𝑛𝑡

𝐶𝑗 5 3 0 0 0
𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4 𝑌5
5 𝑋1 =2 𝑌1 1 1 1 0 0
0 𝑋4 =0 𝑌4 0 -3 -5 1 0
0 𝑋5 =6 𝑌5 0 5 -3 0 1
𝑛
5 5 5 0 0
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗
𝑗 =1
𝑧𝑗 − 𝑐𝑗 0 2 5 0 0

Hence an optimal basic feasible solution is obtained.

Solution is 𝑥1 = 2, 𝑥2 = 0 and 𝑀𝑎𝑥 𝑧 = 10

Exercise:

1. Solve the following LPP using simplex method,

𝑀𝑖𝑛 𝑧 = 𝑥1 − 3𝑥2 + 2𝑥3 , subject to 3𝑥1 − 𝑥2 + 2𝑥3 ≤ 7,
−2𝑥1 + 4𝑥2 ≤ 12,−4𝑥1 + 3𝑥2 + 8𝑥3 ≤ 10 𝑎𝑛𝑑 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2,3.
2. Solve the following LPP using simplex method,
 8

𝑀𝑎𝑥 𝑧 = 25𝑥1 + 20𝑥2 , subject to 16𝑥1 + 12𝑥2 ≤ 100,

8𝑥1 + 16𝑥2 ≤ 80,𝑎𝑛𝑑 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2.

Big-M Method[Method of Penalty]

In the Big-M method huge negative cost (-M) is assigned to
the artificial value, while zero cost to the slack and surplus variables in the
objective function. The usual simplex method is then followed. At any iteration
there can arise any one of the following two cases.

Case(i): There is atleast one vector corresponding to some artificial variable in

the basis at zero level and all the net evaluation 𝑧𝑗 − 𝑐𝑗 ≥ 0. In this case the
given LPP doesnot possess an optimum basic feasible solution.

Case(ii): There is atleast one vector corresponding to some artificial variable in

the basis not at zero level and all the net evaluation 𝑧𝑗 − 𝑐𝑗 ≥ 0. In this case the
given LPP doesnot possess an optimum basic feasible solution.

Whenever an artificial variable leaves from the basis we omit

all the entries corresponding to that variable from the simplex table in the next
iteration. A drawback of this method is the possible computational error that
could result from assigning a very large value to M.

Problem:

Solve the following LPP using Big-M method, 𝑀𝑖𝑛 𝑧 = 12𝑥1 + 20𝑥2 , subject
to 6𝑥1 + 8𝑥2 ≥ 100, 7𝑥1 + 12𝑥2 ≥ 120,𝑎𝑛𝑑 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2.

Solution:

The objective function is converted to maximisation objective function.

The resulting LPP becomes,

𝑀𝑎𝑥 − 𝑧 = −12𝑥1 − 20𝑥2 + 0𝑥3 + 0𝑥4 − 𝑀𝑥5 − 𝑀𝑥6 subject to

6𝑥1 + 8𝑥2 − 𝑥3 + 𝑥5 = 100

7𝑥1 + 12𝑥2 − 𝑥4 + 𝑥6 = 120, 𝑥𝑖 ≥ 0, ∀ 𝑖.

The initial basic feasible solution is obtained by putting 𝑥1 = 𝑥2 = 𝑥3 = 𝑥4 = 0

in the reformulated form of LPP and we get 𝑥5 = 100, 𝑥6 = 120.

Starting Simplex Table:

 9

𝐶𝑗 -12 -20 0 0 -M -M Ratio

𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4 𝑌5 𝑌6 𝑋𝐵𝑖
𝑌𝑖𝑟
-M 𝑋5 =100 𝑌5 6 8 -1 0 1 0 12.5
-M 𝑋6 =120 𝑌6 7 12 0 -1 0 1 10
𝑛
-13M -20M M M -M -M
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗
𝑗 =1
𝑧𝑗 − 𝑐𝑗 -13M+20 - M M 0 0
20M+20

First Iteration:

𝐶𝑗 -12 -20 0 0 -M Ratio

𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4 𝑌5 𝑋𝐵𝑖
𝑌𝑖𝑟
-M 𝑋5 =20 𝑌5 4/3 0 -1 2/3 1 15
-20 𝑋2 =10 𝑌2 7/12 1 0 -1/12 0 120/7
𝑛 -4M/3-35/3 -
-20 M -M
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗 2M/5+5/3
𝑗 =1
𝑧𝑗 − 𝑐𝑗 -4M/3+1/3 0 M - 0
2M/5+5/3

Second Iteration:

𝐶𝑗 -12 -20 0 0
𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4
-12 𝑋1 =15 𝑌1 1 0 -3/4 ½
-20 𝑋2 =5/4 𝑌2 0 1 7/16 -5/12
𝑛
-12 -20 ¼ 7/3
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗
𝑗 =1
𝑧𝑗 − 𝑐𝑗 0 0 ¼ 7/3

Hence an optimal basic feasible solution is obtained.

Solution is 𝑥1 = 15, 𝑥2 = 5/4 and 𝑀𝑎𝑥 − 𝑧 = −205

 10

Therefore 𝑀𝑖𝑛 𝑧 = − −205 = 205

Exercise:

1. Solve the following LPP using Big-M method, 𝑀𝑎𝑥 𝑧 = 𝑥1 + 2𝑥2 ,

subject to 2𝑥1 + 𝑥2 ≤ 2, 3𝑥1 + 4𝑥2 ≥ 12𝑎𝑛𝑑 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2.
2. Solve the following LPP using Big-M method, 𝑀𝑎𝑥 𝑧 = 8𝑥1 − 4𝑥2 ,
subject to 4𝑥1 + 5𝑥2 ≤ 20, 𝑥1 − 3𝑥2 ≤ 23,𝑎𝑛𝑑 𝑥1 ≥ 0, 𝑥2 is
unrestricted.

Two-Phase Simplex Method

Two-phase method consists of two different phases.

Phase(i): In phase (i) we consider the objective objective function min

z*=sum of the artificial variables and cost of the artificial variables are taken
to be -1 in the objective function. Here the cost of the given variables, slack
variables and surplus variables are taken to be zero. Solve the LPP as usual
simplex method. If min z*=0 then we proceed to phase(ii).

Phase(ii): Final table of phase(i) is initial table of phase(ii) but we consider

the original objective function, ie) replace the cost by original cost.

Problem:

Solve the following LPP using Big-M method, 𝑀𝑖𝑛 𝑧 = 60𝑥1 + 80𝑥2 , subject
to 20𝑥1 + 30𝑥2 ≥ 900, 40𝑥1 + 30𝑥2 ≥ 1200,𝑎𝑛𝑑 𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2.

Solution:

The objective function is converted to maximisation objective function.

The resulting LPP becomes,

𝑀𝑎𝑥 − 𝑧 = −60𝑥1 − 80𝑥2 + 0𝑥3 + 0𝑥4 − 𝑀𝑥5 − 𝑀𝑥6 subject to

20𝑥1 − 30𝑥2 − 𝑥3 + 𝑥5 = 900

40𝑥1 + 30𝑥2 − 𝑥4 + 𝑥6 = 1200, 𝑥𝑖 ≥ 0, ∀ 𝑖.

The initial basic feasible solution is obtained by putting 𝑥1 = 𝑥2 = 𝑥3 = 𝑥4 = 0

in the reformulated form of LPP and we get 𝑥5 = 900, 𝑥6 = 1200.

Starting Simplex Table:

 11

𝐶𝑗 -60 -80 0 0 -M -M Ratio

𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4 𝑌5 𝑌6 𝑋𝐵𝑖
𝑌𝑖𝑟
-M 𝑋5 =900 𝑌5 20 30 -1 0 1 0 45
-M 𝑋6 =1200 𝑌6 40 30 0 -1 0 1 30
𝑛
-60M -60M M M -M -M
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗
𝑗 =1
𝑧𝑗 − 𝑐𝑗 -60M+60 - M M 0 0
60M+80

First Iteration:

𝐶𝑗 -60 -80 0 0 -M Ratio

𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4 𝑌5 𝑋𝐵𝑖
𝑌𝑖𝑟
-M 𝑋5 =300 𝑌5 0 15 -1 1/2 1 20
-60 𝑋1 =30 𝑌6 1 ¾ 0 -1/40 0 40
𝑛 - -
-60 M -M
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗 15M-45 M/2+3/2
𝑗 =1
𝑧𝑗 − 𝑐𝑗 0 - M - 0
15M+35 M/2+3/2

Second Iteration:

𝐶𝑗 -60 -80 0 0
𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4
-80 𝑋2 =20 𝑌2 0 1 -1/15 1/30
-60 𝑋1 =15 𝑌1 1 0 1/20 -1/20
𝑛
-60 -80 7/3 1/3
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗
𝑗 =1
𝑧𝑗 − 𝑐𝑗 0 0 7/3 1/3

Hence an optimal basic feasible solution is obtained.

Solution is 𝑥1 = 15, 𝑥2 = 20 and 𝑀𝑎𝑥 − 𝑧 = −2500

 12

Therefore 𝑀𝑖𝑛 𝑧 = − −2500 = 2500

Exercise:

1. Solve the following LPP using Big-M method, 𝑀𝑎𝑥 𝑧 = 2𝑥1 + 3𝑥2 ,
subject to 𝑥1 + 𝑥2 + 2𝑥3 ≤ 5, 2𝑥1 + 3𝑥2 + 4𝑥3 = 12𝑎𝑛𝑑 𝑥𝑖 ≥ 0, ∀ 𝑖 =
1,2,3.
2. Solve the following LPP using Big-M method, 𝑀𝑎𝑥 𝑧 = 4𝑥1 + 𝑥2 ,
subject to 3𝑥1 + 𝑥2 = 3, 𝑥1 + 2𝑥2 ≤ 4,4𝑥1 + 3𝑥2 ≥ 6, 𝑎𝑛𝑑 𝑥1 , 𝑥2 ≥ 0.

Duality
Every LPP associated with another LPP is called the dual of the
problem. The given problem is called primal problem. If the optimal solution
of one problem is known then the optimal solution of other is also known.

Formulation of Dual LPP:

Consider the following LPP in canonical form,

𝑀𝑎𝑥 𝑧 = 𝑐1 𝑥1 + 𝑐2 𝑥2 + ⋯ + 𝑐𝑛 𝑥𝑛

Subject to, 𝑎11 𝑥1 + 𝑎12 𝑥2 + ⋯ + 𝑎1𝑛 𝑥𝑛 ≤ 𝑏1

𝑎21 𝑥1 + 𝑎22 𝑥2 + ⋯ + 𝑎2𝑛 𝑥𝑛 ≤ 𝑏2

………….

𝑎𝑚 1 𝑥1 + 𝑎𝑚 2 𝑥2 + ⋯ + 𝑎𝑚𝑛 𝑥𝑛 ≤ 𝑏𝑚 , 𝑥𝑖 ≥ 0, ∀𝑖.

The above problem is called the primal problem and the variables are called
primal variables and the constraints are called primal constraints.

The dual of the given primal problem is defined as,

𝑀𝑖𝑛 𝑧 ∗ = 𝑏1 𝑦1 + 𝑏2 𝑦2 + ⋯ + 𝑏𝑚 𝑦𝑚

Subject to, 𝑎11 𝑦1 + 𝑎21 𝑦2 + ⋯ + 𝑎𝑚 1 𝑦𝑚 ≥ 𝑐1

𝑎12 𝑦1 + 𝑎22 𝑦2 + ⋯ + 𝑎𝑚 2 𝑦𝑚 ≥ 𝑐2

………….

𝑎1𝑛 𝑦1 + 𝑎2𝑛 𝑦2 + ⋯ + 𝑎𝑚𝑛 𝑦𝑚 ≥ 𝑐𝑛 , 𝑦𝑖 ≥ 0, ∀𝑖.

 13

Here the variables are called dual variables and the constraints are called dual
constraints.

Example: Write the dual of the following LPP.

𝑀𝑎𝑥 𝑧 = 𝑥1 + 2𝑥2 , subject to 2𝑥1 − 3𝑥2 ≤ 3, 4𝑥1 + 𝑥2 ≤ −4, 𝑥1 , 𝑥2 ≥ 0.

The dual is,

𝑀𝑖𝑛 𝑧 ∗ = 3𝑦1 − 4𝑦2 , subject to 2𝑦1 + 4𝑦2 ≥ 1, −3𝑦1 + 𝑦2 ≥ 2, 𝑥1 , 𝑥2 ≥ 0.

Dual Simplex Method

The dual simplex method is moreover same as the
standard simplex method. Here we donot require any artificial variable. Hence a
lot of computational work is minimized in this method. The procedure is as
follows.

Step 1: Convert the minimization LPP into maximization LPP.

Step 2: Convert all ≥ inequalities into ≤ inequalities by multiplying the

corresponding equation by-1.

Step 3: Introduce any slack variables and obtain the starting equation.

Step 4: After finding out the initial basic feasible solution construct the starting
simplex table. In this table calculate the net evaluation 𝑧𝑗 − 𝑐𝑗 and test the
nature of 𝑧𝑗 − 𝑐𝑗 . Here the following subcases arises.

Case(i): If all 𝑧𝑗 − 𝑐𝑗 ≥ 0 and the XBi column vectors are ≥ 0 optimum

basic feasible solution is obtained.

Case(ii): All 𝑧𝑗 − 𝑐𝑗 ≥ 0 and and one of XBi < 0 we go to the bnext stepo.

Case(iii): If one 𝑧𝑗 − 𝑐𝑗 < 0 then the dual simplex method fails.

Step 5: Determine the leaving variable by selecting the most negative XBi . Then
𝑧 𝑗 −𝑐 𝑗
calculate the replacement ratio max⁡
( , 𝑌𝑘𝑗 < 0) and determine the entering
𝑌𝑘𝑗
variable.

Step 6: The leaving row and the entering column intersect at a point and that
element is called pivot element.
 14

Step 7: Now test for optimality and repeat the process until the optimum basic
feasible solution is obtained or there is an indication of no feasible solution.

Problem:

Solve the following LPP using Dual Simplex method, 𝑀𝑖𝑛 𝑧 = 3𝑥1 + 𝑥2 ,
subject to 𝑥1 + 𝑥2 ≥ 1, 2𝑥1 + 3𝑥2 ≥ 2, 𝑥1 , 𝑥2 ≥ 0.

Solution:

The objective function is a minimization problem so convert it into a

maximisation problem. The resulting LPP becomes,

𝑀𝑎𝑥 𝑧 ∗ = −3𝑥1 − 𝑥2 + 0𝑥3 + 0𝑥4 subject to

−𝑥1 − 𝑥2 + 𝑥3 = −1

−2𝑥1 − 3𝑥2 + 𝑥4 = −2

𝑥𝑖 ≥ 0, ∀ 𝑖 = 1,2,3,4.

The initial basic feasible solution is obtained by putting 𝑥1 = 𝑥2 = 0 in the

reformulated form of LPP and we get 𝑥3 = −1, 𝑥4 = −2.

Starting Simplex Table:

𝐶𝑗 -3 -1 0 0
𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4
0 𝑋3 =-1 𝑌3 -1 -1 1 0
0 𝑋4 =-2 𝑌4 -2 -3 0 1
𝑛
0 0 0 0
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗
𝑗 =1
𝑧𝑗 − 𝑐𝑗 3 1 0 0

For calculating the entering variable we find out the replacement ratio
𝑧 𝑗 −𝑐 𝑗
By max⁡
( , 𝑌𝑘𝑗 < 0)=max(-3/2, -1/3, 0/0, 0/1)= -1/3.
𝑌𝑘𝑗

First Iteration:

𝐶𝑗 -3 -1 0 0
𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4
 15

0 𝑋3 =- 𝑌3 -1/3 0 1 -1/3
1/3
-1 𝑋2 =2/3 𝑌2 2/0 1 0 -1/3
𝑛
-2/3 -1 0 1/3
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗
𝑗 =1
𝑧𝑗 − 𝑐𝑗 7/3 0 0 1/3

𝑧 𝑗 −𝑐 𝑗
Calculate the ratio max⁡
( , 𝑌𝑘𝑗 < 0)=max(-7, 0/0, 0/1, -1)= -1.
𝑌𝑘𝑗

Second Iteration:

𝐶𝑗 -3 -1 0 0
𝐶𝐵 𝑋𝐵 𝑌𝐵 𝑌1 𝑌2 𝑌3 𝑌4
0 𝑋4 =1 𝑌4 1 0 -3 1
-1 𝑋2 =1 𝑌2 -1/3 1 -1 0
𝑛
1/3 -1 1 0
𝑧𝑗 = 𝐶𝐵𝑖 𝑌𝑖𝑗
𝑗 =1
𝑧𝑗 − 𝑐𝑗 10/3 0 1 0

Hence an optimal basic feasible solution is obtained.

Solution is 𝑥1 = 0, 𝑥2 = 1 and 𝑀𝑎𝑥 𝑧 ∗ = −1

Min z= - 𝑀𝑎𝑥 𝑧 ∗= -(-1)=1

************************************
 16

UNIT-II
Transportation Problems
Transportation problem is one of the subclasses of Linear
Programming problem and can be regarded as the generalization of Assignment
problem. The TP leads with the determination of minimum cost plan for
transporting a single commodity from a number of sources to a number of
destinations.

Mathematical Formulation
Let O1, O2,…,Om be the m origins and D1, D2,…,Dn be the n
destinations. Let ai be the amount of supply at the origin and bj be the demand at
destination. Let Cij be the unit transportation cost.

The transportation problem is, Min z= 𝐶𝑖𝑗 𝑋𝑗

Subject to 𝑗 𝑥𝑖𝑗 ≤ 𝑎𝑖 ,

𝑗 𝑥𝑖𝑗 ≥ 𝑏𝑗

If the total supply equal to the total demand then the TP is

called balanced TP. Otherwise it is unbalanced. In this case we have either
𝑎𝑖 > 𝑏𝑗 or 𝑎𝑖 < 𝑏𝑗 .

If the transportation table have mn unknowns there are m+n

constraints. These m+n equations are not independent. Hence there are m+n-1
independent constraints.

A feasible solution to a TP is a set of non-negative individual

allocations which satisfies the row and the column.

A feasible solution of a TP is said to be a basic feasible

solution if the total number of positive allocations is exactly equal to m+n-1.

A feasible solution is said to be optimum if it minimizes as

the total transportation cost. If a feasible solution involves exactly m+n-1
independent individual positive allocations, then it is known as non-degenerate
basic feasible solution.
 17

Determination of Feasible Solution of a TP:

1. North West Corner Rule

The procedure is as follows:

Step 1: Start with the cell (1,1) at the north-west corner of the transportation
table and allocate it maximum possible amount x11=min(a1,b1) so that either the
capacity of the origin O1 or the requirement for the destination D1 satisfies it or
both.

Step 2: If x11=a1, cross off the first row of the transportation table and decrease
b1 by a1 and go to step 3. If x11=b1, cross off the first column of the
transportation table and decrease a1 by b1 and go to step 3. If x11=a1= b1, cross
off either the first row or the first column(but not both).

Step 3: Repeat steps 1 and 2 for the resulting reduced transportation table until
all the requirements are satisfied.

Note:

The N-W corner rule of allocating values of the variables eliminates

either a row or column of the table from further consideration and the last
allocation eliminates both a row and column. Hence a feasible solution obtained
by this method has m+n-1 basic variables.

Problem:

Find an initial basic feasible solution to the following TP using North-West

Corner Rule.

1 2 3 4 ai
1 10 0 20 11 15

2 12 7 9 20 25

3 0 14 16 18 5
bj 5 15 15 10 45

Solution:

Step 1: First allocate in (1,1) cell. x11=min(15,5)=5

 18

1 2 3 4
ai
1 10 0 20 11 15

2 12 7 9 20 25

3 0 14 16 18 5
bj 5 15 15 1 45

Step 2: Second allocate in (1,2) cell. x12=min(10,15)=10

2 3 4
1 0 20 11 10

2 7 9 20 25

3 14 16 18 5
5 15 10
Step 3: Third allocate in (2,2) cell. X22=min(25,5)=5

2 3 4

2 7 9 20 25

3 14 16 18 5
5 15 10

Step 4: Fourth allocate in (2,3) cell. X23=min(20,15)=15

3 4

2 9 20 20

3 16 18 15
15 10

Step 5: Fifth allocate in (2,4) cell. X24=min(5,10)=5

 19

2 20 5

3 18 5
10

Step 6: Sixth allocate in (3,4) cell. X34=min(5,5)=5

4
3 18 5
5

The initialz basic feasible solution is given by 𝑥11 = 5, 𝑥12 = 10, 𝑥22 = 5,

𝑥23 = 15, 𝑥24 = 5, 𝑥34 = 5

1 2 3 4
1 5 10

10 0 20 11
2 5 15 5

12 7 9 20
3 5

0 14 16 18
Minimum transportation cost=5*10+10*0+5*7+15*9+5*20+5*18=Rs.410.

Exercise:

1. Using North West Corner rule find out the initial basic feasible solution
to the following TP.

1 2 3 4 ai
1 6 4 1 5 14

2 8 9 2 7 16

3 4 3 6 2 5
bj 6 10 15 4 35
 20

2. Least-Cost Method

The procedure is as follows:

Step 1: Determine the smallest cost in the cost matrix of the transportation table.
Let it be Cij. Allocate xij=min(ai,bj) in the (i,j) cell.

Step 2: If xij=ai cross off ith row of the transportation table and decrease bj by ai.
If xij=bj cross off jth column of the transportation table and decrease ai by bj. If
xij=ai= bj, cross off either the first row or the first column(but not both).

Step 4: Repeat steps 1 and 2 for the resulting table. Reduce transportation table
until all requirements are satisfied.

Problem:

Find an initial basic feasible solution to the following TP using Least Cost
Method.

1 2 3 4
1 10 0 20 11 15

2 12 7 9 20 25

3 0 14 16 18 5
5 15 15 10

Solution:

Step 1: 0 is the minimum element which appears in (1,2) and (3,1) cells. For the
First allocation in x12=min(15,15)=15. Cross out first row or first column.

1 2 3 4

1 10 0 20 11 15

2 12 7 9 20 25

3 0 14 16 18 5
5 15 15 1
 21

Step 2: Second allocate in (3,1) cell. X31=min(5,5)=5

1 2 3 4
2 12 7 9 20 25

3 0 14 16 18 5
5 0 15 10

Step 3: Third allocate in (2,2) cell. X22=min(25,0)=0

2 3 4

2 7 9 20 25

3 14 16 18 0
0 15 10

Step 4: Fourth allocate in (2,3) cell. X23=min(25,15)=15

3 4

2 9 20 25

3 16 18 5
15 10

Step 5: Fifth allocate in (3,4) cell. X34=min(0,10)=0

2 20 10

3 18 0
10

Step 6: Sixth allocate in (2,4) cell. X24=min(10,10)=10

 22

4
2 20 10
10

The initial basic feasible solution is given by 𝑥12 = 15, 𝑥31 = 5, 𝑥22 = 0,

𝑥23 = 15, 𝑥34 = 0, 𝑥24 = 10

1 2 3 4
1 15

10 0 20 11
2 0 15 10

12 7 9 20
3 5 0

0 14 16 18
Minimum transportation cost=15*9+10*20=Rs.335.

Exercise:

1. Using Least Cost Method find out the initial basic feasible solution to the
following TP.

A1 A2 A3

P1 50 30 220 1

P2 90 45 170 3

P3 250 250 50 4
4 2 2 8

3. Vogel’s Approximation Method(VAM)

The procedure is as given below.

Step 1: For each row of the transportation table find the smallest cost and next
to it. Determine the difference between them for each row. Write them within
brackets along the side of the table.
 23

Step 2: For each column do the same step as in step 1.

Step 3: Identify the row or column with the largest difference. Let the greatest
difference correspond to pi and let Cij be the smallest cost in the ith row. Allocate
xij=min(ai,bj) in that cell. If xij=ai cross off ith row of the transportation table and
decrease bj by ai. If xij=bj cross off jth column of the transportation table and
decrease ai by bj. If xij=ai= bj, cross off either the first row or the first
column(but not both).

Step 4: Repeat steps 1,2 and 3 for the resulting table. Reduce transportation
table until all requirements are satisfied.

Problem:

Find an initial basic feasible solution to the following TP using Vogel‟s

Approximation Method.

1 2 3 4
1 10 0 20 11 15

2 12 7 9 20 25

3 0 14 16 18 5
5 15 15 10

Solution:

Step 1:Here the maximum difference is 14 which corresponds to third row of

the table. For the First allocation in x31=min(5,5)=5. Cross out third column or
first column.

1 2 3 4

1 10 0 20 11 15 (10)

2 12 7 9 20 25 (2)

3 0 14 16 18 5 (14)
5 15 15 1
(10) (7) (7) (7)
 24

Step 2: Here the maximum difference is 11 which corresponds to first row.

Second allocate in (1,2) cell. X12=min(15,15)=15

2 3 4
1 0 20 11 15(11)

2 7 9 20 25(2)

3 14 16 18 0 (2)
15 15 10
(7) (7) (7)

Step 3: Here the maximum difference is 11 which corresponds to second

column. Third allocate in (2,3) cell. X23=min(25,15)=15

3 4

1 20 11 0 (9)

2 9 20 25(11)

3 16 18 0 (2)
15 10
(7) (7)

Step 4: Here the maximum difference is 7 which corresponds to fourth column.

Fourth allocate in (1,4) cell. X14=min(0,10)=0

1 11 0 (11)

2 20 10 (20)

3 18 0 (18)
10
(7)
 25

Step 5: Here the maximum difference is 2 which corresponds to fourth column.

Fifth allocate in (3,4) cell. X34=min(0,10)=0

2 20 10(11)

3 18 0 (18)
10
(2)

Step 6: Sixth allocate in (2,4) cell. X24=min(10,10)=10

4
2 20 10(11)
10
(11)

The initial basic feasible solution is given by 𝑥12 = 15, 𝑥23 = 15, 𝑥31 = 5,

𝑥14 = 0, 𝑥34 = 0, 𝑥24 = 10

1 2 3 4
1 15 0

10 0 20 11
2 15 10

12 7 9 20
3 5 0

0 14 16 18
Minimum transportation cost=15*9+10*20=Rs.335.

Exercise:Using VAM Method find out the initial basic feasible solution to the
following TP.

1 2 3 4

1 3 1 7 4 300

2 2 6 5 9 400
 26

3 8 3 3 2 500

250 350 400 200

Assignment Problems
Assignment problem is a special type of a LPP in which the
objective is to assign a number of origins to the equal number of distinct
destination at a minimum cost to find the optimum allocation of a number of
jobs to an equal number of persons.

Mathematical Formulation of AP
Let there be n machines and n jobs. A job i when processed
by machine j at a cost Cij.
1, 𝑖𝑓 𝑡𝑕𝑒 𝑖𝑡𝑕 𝑗𝑜𝑏 𝑖𝑠 𝑎𝑠𝑠𝑖𝑔𝑛𝑒𝑑 𝑡𝑜 𝑗𝑡𝑕 𝑚𝑎𝑐𝑕𝑖𝑛𝑒
𝑥𝑖𝑗 =
0, 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒.
𝑛 𝑛
The total assignment cost is 𝑖=1 𝑗 =1 𝐶𝑖𝑗 𝑋𝑖𝑗

Hence the AP can be stated mathematically as follows so as to

𝑛 𝑛
Minimize z= 𝑖=1 𝑗 =1 𝐶𝑖𝑗 𝑋𝑖𝑗

𝑛 𝑛
Subject to the constraints 𝑖=1 𝑋𝑖𝑗 = 𝑗 =1 𝑋𝑖𝑗 =1

Method of Solution
The procedure for the solution of an AP is known as
Hungarian Assignment Method is as follows.

Step 1: Subtract the minimum cost of each row of the cost matrix from all the
elements of the respective row and get the resultant matrix.

Step 2: Subtract the minimum cost of each column of the cost matrix from all
the elements of the respective column and get the resultant matrix. This gives a
new cost matrix.

Step 3: In the new cost matrix draw minimum possible number of horizontal
and vertical lines to cover all the zeroes. Here two cases arises.
 27

Case (i): If the number of lines=order of cost matrix, optimum assignment

has been obtained. Find out the optimum solution.

Case (ii): If the number of lines<order of cost matrix, we go to step-4.

Step 4: Determine the smallest cost in the new cost matrix not covered by the
lines. Subtract this cost from all the surviving elements(uncovered by lines) of
the new cost matrix and add the same to all those elements of the new cost
matrix that are lying at the intersection of horizontal and vertical lines.

Step 5: Repeat steps 3 and 4 until we get p=n(where p denote the number of
lines and n denote order of the cost matrix).

Step 6: For determining the optimum assignment consider only the zero
elements of the final table. Examine successively the rows(or)columns of the
matrix. To find out one with exactly one zero and encircled this zero and mark a
cross in the remaining zeroes of the row or column.

Step 6: The assignment schedule corresponding to these zeroes is the optimum

assignment schedule and find out the minimum assignment.

Maximisation in AP
In the objective function of the given AP is a maximizing time
we converted into minimizing time as in the case of TP by selecting the largest
element among all the elements of the cost matrix and then subtracting it from
all other elements of the cost matrix. This can equivalently be done by
multiplying all the entries of the cost matrix by -1 and then proceeding as in the
case of minimizing the AP.

Problem:

A shop has 4 workers and 4 works to be performed. The estimate of time(man

hours) each man will take to perform each project is given in the following
table. How should the works to be allotted so as to optimise the total work.

1 2 3 4

1 8 26 17 11

2 13 28 4 26

3 38 19 18 15
 28

4 19 26 24 10

Solution:

Step 1: Given matrix is,

1 2 3 4

1 8 26 17 11

2 13 28 4 26

3 38 19 18 15

4 19 26 24 10

Step 2: Here the given matrix is a square matrix. Hence the assignment problem
is a balanced problem. Subtracting minimum cost of each row of the cost matrix

we obtain the following matrix.

1 2 3 4

1 0 18 9 3

2 9 24 0 22

3 23 4 3 0

4 9 16 14 0

Step 3: Subtracting minimum cost of each column of the modified cost matrix
we obtain the following matrix.

1 2 3 4

1 0 14 9 3

2 9 20 0 22
 29

3 23 0 3 0

4 9 12 14 0

Step 4: Drawing the minimum possible number of horizontal and vertical lines.

1 2 3 4

1 0 14 9 3

2 9 20 0 22

3 23 0 3 0

4 9 12 14 0

Here the number of lines=4=order of cost matrix.Hence optimum assignment

obtained.

1 2 3 4
1 0

2 0

0
3
0
4

Optimum Schedule is 1->1, 2->3, 3->2, 4->4

Minimum assignment cost=8+4+19+10=41.

Exercise:

1. A company has 5 machines and 5 jobs to be done. The return in rupees of

assigning n-machines to n-job as follows. Assign the 5 jobs to the 5
machines so as to maximise the total profit.
 30

1 2 3 4 5

A 5 11 10 12 4

B 2 4 6 2 5

C 3 12 5 14 6

C 6 14 4 11 7

D 7 9 8 12 5

*************************************

UNIT III
SEQUENCING
Sequencing problem consists of selection of an appropriate order for a
finite number of jobs to be done on a finite number of machines/ service
facilities in some prescribed order so as to minimize the total idle time for the
machines. The problem may have some constraints placed on it, such as due
date for each job, processing order of each variable or variable processing
times.
Assumptions on sequencing problems
The following simplifying assumptions are usually made for sequencing
problems:
1) Each job, once started on a machine, is to be performed up to completion
on that machine.
2) The processing time on each machine is known. Such a time is
independent of the order of the jobs in which they are to be processed.
3) The time taken by each job in changing over from one machine to another
machine is negligible.
4) A job gets started on a machine as soon as the job and the machine are
both idle.
5) No machine may process more than one job simultaneously.
6) The order of completion of job has no significance (i.e) no job is to be
given priority.
 31

Processing n jobs in 2 machines:

Suppose there are n jobs 𝐽1 ,𝐽2 ,……….𝐽𝑛 which are to be processed in
two machines say 𝑀1 and 𝑀2 in order 𝑀1 𝑀2 (𝑀1 first and 𝑀2 next).
Let 𝑡𝑖𝑗 be the processing time for 𝑖 𝑡𝑕 job in 𝑗𝑡𝑕 machine. The list of jobs
along with their processing times can be summarized as in the following table
Jobs 𝐽1 𝐽2 ………….. 𝐽𝑛
Processing 𝑡11 𝑡12 …………... 𝑡1𝑛
time in 𝑀1

Processing 𝑡21 𝑡22 …………… 𝑡2𝑛

time in 𝑀2

A job is assigned to machine 𝑀1 first and after it has been completely

processed in machine 𝑀1 it is assigned to the machine 𝑀2 . If the machine 𝑀2
is not free at any moment for processing a particular job then that job has to
wait in a waiting line for its turn on the machine 𝑀2 . In other words passing is
not allowed. Hence machine 𝑀1 will always be busy and will process the 𝑛 jobs
one by one. After processing all the 𝑛 jobs, the machine 𝑀1 remains idle until
all the 𝑛 jobs are completed in the machine 𝑀2 . However 𝑀2 may remain idle
after the completion of some of the 𝑚 jobs and before starting the next job. The
sequencing problem is to minimize the total idle time of the second machine 𝑀2 .
Let 𝑥2𝑗 be the time for which the machine 𝑀2 remains idle after
completing the (𝑗 − 1)𝑡𝑕 job and before starting 𝑗𝑡𝑕 job.
Hence the total idle time for machine 𝑀2 is 𝒏𝒋=𝟏 𝒙𝟐𝒋 . Thus the sequencing
problem is to minimize 𝒏𝒋=𝟏 𝒙𝟐𝒋 . The total elapsed time 𝑇 is given by
𝒏 𝒏
T=Processing time + idle time. (i.e) T = 𝒋=𝟏 𝒕𝟐𝒋 + 𝒋=𝟏 𝒙𝟐𝒋 .
𝑛
Here some of the 𝑥2𝑗 𝑠 may be zeros. We observe that 𝑗 =1 𝑡2𝑗 is constant.
𝑛
Hence minimizing 𝑇 is equivalent to minimizing 𝑗 =1 𝑥2𝑗 .

Algorithm to find the optimum sequence for 𝒏 jobs in 2 machines

Step 1. List the jobs along with their processing times in a table as given
above.
 32

Step 2. Find the minimum (𝑡1𝑗 , 𝑡2𝑗 ) for all 𝑗 = 1,2, … . . , 𝑛.

Step 3. If the smallest processing time is for the first machine 𝑀1 then
place the corresponding job in the first available position in the sequence . If
it is for the second machine 𝑀2 then place the corresponding job in the last
available position in the sequence.
Step 4. If there is a tie in the minimum of all the processing times then
there arises three cases.
Case i. Minimum among all processing times is same for the two
machines (i.e) minimum 𝑡1𝑗 , 𝑡2𝑗 ) = 𝑡1𝑟 = 𝑡2𝑠 then place the 𝑟 𝑡𝑕 job in the first
available position in the sequence and the 𝑠 𝑡𝑕 job in the last available position
in the sequence.
Case ii. If the tie is for the minimum among the processing times
𝑡𝑖𝑗 on machine 𝑀1 only, then place the jobs arbitrarily, one after the other, in
the last available positions in the sequence.
Case iii. If the tie is for the minimum among the processing times
𝑡2𝑗 on machine 𝑀2 only, then place the jobs arbitrarily, one after the other, in
the last available positions in the sequence.
Step 5. Remove the assigned jobs from the table. If the table becomes
empty the optimum sequencing is got and stop the procedure. Other wise go to
step 2.
Note. From the optimal sequencing, calculate the total elapsed time, idle time
on machine 𝑀1 and idle time on machine 𝑀2 .
(a) Total elapsed time = Total time between the starting the first job of the
optimum sequence on machine 𝑀1 and completing the last job on
machine 𝑀2 .
(b) Idle time for 𝑴𝟏 = Total elapsed time –(Time when the last job in the
sequence finishes on 𝑀1 )
(c) Idle time for 𝑴𝟐 = Time at which the first job in the sequence finishes on
𝑀1 + 𝑛𝑗=2(time when the 𝑗𝑡𝑕 job in the sequence starts on 𝑀2 )−(time
when the (𝑗 − 1))𝑡𝑕 job in the sequence finishes on 𝑀2 ).

Problem:
Determine the optimum sequence for the following sequencing problem in
which 5 jobs are done in 2 machines 𝑀1 and 𝑀2 in the order 𝑀1 𝑀2 .
Processing times are given in hours in the following table.
 33

𝑀1 𝑀2
𝑱𝟏 10 4
𝑱𝟐 2 12
𝑱𝟑 18 14
𝑱𝟒 6 16
𝑱𝟓 20 6
Solution:
The minimum processing time among all the jobs in the 2 machines 𝑀1 and
𝑀2 is 2 (which is marked in under lined bold face), which correspond to the job
𝐽2 on machine 𝑀1 . Thus the job 𝐽2 is placed (sequenced) in the first position in
the sequence .
𝐽2

We are now left with 4 jobs and their processing times are given below.
Machines 𝑴𝟏 𝑴𝟐
Jobs
𝑱𝟏 10 4
𝑱𝟑 18 14

𝑱𝟒 6 16

𝑱𝟓 20 6
Among those four
jobs the minimum
processing time between the two machines is 4 which corresponds to the job 𝐽1
on machine 𝑀2 . Hence the job 𝐽1 is placed in the last position in the sequence
shown below.
𝐽2 𝐽1

The remaining 3 jobs and their processing times are given below.
Machines 𝑴𝟏 𝑴𝟐
Jobs
𝑱𝟑 18 14
 34

𝑱𝟒 6 16

𝑱𝟓 20 6

The minimum processing time among these jobs is 6 which corresponds to the
job 𝐽4 on machine 𝑀1 and job 𝐽5 on machine 𝑀2 . Hence 𝐽4 is placed in the first
available position in the sequence as shown below.
𝐽2 𝐽4 𝐽5 𝐽1

Finally, the job 𝑱𝟑 is placed as shown below.

𝐽2 𝐽4 𝐽3 𝐽5 𝐽1

Thus 𝐽2 ⟹ 𝐽4 ⟹ 𝐽3 ⟹ 𝐽5 ⟹ 𝐽1 is the optimal sequence of the five jobs on the

two machines.
Sequencing n Jobs on 3 Machines
Processing n Jobs in m Machines
Let there be n jobs 𝐽1 , 𝐽2 ,……𝐽𝑛 each of which is to be processed in m
machines say 𝑀1 , 𝑀2 ,……𝑀𝑚 in the order 𝑀1 𝑀2 ……𝑀𝑚 . The list of job
numbers 1,2,…..,n with their processing time is given in the following table.
Job numbers
Machines
1 2 3 ……… n
𝑀1 𝑡11 𝑡12 𝑡13 ……… 𝑡1𝑛
𝑀2 𝑡21 𝑡22 𝑡23 ……… 𝑡2𝑛
𝑀3 𝑡31 𝑡32 𝑡33 ……… 𝑡3𝑛
……… ……… ……… ……… ……… ………
𝑀𝑚 𝑡𝑚 1 𝑡𝑚 2 𝑡𝑚 3 ……… 𝑡𝑚𝑛
An optimum solution to this problem can be obtained if either or both of the
following conditions are satisfied.

(a) 𝑚𝑖𝑛 𝑡1𝑗 ≥ 𝑚𝑎𝑥 𝑡𝑖𝑗 for 𝑖 = 2,3, … . . , 𝑘 − 1

(b) 𝑚𝑖𝑛 𝑡𝑚𝑗 ≥ 𝑚𝑎𝑥 𝑡𝑖𝑗 for 𝑖 = 2,3, … … , 𝑘 − 1
 35

If the above condition is satisfied then the problem can be converted to an

equivalent two machines and n jobs problem.
Algorithm for optimum sequence for n jobs in m machines
The iterative procedure for determining the optimal sequence for n
jobs in m machines is given below.
Step 1. Find 𝑚𝑖𝑛 𝑡𝑖𝑗 and 𝑚𝑖𝑛 𝑡𝑚𝑗 . Also find the maximum of each of 𝑡2𝑗 ,𝑡3𝑗
,….𝑡𝑘−1 𝑗 for all 𝑗 = 1,2, … . , 𝑛 .
Step 2. Check the following

(a) 𝑚𝑖𝑛 𝑡1𝑗 ≥ 𝑚𝑎𝑥 𝑡𝑖𝑗 for 𝑖 = 2,3, … . . , 𝑘 − 1

(b) 𝑚𝑖𝑛 𝑡𝑚𝑗 ≥ 𝑚𝑎𝑥 𝑡𝑖𝑗 for 𝑖 = 2,3, … … , 𝑘 − 1

Step 3. If both the inequalities of step 2 are not satisfied this method fails
otherwise go to step 4.
Step 4. Convert the m machines problem into a two machine problem by
introducing two fictious machines H and K such that
𝑡𝐻𝑗 = 𝑡1𝑗 + 𝑡2𝑗 + ⋯ + 𝑡𝑘−1 𝑗 and

𝑡𝐾𝑗 = 𝑡2𝑗 + 𝑡3𝑗 + ⋯ + 𝑡𝑘𝑗

Where 𝑡𝐻𝑗 is the processing time of 𝑗𝑡𝑕 job in machine H and 𝑡𝐾𝑗 is the
processing time of 𝑗𝑡𝑕 job in machine K.
Step 5. Determine the optimal sequence for n jobs and 2 machines sequencing
problem .
Problem :
Find the sequence that minimizes the total elapsed time required to
complete the following six jobs on three machines 𝑀1 , 𝑀2 and 𝑀3 in the order
𝑀1 ⟹ 𝑀2 ⟹ 𝑀3

Jobs⟹ 𝑱𝟏 𝑱𝟐 𝑱𝟑 𝑱𝟒 𝑱𝟓 𝑱𝟔

𝑴𝟏 3 12 5 2 9 11
Machines⟹
𝑴𝟐 8 6 4 6 3 1

𝑴𝟑 13 14 9 12 8 13
 36

Solution:
We are given 6 jobs each of which is to be processed in 3 machines
𝑴𝟏 , 𝑴𝟐 , 𝑴𝟑 in the order 𝑴𝟏 𝑴𝟐 𝑴𝟑 . Hence n=6 and m=3. Hence the given
problem becomes 6 jobs in 3 machines problem. For optimum sequence the
required condition is either or both of the following conditions must be true.

(a) 𝑚𝑖𝑛 𝑡1𝑗 ≥ 𝑚𝑎𝑥 𝑡2𝑗 𝑗 = 2, … . . ,5

(b) 𝑚𝑖𝑛 𝑡3𝑗 ≥ 𝑚𝑎𝑥 𝑡2𝑗 𝑗 = 2, … . . ,5

𝑴𝟏 𝑴𝟐 𝑴𝟑
Minimum time 2 1 8
Maximum time 12 8 14
𝑀𝑖𝑛 𝑡1𝑗 = 2 and 𝑛 𝑡3𝑗 = 8 ; 𝑀𝑎𝑥 𝑡2𝑗 = 8
Clearly 𝑀𝑖𝑛 𝑡3𝑗 = 𝑀𝑎𝑥 𝑡2𝑗 = 8 . Hence the condition to reduce 3 machines
problem to 2 maxhines problem is satisfied. Hence the given problem can be
converted to an equivalent two machines for six jobs problem.
Let H and K be two fictious machines such that the processing times are
got from the following relations
𝐻𝑖 = 𝑀𝑖1 + 𝑀𝑖2 𝑖. 𝑒. 𝐻𝑖 = 𝑡𝐻𝑗 = 𝑡1𝑗 + 𝑡2𝑗
𝐾𝑖 = 𝑀𝑖2 + 𝑀𝑖3 𝑖. 𝑒. 𝐾𝑖 = 𝑡𝐾𝑗 = 𝑡2𝑗 + 𝑡3𝑗

Where 1 ≤ 𝑖 ≤ 6.
Hence we have the processing table for the two machines H and K.
Jobs⟹ 𝑱𝟏 𝑱𝟐 𝑱𝟑 𝑱𝟒 𝑱𝟓 𝑱𝟔

H 11 18 9 8 12 12
Machines⟹ (𝑴𝟏
+ 𝑴𝟐 )
K 21 20 13 18 11 14
𝑴𝟐 + 𝑴 𝟑 )
 37

Now we find the optimum sequence of the 6 jobs in two machines H and K as
usual in the following tables.
𝐽4

𝐽4 𝐽3

𝐽4 𝐽3 𝐽1 𝐽5

𝐽4 𝐽3 𝐽1 𝐽6 𝐽5

𝐽4 𝐽3 𝐽1 𝐽6 𝐽2 𝐽5

Hence the optimal sequencing is 𝐽4 ⟹ 𝐽3 ⟹ 𝐽1 ⟹ 𝐽6 ⟹ 𝐽2 ⟹ 𝐽5 .

Table to find minimum elapsed time

Job 𝑴𝟏 𝑴𝟐 𝑴3 Idle Idle

seq Time in Time in Time in time time
Time out Time Time 𝑀2 𝑀3
out out
𝑱𝟒 0 2 2 8 8 20 2 8
𝑱𝟑 2 7 8 12 20 29 - 0
𝑱𝟏 7 10 12 20 29 42 - 0
𝑱𝟔 10 21 21 22 42 55 1 0
𝑱𝟐 21 33 33 39 55 69 11 0
𝑱𝟓 33 42 42 45 69 77 3 0
Total 17 8

Minimum total elapsed time = 77 Hrs

Idle time for 𝑀1 =77-42=35
Idle time for 𝑀2 =17+(77-45)=17+32=49
 38

Idle time for 𝑀3 =8.

Processing 2 jobs in m machines
Suppose there are 2 jobs 𝐽1 , 𝐽2 which are processed in m machines 𝑀1 ,
𝑀2 , … 𝑀𝑚 in two different orders. The ordering of each of the two jobs in m
machines in known in advance. The ordering may not be the same for both the
jobs. Also the processing times for each of the two jobs are also known. Let 𝑡𝑖𝑗
be the processing time of job 𝐽𝑖 in machine 𝑀𝑗 where i=1,2 and j=1,2,…,m. The
optimal sequence can be obtained by the following graphical method. The
solution procedure is summarized in the following steps.
Step 1. Draw two perpendicular lines, horizontal and vertical. Horizontal line
represents the processing time for 𝐽1 while the job 𝐽2 remains idle. Vertical line
represents the processing time for job 𝐽2 while the job 𝐽1 remains idle.
Step 2. Mark the processing time for jobs 𝐽1 and 𝐽2 on the horizontal and
vertical lines respectively according to the given order of machines.
Step 3. Construct various blocks starting from the orgin ( starting point) by
pairing the same machines until the end point.
Step 4. Draw the line starting from the origin to end point by moving
horizontally, vertically and diagonally along a line which makes an angle of
45° with the horizontal line (base). The horizontal segment of this line indicates
that first job is under process while the second job is idle. Similarly, the vertical
segment of this line indicates that the second job is under process while the first
job is idle. The diagonal segment of the line shows that both the jobs are under
process simultaneously.
Step 5. An optimum path is one that minimizes the idle time for both the jobs.
Thus we must choose the path on which diagonal movement is maximum.
Step 6. The total elapsed time is obtained by adding the idle time for either job
to the processing time for that job.
The total elapsed time =Processing time for job 𝑱𝟏 + idle time for 𝑱𝟏
Or
The total elapsed time =Processing time for job 𝑱𝟐 + idle time for 𝑱𝟐
Problem:
Use graphical method to find the minimum total elapsed time needed to process
the following two jobs on four machines A,B,C,D given the processing times
and the sequences.
 39

Job 1 Sequence: A B C D
Time in Hrs: 2 3 5 2
Job 2 Sequence: D C A B
Time in Hrs: 6 2 3 1
Solution:
The graph representing the data is given below. The rectangular blocks in the
graph are obtained by pairing the same machines. The path from the origin to
the end point, which uses maximum diagonal segments, is shown in the figure.
For other details the figure is self- explanatory.

The minimum total elapsed time = Processing time for Job 1+ idle time for Job
1
=12+3=15 hours (or)
The minimum total elapsed time = Processing time for Job 2+ idle time for Job
2
=12+3=15 hours

Queueing Theory
In every life, it is seen that a number of people arrive at a cinema ticket
window. If the people arrive “too frequently” they will have to wait for getting their
tickets or sometimes do without it. Under such circumstances the only alternative is to
form a queue, called the waiting line, in order to maintain a proper discipline.
Occasionally, it also happens that the person issuing tickets wil have to wait,(i.e,
remains idle), until additional people arrive. Here the arriving people are called the
customers and the person issuing the tickets is called a server.
 40

Arrivals Service Service Departures

Centre 1 Centre 2
Service centre 3

Service Service Centre Departures

Centre 4 5
Queuing theory is concerned with the statistical description of the behavior of queues
with finding, e.g., the probability distributed of the number in the queue from which
the mean and variance of queue length an probability distribution of waiting time for
a customer, or the distribution of a server‟s busy periods can be found. In operational
research problems involving queues, investigators must measure the existing system
to make an objective assessment of its characteristics, and must determine how
changes may be made to the system, what effects of various system‟s kinds of
changes in the system‟s characteristics would be, and whether , in the light of the
queuing system under study must be constructed in this kind of analysis and the
results of queuing theory are required to obtain the characteristics of the model and to
assess the effects, such as the addition of an extra server or a reduction in mean
service time.

QUEUEING SYSTEM
A queuing system can be completely described by

(a) The input (or arrival pattern),

(b) The service mechanism ( or service pattern)
(c) The „ queue discipline „ and
(d) Customer‟s behavior.

Queuing System

Queue or waiting time Servicing

Station

Arriving units system Departures

 41

(a ) The input ( or arrival pattern ). The input describes the way in which the
customers arrive and join the system. Generally, the customers arrive in a more or less
random fashion which is not worth making the prediction. Thus, the arrival patteren
can best be described in terms of probabilities and consequently the probability
distribution for inter – arrival times ( the time between two successive arrivals ) or the
distribution of number customers arriving in unit time must be defined.

(b) The service mechanism ( or service pattern). It is specified when it is

known how many customers can be served at a time, what the statistical distribution
of service time is, and when service is available. It is true in most situations that
service time is a random variable with the same distribution for all arrivals, but cases
occur where there are clearly two or more classes of customers, (e.g., machines
waiting repair) , each with a different service time distribution. Service time which are
important in practice are the „ negative exponential distribution‟ and the related „
Erlang ( gamma) distribution „. Queues with the negative exponential service time
distribution are studied in the following sections.

(c) The queue discipline. The queue discipline is the rule determining the
formation of the queue, the manner of the customer‟s behavior while waiting, and the
manner in which they are chosen for service. The simplest discipline is “ First come,
first served”, according to which the customers are served in the order of their arrival.
For example, such type of queue discipline is observed at a ration shop, at cinema
ticket windows, at inrailway stations, etc., If the order is reversed we have the “ last
come, first served “ discipline, as in the case of a big go down the items which come
last or taken out first. An extremely difficult queue discipline to handle might be “
service in random order “ or “ might is right “.

Notations:
FIFO ⟹ First In, First Out

FCFS ⟹ First come, First Served

LIFO ⟹ Last In, First Out

SIRO ⟹Service in Random Order

FILO⟹ First In, Last Out.

(d) Customer’s behavior. The customers generally behave in four ways:

(i) Balking . A customer may leave the queue because the queue is too long
and he has no time to wait, or there is not sufficient waiting space.
 42

(ii) Reneging . This occurs when a waiting customer leaves the queue due to
impatience.
(iii) Priorities. In certain applications some customers are served before others
regardless of their order of arrival. There customers have priority over
others.
(iv) Jockeying. Customers may jockey from one waiting line to another. It may
be seen that this occurs in the supermarket.

Transient and steady states

Queuing theory analysis involves the study of a system‟s behavior over time. A
system is said to be in “ transient state “ when its operating characteristics ( behavior )
are dependent on time . This usually occurs at the early stages of the operations of the
system where its behavior is still dependent on the initial conditions. How ever, since
we are mostly interested in the “ long run “ behavior been paid toward “ steady state” .

A steady state conditions is said to prevail when the behavior of the system
becomes independent of time. Let 𝑃𝑛 (𝑡) denote the probability that there are n units
in the system at time t. In fact, the change of 𝑃𝑛 (𝑡) with respect to t is described by
𝑑𝑃𝑛 𝑡
the derivative 𝑜𝑟𝑃𝑛 ′ (𝑡). Then the queuing system is said to become „state‟
𝑑𝑡
eventually, in the sense that the probability 𝑃𝑛 (𝑡) is independent of time, that is,
remains the same as time passes (𝑡 → ∞) . Mathematically, in steady state

lim 𝑃𝑛 𝑡 = 𝑃𝑛 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡𝑜𝑓𝑡
𝑛→∞

𝑑𝑃𝑛 𝑡 𝑑𝑃𝑛
⟹lim𝑡→∞ =
𝑑𝑡 𝑑𝑡

⟹log 𝑡→∞ 𝑃𝑛 ′ 𝑡 = 0.

In some situations, if the arrival rate of the system is larger than its service rate, a
steady state cannot be reached regardless of the length of the elapsed time. In, fact, in
this case the queue length will increase with time and theoretically it could bulid up to
infinity. Such case is called the “ explosive state".

In this chapter, only the steady state analysis will be considered. We shall not
treat the „ transient‟ and „ explosive‟ states.

A list of symbols
 43

Unless otherwise stated, the following symbols and terminology will be used
hence forth in connection with the queueing models. The reader is reminded that a
queueing that a queueing system is defined to include the queue and the service station
both.

n=number of units in the system.

𝑃𝑛 (𝑡)= transient state probability that exactly n calling units are in the queueing
system at time t

𝐸𝑛 = the state in which there are n calling units in the system.

𝑃𝑛 = steady state probability of having n units in the system.

𝜆𝑛 = mean arrival rate ( expected number of arrivals per unit time) of customers ( when
n units are present in the system)

𝜇𝑛 =mean service rate ( expected number of customers served per unit time when
there are n units in the system )

λ = mean arrival rate when 𝜆𝑛 is constant for all n

𝜇 = mean service rate when 𝜇𝑛 is constant for all 𝑛 ≥ 1

s= number of parallel service stations.

𝜆
𝜌 = = traffic intensity ( or utilization factors) for servers facility, that is, the expected
𝜇𝑠
fraction of time the servers are busy

𝜑𝑇 𝑛 = probability of n services in time T, given that servicing is going on throughout

T.

Line length ( or queue size)= number of customers in the queuing system.

Queue length= line length ( or queue size) – (number of units being served)

𝜑(𝑤) = probability density function (p.d.f.) of waiting time in the system.

𝐿𝑠 =expected line length, i.e., expected number of customers in the system

𝐿𝑞 =expected queue length, i.e., expected number of customers in the queue

𝑊𝑠 = expected waiting time per customer in the system

𝑊𝑞 = expected waiting time per customer in the queue

(W|W>0) = expected waiting time of a customer who has to wait

 44

(L|L>0)= expected length of non-empty queues. i.e., expected number of customers in

the queue when there is a queue

P(W>0)= probability of a customer having to wait for service

𝑛 𝑛! 𝑛 𝑛−1 …(𝑛−𝑟+1)
𝑟
= denotes the binomial coefficient 𝑛𝐶𝑟 = = for r and n
𝑟!(𝑛−𝑟)! 𝑟!
non- negative integers (𝑟 ≤ 𝑛).

DISTRIBUTIONS IN QUEUEING SYSTEMS

In particular, we show that, if arrivals are „ completely random‟, the number of

arrivals in unit time has a Poisson distribution, and the intervals between successive
arrivals are distributed negative exponentially.

Distribution of Arrivals „The Poisson Process‟ (Pure Birth Process).

Arrival Distribution Theorem.

If the arrivals are completely random, then the probability distribution of number of
arrivals in a fixed time – interval follows a Poisson distribution.

Proof: In order to derive the arrival distribution in queue, we make the following three
assumptions

1. Assume that there are n units in the at time t, and the probability that exactly
one arrival will occur during small time interval ∆𝑡 be given by 𝜆∆𝑡 + 𝑜 ∆𝑡 ,
where λ is the arrival rate independent of t and O(∆𝑡) includes the terms of
higher order of ∆𝑡.
2. Further assume that the time ∆𝑡 is so small that the probability of more than
one arrival in time ∆𝑡 is 𝑂(∆𝑡)2 , i.e., almost zero.
3. The number of arrivals in non- overlapping intervals are statistically
independent, i.e., the process has independent increments.

We now wish to determine the probability of n arrivals in a time interval of length

t, denoted by 𝑃𝑛(𝑡) . Clearly, n will be an integer greater than or equal to zero. To
do so, we shall first develop the differential equations governing the process in two
different situations.

Case 1. When n >0

For n>0 ,there may be two mutually exclusive ways of having n units at time
𝑡 + ∆𝑡
 45

(i) There are n units in the system at time t and no arrival takes place during
time interval ∆𝑡. Hence , there will be n units at time 𝑡 + ∆𝑡 also.

Therefore, the probability of these two combined events will be = prob. Of n units
at time t x prob. of no arrival during ∆𝑡 = 𝑃𝑛 𝑡 . (1 − 𝜆∆𝑡)

(ii) Alternately, there are (n-1) units in the system at time t, and one arrival
takes place during ∆𝑡. Hence there will remain n units in the system at time
t+∆𝑡.

Therefore, the probability of these two combined events will be = prob.of (n-1)
units at time t x prob.of one arrival in time ∆𝑡

=𝑃𝑛 −1 𝑡 . 𝜆. ∆𝑡

We get the probability of n arrivals at time t+∆𝑡

𝑃𝑛 𝑡 + ∆𝑡 = 𝑃𝑛 𝑡 1 − 𝜆∆𝑡 + 𝑃𝑛−1 (𝑡)𝜆∆𝑡

Case 2. When n=0

𝑃0 𝑡 + ∆𝑡 = 𝑃0 𝑡 1 − 𝜆∆𝑡

Dividing both sides by ∆𝑡 and then taking limit as ∆𝑡 → 0

𝑃𝑛 𝑡 + ∆𝑡 − 𝑃𝑛 𝑡
lim = −𝜆𝑃𝑛 𝑡 + 𝑃𝑛 −1 (𝑡)
∆𝑡→0 ∆𝑡
𝑃0 𝑡 + ∆𝑡 − 𝑃0 𝑡
lim = −𝜆𝑃0 𝑡
∆𝑡→0 ∆𝑡
Pn′ t = −𝜆𝑃𝑛 𝑡 + 𝜆𝑃𝑛−1 (𝑡) n>0

P0′ t = −𝜆𝑃0 𝑡 n=0

∞

P 𝑧, 𝑡 = 𝑃𝑛 (𝑡)𝑧 𝑛
𝑛=0

P′ 𝑧, 𝑡 = 𝑃𝑛 ′(𝑡)𝑧 𝑛
𝑛=0

Multiplying both sides of (A) by 𝑧 𝑛 and taking summation for n=1,2,…..,∞ we get
∞ 𝑛 ∞ 𝑛 ∞ 𝑛
𝑛=0 𝑃𝑛 ′(𝑡)𝑧 =-λ 𝑛=1 𝑃𝑛 (𝑡)𝑧 +λ 𝑛=1 𝑃𝑛−1 (𝑡)𝑧
 46

P′ 𝑧, 𝑡
= 𝜆(𝑧 − 1)
P 𝑧, 𝑡

Log P(z,t)=λ(z-1)t+E

To determine E, we put t=0 to get

Log P(z,0)=𝐸 ′
∞ 𝑛
P(z,0)= 𝑛=0 𝑧 𝑃𝑛 0

=1+0

=1

P(z,t)=𝑒 𝜆 𝑧−1 𝑡

Now , 𝑃𝑛 (𝑡) can be defined as

1 𝑑 𝑛 𝑃 𝑧,𝑡
𝑃𝑛 𝑡 = t=0
𝑛! 𝑑𝑧𝑛

(𝜆𝑡 )𝑛 𝑒 −𝜆𝑡
In general, 𝑃𝑛 𝑡 =
𝑛!

Classification of queueing models

I- Probabilistic queueing models

Model I. (Erlang Model). This model is symbolically represented by

(M|M|I):( ∞ 𝐹𝐶𝐹𝑆 . this denotes poisson arrival ( exponential inter -arrival)
poisson departure (exponential service time) single server , infinite capacity and “
first come , first served “ service discipline.

Model II. (General Erlang Model) . Although this model is also represented by
(M|M|I):(∞ 𝐹𝐶𝐹𝑆 , but this is a general queueing model in which the rate of
arrival and service depend on the length n of the line.

Model III this model represented by (M|M|I):(𝑁 𝐹𝐶𝐹𝑆 . In this model capacity of
the system is limited (finite) , say n. Obviously , the number of arrivals will not
exceed the number N in any case .

Model IV. This model is represented by (M|M|S):(∞ 𝐹𝐶𝐹𝑆 , in which the number
of stations is s in parallel .

Model V. This model is represented by (M|𝐸𝑘 |I):(∞ 𝐹𝐶𝐹𝑆 , that is , poisson

arrivals , Erlangian service time for k phases and a single server.
 47

Model VI. (Machine servicing model) .this model is represented by

(M|M|R):(𝐾 𝐺𝐷 , 𝑘 > 𝑅, that is , poission arrivals , exponential service time , R
repairmen , and K machines in the system and general queue discipline.

Model VII. Power -supply model. Models.

Model VIII. Economic cost profit models.

Model IX. (M|G|I):(∞ 𝐺𝐷 , where G is the general output distribution , and GD

represents a general service discipline

II- Mixed Queueing Model

Model X. (M|D|I):(∞ 𝐹𝐶𝐹𝑆 , where D stands for deterministic service
time.
III- Deterministic Queueing Model
Model XI. (D|D|I):(𝐾 − 1 𝐹𝐶𝐹𝑆 , where
D→ Deterministic arrivals , i.e., interarrival time distribution is constant or
regular.
D→ Deterministic service time distribution.

Solution of queueing models

(a) To obtain system of steady state equations goverening the queue

(b) To solve this equations for finding out the probability distribution of
queue length.
(c) To obtain probability density function for waiting time distribution
(d) To find the busy period distribution
(e) To derive formula for 𝐿𝑠 , 𝐿𝑞 , 𝐿 𝐿 > 0 , 𝑊𝑠 , 𝑊𝑞 , 𝑊 𝑊 > 0 , and var (n)
etc.,
(f) Also , to obtain the probability of arrival during the service time of any
Customer

To solve the system of difference equations.

To solve difference equations:

0=-(λ+𝜇)𝑃𝑛 + 𝜆𝑃𝑛+1 + 𝜇𝑃𝑛 +1 𝑖𝑓 𝑛 > 0

0=-λ𝑃𝑜 + 𝜇𝑃1, if n=0

 48

We use the technique of successive substituation

Since 𝑃0 = 𝑃0

𝜆
𝑃1 = 𝑃
𝜇 0
𝜆
𝑃2 = 𝑃
𝜇 1

𝜆
𝑃𝑛 = 𝑃
𝜇 𝑛 −1
∞
Now using the fact that 𝑛=0 𝑃𝑛 =1
𝜆 𝜆
𝑃0 [1 + + ( )2 ……]=1
𝜇 𝜇

1
𝑃0 =1
1−𝜆 𝜇

=1-p

Now substituting the value of 𝑃0 from

𝜆 𝜆
𝑃𝑛 = ( )2 (1- )
𝜇 𝜇

=𝜌𝑛 (1-𝜌)

Measure of Model I:

To find expected number of units in the system 𝐿𝑠

By definition of expected value,

∞
𝐿𝑠 = 𝑛𝑃𝑛
𝑛 =1
 49

∞ 𝜆 𝑛 𝜆
= 𝑛=1 𝑛(𝜇 ) (1- )
𝜇

𝜆 𝜆 ∞ 𝜆 𝑛
= (1- ) 𝑛=1 𝑛(𝜇 ) −1
𝜇 𝜇

𝜆 𝜆 1
=(1- ) [ ]
𝜇 𝜇 1−𝜆 𝜇

𝜆
𝜇
𝜌 𝜆
𝐿𝑠 = = , 𝑤𝑕𝑒𝑟𝑒 𝜌 = < 1)
(1 − 𝜆 𝜇) 1−𝜌 𝜇

Which is the required formula

To find expected queue length 𝐿𝑞

Since there are (n-1) units in the queue excluding one being serviced,
∞
𝐿𝑞 = (𝑛−1)𝑃𝑛
𝑛 =1

∞ ∞
= 𝑛=1 𝑛𝑃𝑛 - 𝑛=1 𝑃𝑛

∞ ∞
= 𝑛=1 𝑛𝑃𝑛 -[ 𝑛=1 𝑛𝑃𝑛 − 𝑃0 ]

=𝐿𝑠 = [1 − 𝑃0 ]

Substituting the value of 𝑃0

𝜆
𝐿𝑞 = 𝐿𝑠 − 1 + (1 − )
𝜇

𝜆 𝜌2 𝜆 𝜌
𝜇
𝐿𝑞 = 𝐿𝑠 − = where 𝐿𝑠 = =
𝜇 1−𝜌 (1−𝜆 𝜇 ) 1−𝜌

To find expected waiting time in the system 𝑊𝑠

Since expected waiting time in the system

= Expected waiting time in queue + expected time, 𝑊𝑠 = 𝑊𝑞 + 1 𝜇

𝜆 1 1
𝑊𝑠 = + =
𝜇 𝜇−𝜆 𝜇 𝜇−𝜆

To find variance of queue length

By definition
 50

∞ 2 ∞ 2
Var. {n}= 𝑛=1 𝑛 𝑃𝑛 −[ 𝑛=1 𝑛𝑃𝑛 ]

∞ 2 2
= 𝑛=1 𝑛 𝑃𝑛 -[𝐿𝑠 ]

∞ 2 𝜌
= 𝑛=1 𝑛 1 − 𝜌 𝜌𝑛 − [ ]2
1−𝜌

𝜌
= (1 − 𝜌)2

Inter- Relationship Between 𝑳𝒔 , 𝑳𝒒 , 𝑾𝒔 , 𝑾𝒒

It can be proved under general conditions of arrival, departure, and discipline that the
formulae

𝐿𝑠 = 𝜆𝑊𝑠,

And 𝐿𝑞 = 𝜆𝑊𝑞

Will hold in general. These formulae act As key poits in establishing the strong
relationship between 𝐿𝑠 , 𝐿𝑞 , 𝑊𝑠 , 𝑊𝑞 which can be found as follows

By definition,

𝑊𝑞 = 𝑊𝑠 − 1 𝜇

Thus, multiplying both sides by and substituting the values

𝐿𝑞 = 𝐿𝑠 − 𝜆 𝜇

This means that one of the four expected values should immediately yield the
remaining three values.

MODEL IV (A) : (M|M|S): (∞ 𝐹𝐶𝐹𝑆

In this more realistic queueing system the customers arrive in a poission fashion with
mean arrival λ. There are s number of counters arranged in parallel, and a customer
can go tc any of the free counters for his service, where the service time at each
counter is identical and follows the same exponentioal distribution law. The mean
service rate per busy service is 𝜇. Therefore over all service rate, when there are n
units in the system may be obtained in the following two situations.

(i) If 𝑛 ≤ 𝑠, all the customers may be served simulataneously. There will be no

queue , (s-n) number of servers may remain idle, and then
 51

𝜇𝑛 = 𝑛𝜇, 𝑛 = 0,1,2, … . , 𝑠;

(𝑖𝑖) If 𝑛 ≥ 𝑠, all the servers are busy , maximum number of customers waiting in
queue will be (n-s), then 𝜇𝑛 = 𝑛𝜇

𝑀𝑂𝐷𝐸𝐿 IV (B) : (M|M|S): (𝑁 𝐹𝐶𝐹𝑆

In model IV(A) , if the maximum number in the system is limited to N, then as in

model III,

𝜆 𝑓𝑜𝑟 0 ≤ 𝑛 ≤ 𝑁
𝜆𝑛 =
0 𝑓𝑜𝑟 𝑛 > 𝑁

And
𝑛𝜇 𝑓𝑜𝑟 0 ≤ 𝑛 ≤ 𝑠
𝜇𝑛 =
𝑠𝜇 𝑓𝑜𝑟 𝑠 ≤ 𝑛 ≤ 𝑁

Virtually the same relationships hold between 𝑃𝑛 𝑎𝑛𝑑 𝑃0 as in model IV(A) with
infinite capacity. Therefore,
𝑃0
(𝑠𝜌)𝑛 𝑛! 𝑓𝑜𝑟 𝑜 ≤ 𝑛 ≤ 𝑠
𝑃0 = 𝑃0 𝑓𝑜𝑟 𝑠 ≤ 𝑛 ≤ 𝑁
𝑠 𝑛 𝜌𝑛 𝑠!
0 𝑓𝑜𝑟 𝑛 > 𝑁

Where 𝑃0 may be written as

𝑠−1 𝑁
𝑠𝑝 𝑛 𝑠𝑝 𝑛
𝑛 −1
𝑃0 = [ + 𝑠! (𝑠 − 𝑠)]
𝑛!
𝑛=1 𝑛 =𝑠

𝑛 𝑠𝜌 𝑠
𝑠−1 𝑠𝑝 𝜆
[ 𝑛=0 𝑛! + (1 − 𝜌𝑁−𝑠+1) ]−1 , 𝜌 = ≠1
𝑠! 1−𝜌 𝑠𝜇
= 𝑠𝑝 𝑛 𝑠𝜌 𝑠 𝜆
[ 𝑠−1
𝑛=0 𝑛 ! + (𝑁 − 𝑠 + 1)]−1 , 𝜌 = =1
𝑠! 𝑠𝜇

This queueing modrl with limited waiting room is valuable because of its relevance to
many real situations and the fact that changes may be made to its properties by
adjusting the number of servers or the capacity of the waiting room. However, while
poission arrivals are common in practice, negative exponential service times are less
so , and it is the second assumption in the system M|M|s that limits its usefulness.

Example;
 52

A supermarket has two girls ringing up stales at the counters. If the service time for
each customer is exponential with mean 4 minutes and if people arrive in a poission
fashion at the counter at the rate of 10 per hour, then calculate.

(a) The probability of having to wait for service;

(b) The expected percentage of idle time for each girl;
(c) If a customer has to wait, find the expected length of his waiting time.

Solution

(a) Probability of having to wait for service is

(𝜆 𝑠
𝜇)
P(W>0)= 𝑃
𝑠!(1−𝜌) 0
1
λ=1/6,𝜇 = , 𝑠 = 2
4
1
𝜌 = 𝜆 𝜇𝑠 =
3
Now, compute
𝑠−1
𝑠𝑝 𝑛 𝑠𝜌 𝑠 −1
𝑃0 = [ + ]
𝑛! 𝑠! 1 − 𝜌
𝑛=0
−1
1 (2. 1 3)𝑛 (2. 1 3)2
= [ + ]
𝑛=0
𝑛! 2! (1 − 1 3)

=[1+2/3+4/9/2x2/3]-1=1/2
(4 6 )2 .0.5
Thusprob.(W>0)= =1 6
2!(1−1 3 )

(b) The fraction of the time the service remains busy is given by 𝜌 = 𝜆 𝜇𝑠 =1/3

Theefore, the fraction of the time thee service remains idle is= (1-
1/3)=2/3=67%(nearly)
1 1 1 1
(c) (W|W>0) = . = . = 3 𝑚𝑖𝑛𝑢𝑡𝑒𝑠
1−𝜌 𝑠𝜇 1−1 3 2×1 4

**************************************

UNIT- IV
 53

Inventory Control
Introduction

A inventory may be defined as an idle resources that possesses economic

value. It is an item stored or reserved for meeting future demand. Such items
may be materials machines, many or even human resources.

Reasons for holding storks

The main reasons are

(1) To ensure sufficient goods one are available to meet anticipated

demands.
(2) To absorb variations in demand and production.
(3) To prove a buffer between production processor.
(4) To take advantage of bulk purchasing discounts.
(5) To meet possible shortages in the future.
(6) To enable produces process to flow smoothly and efficiently
(7) As deliberate investment policy particularly in times of inflation or
possible shortages.

The Objective of Inventory Control

The objective of inventory control is to maintain stock levels so that the

combined costs, mentioned earlier are at amenities. This is done by establishing
two sectors.

“how to order? When to order?”Inventory control terminologies.

(1) Demand : The amount of quantity required by sales or products usually

expressed as the rate of demand for week or month or year etc.

(2) Economic order quantity (EOQ)

This is a calculated ordering quantity which minimized the balance
between inventory holding costs and re-order costs.
(3) Lead time: The period of time between ordering and replenishment.
(4) Butter stock (or safety stock). It is a stock allowance to cover errors in
forecasting the lead time on the demand during the lead time.
 54

(5) Maximum stock : A stock level as the maximum desirable which is used
as an indicator.
(6) Reorder level: The level of stock of which culture replacement order
should be placed. The re-order level 9 independent upon the lead time
and the demand upon the lead time and the demand during the lead time .

Types of Inventory

Models:

There are two types of models can be used for inventory control.

They are (i) deterministic model (ii) stochastic model.

A deterministic model is one which assumes complete continuity. The

values of all factors like demand, usage, lead time, carrying costs etc are known
exactly and there is no element of risks and uncertainty.

Stochastic model exists where some of all the factors are not known with
costmity and only they are expressed with probabilities.

Model: (Purchasing model method shortage)

In this model, the following assumptories, are made

1) The annual demand, D is uniform and is known.

2) The replenishment is made instantaneity.
(ie) the whole batch is delivered at once.
3) There is no shortages and no lead time.
4) There is known constant inventory carrying cent (C1).
5) There is known constant ordering cost (C3).
6) There is known constant unit prince.

Let Q be the ordering quantity per order.

Let of be the time interval between

Inventory level
 55

The total cost per year

= Total ordering cost + total – inventory carrying cost.

1
C4(Q) = G(Cs+ Q C1)
2

𝐷 1
= cs + C1
𝑄 2

(/D = GQ)

Differenhating w .r. to Q

We get
𝑑 𝐷 1
(CA) = (s + C1
𝑑𝑄 𝑄2 2

𝑑 2 C 𝐴 (𝑄) 2𝐷
= G is always posture
𝑑𝑄 2 𝑄3

𝑑 2𝐷𝐺
= 0 is given by Q2 =
𝑑𝑄 𝐶1

2𝐷𝐺
/Q=
𝐶1

𝑑2
For this value of Q, CA>0
𝑑𝑄 2

“ CA (Q) in minimum

2𝐷𝐺
When Q =
𝐺

This ordering quantity is denoted by Q.

 56

2𝐷𝐺
/toQ us Qo =
𝐶1

Substituting this in (1) we get the total minimum cost as

2𝐷𝐺 1 2𝐷𝐺
CA (Qo) = DG + C1
2𝐷𝐶𝑠 2 𝐶1

1 1
= 2D𝐶1 𝐶𝑠 + 2D𝐶1 𝐶𝑠
2 2

= 2D𝐶1 𝐶𝑠
𝑄𝑜
The optimum time interval between order is given by to= .
𝐷

𝑄𝑜
The optimum number of orders in a year is n to = .
𝐷

Note:-

(i) Total minimum cent is also by C4(Qo) = Qo*C1,

(ii) For calculating the total minimum cost we consider only inventory
carrying cost. The cost of material is always a constant

Worked Example

1. Annual demand for an item is 3200 units. The unit cent is Rs.6 and
inventory carrying changes 25% per annum. If the cost of one
procurement is Rs.150, determine.
(i) Economic order quantity.
(ii) Number of orders per year.
(iii) Time between the two consecutive order.
(iv) The optimal cost.
(v) The optimal cost

Solution:

Annual Demand (D) = 3200units.

Procurement cost (Cs) = Rs. 150

Inventory carrying cost (C1) = Rs. 6x25=100, Rs.150

2𝐷𝐶𝑠 2𝑥3200𝑥150
(i) EOQ is Qo = = = 800 units
𝐶1 150
 57

(ii) Number of orders per year (u)

𝐷 3200
= = =4
𝑄𝑜 800
𝑄𝑜 800 1
(iii) Optical time between two consecutive orders to = = = yr or
𝐷 3200 4
3months.
(iv) The optical cost 2𝑥3200𝑥1.50𝑥150
= Rs. 1200
2. A company buys its annual relativements of 36,000 unitsin six
instalments. Each unit costs Rs.1 and ordering cost is Rs.25. The
inventory carrying cost is estimated at 20% of unit value Find the total
annual cost of the existing inventory policy. How much money can be
saved by using EOQ.

Solution:-

D = 36000 units per year

Number of orders = 6
36000
Ordering quantity per order ()
5

Inventory carrying cost = 1x0.20 = Rs. 0.20

Ordering cost (Cs) = Rs. 25

Total Cost as per the existing policy

= Total ordering cost + total carrying cost

1
= n x Cs + = QxC1
2

1
= 6x25+ x6000x20
2

=150+600=Rs. 750

2𝐷𝐶𝑠 2𝑥3200𝑥150
EOQ= =
𝐶1 0.20

= 3000 units

Total minimum cost 2𝐷𝐺𝐶𝑠

= 2𝑥3600𝑥0.20𝑥25 = Rs. 600

 58

Amount of savings by following EOQ policy = Rs.750-Rs.600=Rs.150

Note: Total minimum cost = oxC1

= 3000x02=600 Rs.
36000
Total ordering cost = x 25
3000

= Rs.300

Model II

Production model without shortages

(Manufacturing model – No Shortaged

In this model we assume that the ordering quaintly is received only over a
period of time. Further is assumed that the production rate is greater than the
demand rate, otherwise there will be no inventory build up and stock only
will occur. These two assumptions are made in addition to the assumption of
model I. Let k be the production rate, r be the demand rate, otherwise there
will be no inventory build up and stock only will occur. There two
assumptions are made in addition to the consumption of model I.

Let k be the production rate, r be the demand rate. Let tbe the length of
production run rate, r be the demand rate. Let t be the length of production
run

𝑄
Total production = Q = kt t=
𝐾
 59

The average height is half the height of the triangle and the height in
determined by the rate of replenishment less the demand over the
replimshment.
𝑘𝑡 𝑟𝑡
period= -
2 2

𝑘−𝑟
= G
𝑘

Total cost = total Ordering cost + Total Carrying cost

𝑄 𝑟
= u Cs + (1- ) G
2 𝐾

𝑟 𝑄 𝑟
CA(Q) = cs + (1- )G
 2 𝐾

𝑑 2𝑟𝐶𝑠
CA = > 0 always
𝑑𝑎 𝑄2

𝑑 𝑟
CA = OQ2=2rCs
𝑑𝑄 𝐾−𝑟

Total minimum cost is given by

𝑘−𝑟 1 𝑘−𝑟 2𝐺𝑟 (𝑘−𝑟)

CA (Qo) = rcs + G
2𝑟𝐺𝐾 2 2𝑟𝐺𝐾 𝐺𝐾

2𝐺𝑟 (𝑘−𝑟)
=
𝐺𝐾

𝑄𝑜
The optimum time interval between orders is to
𝑟

The optimum of number of order is

𝑟
n=
𝑜

Worked Example

A manufacturer has to supply his customer with 600 units of his product
per year. Shortages are not allowed and storage cost amounts to 60paise per
unit per unit per year. The set up cost associated with ordering 200% higher
than EOQ.

Solution
 60

D = 600 units per year

Cs= Rs.80 per setup

C1 = Rs. 0.60

2𝐷𝐺 2𝑥600𝑥80
(i) EOQ = = = 4000 units.
𝐺 0.60
(ii) The minimum average yearly cost
= QoxC1 = 4000x0.6 = Rs. 240
(iii) The optimum number of per year
D 600
= = 1.5
𝑜 400

𝑜 400 2
(iv) The optimum period of supply = = = year = 8 months.
D 600 3
20
(v) 200% of EOQ = 400 x = 80
100

/ New ordering quality = 480 units

1
T.C. nCs + Q C1
2

600 1
= x 80 + x 480 x 0.60
480 2

Increase is cent

= Rs. 244 – Rs. 240

= Rs. 4

Model III [Purchasing model with shortage]

Assumptions

(1) Annual demand „D‟ is uniform at a rate of r units per unit time.
(2) Production is instantaneous.
(3) Zero lead time
(4) G is the inventory carrying cost per units per year.
(5) Cs is the setup cost
(6) C2 is the shortage cost per unit/yr
 61

2𝐷𝐶𝑠 𝐶1 +𝐶2
Q=
𝐶1 𝐶2
𝐶2 
Maximum inventory level = x and to =
𝐶1 +𝐶2 𝐷
𝐶2
Minimum cost CA () = 2𝐷𝐶1 𝐶𝑠
𝐶1 +𝐶2

Worked Example

Consider the following data

Unit cost = Rs. 100

Order Cost = Rs. 160, Inventory carrying

Cost = Rs. 20, Back order (stock out cost) Rs. 10

Annual demand = 1000 unit

Computed the following

(i) Minimum order quantity

(ii) Time between
(iii) Maximum number of backorder
(iv) Maximum inventory level
(v) Overall annual cost

Solution:

D = 1000 units, Cs=Rs.160, C1=Rs.20 C2= Rs.10.

2𝑥1000𝑥160
(i) Minimum cost order quantity =
10
(ii) = 219 units.
∗ 219
(iii) Time between orders: f* = = x12
𝐷 1000
= 2.628 months

Maximum number of backorders

=*-* = 219-73 = 146 units

(iv) Maximum Inventory level
𝐺 10
Q1* = Q* = 219 = 73 units
𝐶1 +𝐶2 20+10
 62

𝐺
v) Over all annual cent = 1000x100 2𝐷𝐶1 𝐶𝑠
𝐶1 +𝐶2

10
= 100000+ 2𝑥1000𝑥20𝑥160𝑥 = Rs.
30
1,01,306

Model IV [Manufacturing model with shortage]

1. Demand D is uniform at a rate of „runits‟ per unit time.

2. Production rate is finite say k>r units per unit time
3. Shortages are allowed and have logged
4. Zero lead time
5. C1= inventory carrying cost per unit time.
6. C2 = shortage cost per unit line
7. Cs = set up cost per run.

Optimal order quantity,

2𝑐𝑠𝑟 (𝐶1 +𝐶2 )

Q=
𝐶1 𝐶 𝑟
2 1−
𝑘

2𝑐𝑠𝑟 (𝐶1 +𝐶2 )

Minimum cost Q* =
𝐶1 𝐶 𝑟
2 1−
𝑘

𝑟
𝑄 ∗ 𝐶1 (1− )
𝑘
S* =
𝐶1 +𝐶2

𝑄∗
t*= , minimum cost in
𝑟

𝑟
2𝑟 𝐶1 𝐶2 𝐶𝑠 (1− )
𝑄
C* =
𝐶1 +𝐶2

Worked Example:-

1. An item is produced at the rate of 50 items per day. The demand occuss at
the rate of 25 items per day. If the setup cost is Rs.100 per set up and
holding cost is Rs. 0.01 per unit of item per day. Find the economics lot
size for one run, assumes that the shortages are not allowed. Also find
the time fo cycle and the minimum total cost of one run.
 63

Solution

k = 50 perday

r =25 per days

Cs = Rs. 100

C1 = Rs. 0.01

2𝐶𝑠𝑟 𝑘
Qo =
𝐶1 𝑘−𝑟

2𝑥100𝑥25 50
Q* = = 100 units
0.01 50−25

𝑄∗ 1000
t* = = = 40 days
𝑟 25

𝑘
CA (Q*) = 24𝐺𝐶𝑠
𝑘−𝑟

50−25
= 2𝑥25𝑥0.01𝑥100 = Rs.5
50

Total variable cost per run = 5x40 = Rs. 200

Model V- Price Breaks Model we consider the case that the unit price
vary with ordering quantity usually some discoints in unit price are offered
to encourage having large purchasing quantity. Above the rates of discoint
may vary depending on the purchasing quantity. These types of inventory
models or prove break models. The general pattern of price breaks is given
in the following table showing the ordering quantity and the price breaks.

Ordering quantity unit price

0{1<b1 P1

b1{2<b2 P2

b2{3<b3 P3
------------ -----

Bn-1{n<bn Pn
 64

Hence P1>P2……..>Pn. The price of item falls as the ordering quantity is b1b2., bn-1
and so these ordering quantity level are called price breaks consider the inventory
model with one price break.

Ordering quantity unit price

0Q1<b P1

b1Q2<b2 P2

b2Q3 P3

Working rule

1. Calculate Q3* taking the lead unit price P3. If Q3*>b2 then Q3* is the EOQ.
2. If Q3* <b2 there calculate Q2*. Now compare units C(Q2*) and b2 If C(Q2*)
in the EOQ. If (b2) is the minimum then b2 is the EOQ.
3. If Q2*<b1 then, calculate Q1* New compare the costs (Q1*)) and (b1) and
((b2)

If C(Q*) is minimum EOQ is Q1*

If C(b1) is minimum EOQ is b1
If C(b2) is minimum EOQ is b2

Example:

Find the optimal order quantity for a product for which the prince leaks are us
follows.

Quantity Purchasing cost

01<100 Rs. 20/ unit
1002<200 Rs. 18/unit
200  Qs Rs.14/unit
The marthly demand for the provided is 400 units. The storage cost is
Rs. 20% of the unit cent of the product/month and cost of undering is
Rs.25/-

Solution

D = 400 units permonth

Cs = Rs.25
C1 = 20% of unit per month
 65

Q3* does not lie in the last range

/Q3* can not be the EOQ.

2𝑥400𝑥25
Q3 * = = 75 units
0.20𝑥18

This does not lie in the range

100Q2<200

2𝑥20𝑥25
Q1 * = = 75 units
0.20𝑥20

Now we have to compare

CA(Q1*) , CA(100), CA(200) to find EOQ.
CA (Q1*) = cost of the material +total inventory cost

= 400x20+ 2𝐷𝐶1 𝐶2

= 8000+ 2𝑥400𝑥4𝑥25
= 8283 nearly
400 1
CA(100) = 400x18 + x25 x 100 x 3.6
100 2

= 7200 + 100+180
= Rs. 7480
400 1
CA(200) = 400 x 16 + x25x x200x3.2 = Rs. 6770.
100 2

Since total, cost is minimum for 200 units, EO = 200 units.

Models : Probabilistic model (stochastic model)

In this model, we have to determine the optimum inventory evel which
mentioned the expected cost. The following assumptions are made.
(i) T is constant internalbetween orders.
(ii) Z is the stock level at the beginning of each period of time „t‟
(iii) Lead time is zero.
(iv) C1 is the holding cent per unit time.
(v) C2 is the shortage cost per unit time.
(vi) Also assume that the demand is instantionous „r‟ be the demand, p(r) be the
probability that demand is r units in time „t‟.
 66

Two cases: discrete of continuous

𝑧−1 𝐶2 𝑍
Discrete case: 4 is discrete If 𝑟=0 𝑝(𝑟)<𝐶 +𝐶 < 𝑟=0 𝑝(𝑟) then C(z+1) = C(z) and the
1 2
optimum stock level is either Zo or Zo+1
Newspaper boy problem
When r and z are discrete values, the discrete stochastic model is hand as news-
paper boy problem. The news boy buys certain number of papers Z and sell some or
all of them. He make profit for each sold paper and less for each unsold paper. All of
demand are than the stock on anyday the ne is in the problem to find the optimum
number of newspaper he was us buy each day so as to maximize his profit.

ABC Analysis
The university of an organization may consist of several items varying with
cents usage, lead time, procurements cents and so on. Here we should pay more
attention on claims with low usage value. So we should be selective to our approach
towards inventory central. This is called relative inventory control and hence we have
ABC analysis (always better control analysis).
Example:
The items kept in inventory by the school management studies at state
university are instead below which claims should be classified as „A‟ claims, „B‟ less
„c‟ claims? What percentage of claims in each land.
Item Annual usages Value per unit
1 200 400.00
2. 100 366.00
3. 2000 0.20
4. 400 20.00
5. 6000 20.00
6. 1200 0.04
7. 120 0.80
8. 2000 100.00
9. 1000 0.70
10. 80 400.00
Solution:
In order to classify the items kept in the A,B,C Class items the first step into
calculate the annual usuage value in the descending order.
Value per Ranking
S.No Annual Usage Annual usage
item Rs.
 67

1 200 40 8000 IV
2 100 360.00 36000 I
3 2000 0.20 400 IX
4 400 20.00 8000 V
5 8000 0.04 240 X
6 1200 0.80 960 VIII
7 120 100.00 1200 III
8 2000 0.70 1400 VI
9 1000 1.00 1000 VII
10 80 400.00 32000 II
The table given below classifies the above items under three class A,B,C
Annual
Annual
36000 Usage % of total
Item Rank Usage Class
32000 Value items
Value
ingots
2 I 36000 2 68000𝑥10
A 68000 =206 =68%
10 II 32000 10 10000
7 III 12000
3 68000𝑥10
4 IV 8000 B 28000 =30% =68%
10 100000
1 V 8000
8 VI 1400
9 VII 1000 C 4000 5 4000𝑥10
=50% =68
10 100000
6 VIII 960
= 100
3 IX 400
5 X 240
Iron the table it is clear that 20% of inventory claims belonging to A class,
controller 68% of total annual usage value 30% falls under class B, 50% falls in C.

*********************************
UNIT-V
Network Analysis
Network Analysis is a technique which determines the
various sequences of jobs concerning a project and the project completion
time. Network analysis has been successfully used to a wide range of
significant management problems. Two popular methods of this technique
which are widely used are 1) The Critical Path Method (CPM) and 2) The
Programme Evaluation and Review Technique(PERT).
 68

CPM: This method differentiates between planning and scheduling. Planning

refers to the determination of activity and the order in which such activities
should be performed to achieve the objective of the project. Scheduling
refers to introduction of time in performing the jobs concerning the project
taken in the plan. CPM technique is generally applied to well known projects
where the time schedule to perform the activities can exactly be determined.
In CPM two types of times are used. They are i) Normal time(time taken to
perform an activity), ii) Crash time(the shortest time in which an activity can
be finished)

PERT: This was specially developed for planning and controlling the polaris
missile programme. Using this technique the management was able to obtain
the expected project completion time and the bottle neck activities in a
project. In PERT three types of times are used. They are i) Optimistic time
estimate, ii) Pessimistic time estimate, iii) Most likely time estimate. In
PERT, statistical analysis is used in determining the time estimates.

A path of network is the sequence of activities starting from the

initial events to the final event proceeding in the direction of arrows. The
duration of a path is the sum of activities coming along the path.

The path that takes the longest duration is called the critical path.
The duration of this path is called the project duration. We introduce the
following definitions.

(1) EST: It is the earlier possible time at which an activity can be started
assuming that all the preceding activities can be started at the earliest start
times.
(2) EFT: The earliest finish time for an activity is the earliest start time plus
the duration of the activity. Ie)EFT=EST+Duration.
(3) LFT: The latest finish time for an activity is the latest possible time at
which an activity can be finished assuming that all the subsequent
activities can also be finished at the latest possible times.
(4) LST: The largest start time of an activity is equal to the latest finish time
minus the duration of the activity.
(5) Total Float(TF): TF= Head event L-Tail event E-Duration

TF=LFT-EFT=LST-EST.

(6) Free Float(FF): FF=Head event L-Tail event E-Duration

 69

FF=TF-Head event Slack(L-E)

(7) Independent Float(IF): IF=Head event L-Tail event E-Duration
=FF-Tail event slack(L-E).

Dummy activity: This is the activity which doesnot consume time or resources.
It is used only to show logical dependencies between activities so as not to
violate the rules for drawing networks. It is represented by a dotted arrow line in
a network diagram.

PERT Network: A PERT network is drawn in the same way as a CPM network.
In PERT network we need only to estimate the duration of each activity. There
are 3 types of time estimates.

1. The Optimistic time estimate(T0): It is the shortest possible time estimate

for finishing an activity. The chance of occurrence of this is very small.
2. The Pessimistic time estimate(Tp): It is the longest time conceivable for
an activity. The chance of this occurrence is also very small.
3. The most likely time estimate(Tm): This is the the time estimate to be
executed under normal conditions of the activity. This is a reasonable
time to estimate.

Problem 1:

The following table gives the activities in a construction project and other
relevant information.

Activity 1-2 1-3 2-3 2-4 3-4 4-5

Duration(Days) 20 25 10 12 6 10
(i) Draw the network for the project.
(ii) Find the critical path and the project duration.
(iii) Find the total float for each activity.

Solution:

L=20, E=20

L=0, E=0
1 4 5
 70

E=46, L=46

E=36, L=36

3
E=30, L=30

Activity Duration EST EFT LST LFT TF

1-2 20 0 20 0 20 0
1-3 25 0 25 5 30 5
2-3 10 20 30 20 30 0
2-4 12 20 32 24 36 4
3-4 6 30 36 30 36 0
4-5 10 36 46 36 46 0

The critical path is 1-2-3-4-5

The project duration is 46 days.

Problem 2:

A project has the following time schedule.

Activity Time in months Activity Time in months

1-2 2 3-7 5
1-3 2 4-6 3
1-4 1 5-8 1
2-5 4 6-9 5
3-6 8 7-8 4
8-9 3

(i) Construct the network

(ii) Find the total float for each activity
(iii) Find the critical path and the project duration.

Solution: L=11, E=6 L=12, E=11

L=7,E=2 5
2 8

1 3 7
 71

L=0, E=0 L=2, E=2 L=8, E=7

4 6 9
L=7, E=1 L=10, E=10 L=15, E=15

Activity Duration EST LFT LST EFT TF

1-2 2 0 7 5 2 5
1-3 2 0 2 0 2 0
1-4 1 0 7 6 1 6
2-5 4 2 11 7 6 5
3-6 8 2 10 2 10 0
3-7 5 2 8 3 7 1
4-6 3 1 10 7 4 6
5-8 1 6 12 11 7 5
6-9 5 10 15 10 15 0
7-8 4 7 12 8 11 1
8-9 3 11 15 12 14 1

The critical path is 1-3-6-9

The project duration is 15 months.

Problem 3:

The activities of a project have the following PERT time estimates.

Job Optimistic time Most likely time Pessimistic time

1-2 3 6 15
7-8 4 19 28
2-3 6 12 30
3-5 5 11 17
5-8 1 4 7
6-7 3 9 27
4-5 3 6 15
2-4 2 5 8
1-6 2 5 14

(i) Draw the network diagram and determine the critical path.
(ii) Find the project completion time and its variance.
 72

Solution:

2 3 5 8

1 4

6 7

Job 𝑇𝐸 𝑇𝑃 − 𝑇𝑂 Variance
𝑇𝐷 + 𝑇𝑝 + 4𝑇𝑚 𝑆𝐷( )
6
=
6
1-2 42/6=7 2
7-8 108/6=18 4 16
2-3 84/6=14 4 16
3-5 66/6=11 2 4
5-8 24/6=4 1 1
6-7 66/6=11 4 16
4-5 42/6=7 2 4
2-4 30/6=5 1 1
1-6 36/6=6 2 4

The various paths are

Duration
1. 1-2-3-5-8 36
2. 1-2-4-5-8 23
3. 1-6-7-8 35

The critical path is 1-2-3-5-8

The project duration is 36 days.

Variance of the critical path=4+16+4+1=25.

 73

Problem 4:

The following table lists the jobs of a network with their time estimates

Job Optimistic Duration days Pessimistic

Most Likely
1-2 3 6 15
1-6 2 5 14
2-3 6 12 30
2-4 2 5 8
3-5 5 11 17
4-5 3 6 15
6-7 3 9 27
5-8 1 4 7
7-8 4 19 28

Draw the project network and calculate the length and variance of the critical
path.

Solution:

Job 𝑇𝐸 𝑇𝑃 − 𝑇𝑂 Variance
𝑇𝐷 + 𝑇𝑝 + 4𝑇𝑚 𝑆𝐷( )
6
=
6
1-2 42/6=7 12/6=2 4
1-6 36/6=6 12/6=2 4
2-3 84/6=12 24/6=4 16
2-4 30/6=5 6/6=1 1
3-5 66/6=11 12/6=2 4
4-5 42/6=7 12/6=2 4
6-7 66/6=11 24/6=4 16
5-8 24/6=4 6/6=1 1
7-8 108/6=18 24/6=4 16
1 2 4

3 5 8

6 7
 74

The various paths are

Duration
1. 1-2-4-5-8 23
2. 1-2-3-5-8 36
3. 1-6-7-8 36

The critical path is 1-2-3-5-8 and 1-6-7-8.

The project duration is 36 days.

Variance of the critical path=4+16+16=36.

Exercises:

1. A project consists of 12 jobs. Draw a project network and determine the

critical path.
Job Duration Job Duration
1-2 2 6-7 8
2-3 7 6-10 4
2-4 3 7-9 4
3-4 3 8-9 1
3-5 5 9-10 7
4-6 3
5-8 5

2. Consider the following project whose activities along with PERT time
estimates the optimistic time (a), most likely time (m), and the pessimistic
time (b) are given as follows.
Activity a (days) m (days) b (days)
1-2 12 14 21
1-3 7 10 16
3-5 4 6 10
3-4 36 40 60
4-6 12 15 24
5-6 6 8 12
6-7 9 12 18
 75

6-8 6 10 15
7-8 4 5 7
8-9 8 10 14

*******************************