Operations Research

II
STA 412
 Course Outline

 Queue Theory
 Inventory Theory
 Simulation

2
 Lecture One
Queue Theory
Learning Objective
 Understanding the nature and scope of queuing system
 Understanding Queuing models and the solution to queuing model problems

INTRODUCTION

Queue when rate of arrival of people or customer at service point is higher than the rate of
departure at the service point. Places where we can experience queue include supermarkets,
banks, communication systems, information processing systems etc.

Queuing System

Queuing system can be characterised by the following

1. Arrival (Inter-arrival) Distribution: This represents the pattern in which the number
of customers arrive at the service facility. Arrivals may also be represented by the inter-
arrival time, which is the time between two successive arrivals. The inter-arrival time
may be equal interval of time or unequal but definitely known interval of time or unequal
interval of time whose probabilities are known. These are called random arrivals. The
best probability distribution that describes arrival rate is Poisson distribution. Here the
mean value of arrival rate is given as λ . Other probability distribution like (negative)
1
exponential distribution with mean value of arrival rate given as can also be used.
λ
2. Service (Departure) Distribution: This represents the pattern in which the number of
customers depart the service facility. Departures may also be represented by the inter-
departure time, which is the time between two successive departures. Service time may
be of constant or variable with known probability distribution. When service time is
random, the best probability distribution that describes service rate is exponential
distribution with mean service rate is given as μ.
3. Service Channels (Number of Servers): This is number of service facilities provided
at the service centre. It may single or multi-service channels, which may be arranged in
parallel or in series or a complex combination of both.
i. Parallel Channels: Several customers may be served simultaneously as in
receiving and paying cashiers in banking hall.
ii. Series Channels: A customer must pass successively through all the channels
before service is completed, eg a product undergoing different processes over
different machines or fresh student, during admission going through one office
after another before all admission formalities are completed.
4. System Capacity: Maximum number of customers in the system can be finite or
infinite. In some facilities, only a limited number of customers are allowed in the system

3
 and new arriving customers are not allowed to join the system unless the number
becomes less than the limiting value.
5. Service Discipline: Service discipline or queue discipline is the order with which
waiting customers are served. There are many possibilities which include
 First Come First Serve (FCFS)
 Last Come First Serve (LCFC)
 Service in Random Order (SIRO)
6. Customer’s Behaviour: The customer’s behaviour is also very important in the study of
queues and they include
i. Balking: The tendency of a customer not joining the queue because of the fear of
delay.
ii. Reneging: The tendency of a customer joining the queue but leaves the queue after
some time probably due to impatient.
iii.Jockeying: The tendency of people joining the queue but after sometime joins
another.

OPERATING CHARACTERISTICS OF A QUEUING SYSTEM

Analysis of a queuing system involves a study of its different operating characteristics. Some of
them are

1. Queue Length (Lq): The average number of customers in the queue waiting to get
service. This excludes the customer(s) being served.
2. System Length (Ls): The average number of customers in the system including those
waiting as well as those being served.
3. Waiting Time in The Queue (Wq): The average time for which a customer has to wait
in the queue to get service.
4. Waiting Time in The System (Ws): The average total time spent by a customer in the
system from the moment he arrives till he leaves the system. It is taken to be the waiting
time plus the service time.
5. Utilization factor (ρ): It is the proportion of time a server actually spends with the
customers. It also called traffic intensity.

KENDALL’S NOTATION FOR REPRESENTING QUEUING MODELS

Notation for queuing model can be expressed using Kendall’s notation as

( a /b/c ) : ( d /e/ f )

where,

a = arrival (or inter-arrival) distribution

b = departure (or service time) distribution

c = number of parallel service channels in the system

4
 d = service discipline

e = maximum number of customers allowed in the system

f = calling source or population

The following standard notation replaces the symbols a, b, and d:

The symbols a and b can be replaced by any of the following:

M = Markovian (Poisson) arrival or departure distribution (or exponential

inter-arrival or service time distribution)

Ek = Erlangian or gamma inter-arrival or service time distribution with

parameter k

GI = general independent arrival distribution

G = general departure disrtribution

D = deterministic inter-arrival or service times

The symbol d can be replaced by any of the following:

FCFS = First Come First Serve

LCFS = Last Come First Serve

SIRO = Service in Random Order

GD = general service discipline.

The symbols e and f represent a finite (N) or infinite (∞) number of customers in the system and
calling source respectively.

Illustration

Consider the notation (m/m/10) : (GD/N/∞)

This means Poisson arrivals, exponential service time (departure), and 10 parallel servers in the
facility. The service discipline is general service discipline (GD). That is it can be either be
FCFS, LCFS, SIRO or whatever procedure the servers might wish to adopt. The system capacity

5
can only hold maximum of N customers, other can seek service elsewhere. The population size
is infinite.

SINGLE CHANNEL QUEUING THEORY

A single channel queuing problem results from random inter-arrival time and random service
time at a single service station. The random arrival time can be described mathematically by
probability distribution. The most common distribution found in queuing problems is Poisson
distribution. This is used in single channel queuing problems for random arrivals where the
service time is exponentially distributed.

Queuing Model For Single Channel Poisson Arrivals With Exponential Service, Infinite
Population [(M/M/1) : (FCFS/∞/∞)]

Let us consider a single channel system with Poisson arrivals and exponential service time
distribution. Both the arrivals and service rates are independent of number of customers in the
waiting line. Arrivals are handled on ‘First come, first served’ basis. Also the arrival rate λ is
less than the service rate μ.

The following mathematical notations (symbols) will be used in connection with queuing
models:

n = number of customers in the system (waiting line + service

facility) at time t.

λ = mean arrival rate (number of arrivals per unit time).

μ = mean service rate per busy server (number of customers served

per unit of time).

λdt = probability that an arrival enters the system between t and t + dt

time interval i.e within time interval dt.

1 – λdt = probability that no arrival enters the system within interval dt plus

higher order in dt

μdt = probability of one service completion between t and t + dt

time interval i.e within time interval dt.

1 – μdt = probability that no service rendered during the interval dt plus

higher order in dt

6
 Pn = steady state probability of exactly n customers in the system.

Pn(t) = transient state probability of exactly n customers in the system at

time t, assuming the system started its operation at time zero.

Pn+1(t) = transient state probability of having n + 1 customers in the system

at time t.

Pn–1(t) = transient state probability of having n – 1 customers in the system

at time t.

Pn(t+dt) = probability of having n customers in the system at time t + dt.

To determine the properties of the single channel system, it is necessary to find an expression for
the probability of n customers in the system at time t i.e Pn(t), for if Pn(t) is known, the expected
number of customers in the system and hence the other characteristics can be calculated. In place
of finding an expression for Pn(t), we shall first find the expression for Pn(t+dt).

The probability of having n customers in the system at time t + dt can be determined by

summing up probabilities of all the ways this event could occur. The event can occur in four
mutually exclusive ways.

Event No. of Units at No. of Arrivals No. of Services No. of Units at

Time t in Time dt in Time dt Time t + dt

1 N 0 0 n

2 n–1 1 0 n

3 n+1 0 1 n

4 N 1 1 n

Now we can calculate the probability of occurrence of each of the events. Note that ( dt )2 → 0

Prob . of event 1=Prob . of having n customers at timet × Prob . of no arrivals × Prob . of no services

¿ Pn ( t ) × ( 1− λdt ) × ( 1−μdt )

7
 ¿ Pn ( t ) [ 1−λdt −μdt + λμ ( dt )2 ]

¿ Pn ( t ) [ 1−λdt −μdt ]

Prob . of event 2=Pn−1 (t ) × ( 1−μdt ) × λdt

¿ Pn−1 ( t ) × λdt

Prob . of event 3=Pn +1 ( t ) × ( 1−λdt ) × μdt

¿ Pn+1 ( t ) × μdt

Prob . of event 4=Pn ( t ) × λdt × μdt

2
¿ Pn ( t ) × λμ ( dt ) =0

Since one and only one of the above events can happen, we can obtain Pn ( t+ dt ) [where n>0] by
adding up the above four events.

∴ P n ( t +dt )=Pn ( t ) [ 1−λdt −μdt ] + Pn +1 ( t ) × μdt + Pn−1 (t ) × λdt+ 0

¿ Pn ( t )−P n ( t ) [ λdt + μdt ] + Pn+1 ( t ) [ μdt ] + Pn−1 ( t ) [ λdt ]

P n ( t +dt )−Pn ( t )
=−( λ+ μ ) . Pn ( t )+ μ . Pn+1 ( t )+ λ P n−1 ( t )
dt

Taking the limit whendt → 0 , we get the following differential equation which gives the
relationship between Pn ( t ) , Pn−1 ( t ) , Pn +1 ( t ) at time t , mean arrival rate λ , and mean service rate μ
:

dPn ( t )
=λ Pn −1 ( t )+ μ . Pn+1 ( t ) −( λ+ μ ) . Pn ( t ) (1)
dt

where, n> 0

Next, we solve for Pn ( t+ dt ) when n=0. If n=0, only two mutually exclusive and exhaustive
events can occur as shown in table below:

Event No. of Units at No. of Arrivals No. of Services No. of Units at

8
 Time t in Time dt in Time dt Time t + dt

1 0 0 0 0

2 1 0 1 0

Prob . of event 1=Prob . of having no customers at time t × Prob . of no arrivals × Prob . of no services

¿ P0 ( t ) × ( 1− λdt ) ×1

Prob . of event 2=Prob . of having 1customers at timet × Prob . of no arrivals × Prob . of one services

¿ P1 ( t ) × ( 1−λdt ) × ( μdt )

The probability of having no customer in the system at time t+ dt is given by summing up the
probabilities of the above two events.

P0 ( t+ dt )=P 0 ( t ) × ( 1−λdt ) × 1+ P 1 ( t ) × (1−λdt ) × ( μdt )

¿ P0 ( t +dt )−P 0 ( t )=−P0 ( t ) . ( λdt )+ P1 ( t ) . ( μdt )

P 0 ( t +dt )−P0 ( t )
=P1 ( t ) . ( μ )−P0 ( t ) . ( λ )
dt

whendt → 0 , we get the following differential equation which gives the relationship between
P0 ( t ) ∧P 1 ( t ) at time t , mean arrival rate λ , and mean service rate μ:

d P0 ( t )
=P 1 ( t ) . ( μ )−P 0 ( t ) . ( λ )(2)
dt

Where n=0

Assuming the steady state condition for the system, when the probability of having n customers
in the system becomes independent of time, we get

dPn ( t )
Pn ( t ) =P n , =0
dt

Therefore, for steady state system the differential equation (1) and (2) reduce to difference
equations

λ Pn−1 + μ . P n+1−( λ+ μ ) . Pn=0 , where n> 0(3)

μ . P 1−λ P0=0 , where n=0( 4)

9
From (4), we have

λ
P 1= P 0
μ

Putting n=1, in (3), we have

λ P0 + μ . P2−( λ+ μ ) . P1=0

( λ+ μ ) λ ( λ+ μ ) λ λ
P 2= . P 1 − P 0= P0 − P 0
μ μ μ μ μ

λ
μ
P0 (
( λ+ μ )
μ
−1 =
λ 2
μ )()
P0

Similarly, for n=2, (3) gives

()
3
λ
P 3= P0
μ

()
n
λ
Pn= P 0 (5)
μ

(5) gives Pn in terms of P0, λ∧μ. Finally, an expression for P0 in terms of λ∧μ must be
obtained. To do this, note that the probability that system is busy is ratio of the arrival rate and
service rate. That is

λ
μ

Thus

λ
P0=1−
μ

Hence

( ) (1− μλ )
n
λ
P n=
μ

Having known the value of Pn, we can find the various operating characteristics of the system.

OPERATING CHARACTERISTICS OF THE SYSTEM

10
1. Expected Number of Customers In The System (Ls):
By definition,
∞
E ( x )= ∑ x i Pi
i=0

( )( ) ( ) ∑ n( μλ )
∞ n ∞ n
λ λ λ
Ls =E ( ns ) =∑ n 1− = 1−
i=0 μ μ μ i=0

( )[ () () ]
2 3
λ λ λ λ
1− 0+ +2 +3 +…
μ μ μ μ
This is sum of infinite series and the formula for infinite series is given as
a
S∞ =
( 1−a )2

[( ) ]
λ λ
∴ Ls= 1− ( λμ ) μ

1−
λ
2
=
1−
μ
λ
=
λ
μ−λ
μ μ
2. Expected Number of Customer in the Queue (Lq):
By definition,
∞
Lq= E ( nq ) =∑ ( n−1 ) Pn
n=1

∞ ∞
¿ ∑ n Pn − ∑ P n
n=0 n=1

¿
λ λ
× =λ
μ−λ μ
μ−μ+ λ
=
λ
(
λ λ
= Ls
μ ( μ−λ ) μ μ−λ μ ) ( )
3. Mean Response Time or Mean Delay or Mean Waiting Time in the System (Ws)
This is defined as
Expected number of Customers∈the System
W s=
Arrival rate
By Little’s law,
Ls =λ W s

( )
L 1 λ 1
∴ W s= s = =
λ λ μ− λ μ−λ

4. Mean Delay or Mean Waiting Time in the Queue (Wq)

By Little’s law,
Lq = λ W q

11
 Lq 1
( ) ( )
2
λ 1 λ
∴ W q= = 2
=
λ λ μ −μλ μ μ− λ

5. Average Length of Non-Empty Queue Ln:

This is the length of queue that is formed from time to time to time. For a non-empty
queue, the number of customers in the system should be at least 2 (one in the queue and
the other in service).
Probability of a non-empty queue
∞
Probability of a non−empty queue=∑ P n−( P0 + P1 )
n=0

( λ
)
¿ 1− P0 + P0 =1−P0 1+
μ
λ
μ ( )
( )( ) ( )
2
λ λ λ
¿ 1− 1− 1+ =
μ μ μ

Now average length of non-empty queue,

Ln=
Average length of queue
=
1
( )= μ
λ
μ μ−λ

( λμ ) λ ( μ−λ )
2
Probability of non−empty queue

6. Average Waiting Time in Non-Empty Queue Wn:

This is waiting time of an arrival who waits or expected waiting time per busy period.
1
W n=
μ−λ

7. Probability of k or more Customers in the System

()
k ❑ n
λ
P ( ≥ k )=∑ Pn =∑ ( 1−ρ ) ρ =ρ =
j n

n=1 ❑ μ

8. Probability of Queue Greater Than n

P ( ¿ n )=
μ()
λ n+1

9. Probability that the Queue is Non-Empty

()
2
λ
P ( ¿ 1 )=
μ

10. Traffic Intensity ( ρ ) : Traffic intensity is defined as the ratio of arrival rate λ to service
rate μ, mathematically,
λ
ρ=1−P0=
μ

12
 It is also called utilization factor and it determines the degree to which the capacity of the
service station is utilized (expected fraction of time the service facility is busy).

Example:

Suppose that customer arrive on ( m/ m/1 ) model at a Poisson rate of 1 person per 12mins and that
the service time is exponential at the rate of 1 person per 8mins. Find Ls , Lq ,W s ,∧W q

Solution:

Here

1 1
λ= ∧μ=
12 8

a) Expected number of people in the system (Ls )

λ ρ λ
Ls = = But ρ=
μ−λ 1−ρ μ

1 8 2
ρ= × =
12 1 3

2
3 2 3
∴ Ls = = × =2 persons
2 3 1
1−
3
b) Expected number of people in the queue (Lq )

Lq =
λ λ
( )
μ μ−λ
λ
= Ls
μ
2 4
¿ ×2= =1.33 ≅ 1 person
3 3
c) Mean waiting time in the system (W s)

( )
Ls 1 λ 1
W s= = =
λ λ μ−λ μ−λ
1 1
¿ = =24 mins
1 1 1
−
8 12 24

d) Mean waiting time in the queue (W q)

Lq 1
( ) ( )
2
λ 1 λ
∴ W q= = 2
=
λ λ μ −μλ μ μ− λ

13
 ( )
1
12 1 24
¿8× =8 × × =16 mins
1 1 12 1
−
8 12

Example:

Suppose arrival at an ATM machine follows a Poisson process with mean of 12 persons per
60mins. The length of withdrawal follows exponential distribution with mean 2mins. What is

i. The probability that a new customer coming in finds the ATM machine engaged?
ii. The average length of queue when it is formed?
iii. If it is the bank’s policy to install new ATM machine, if on the average customer waits
for at least 3mins for the ATM, by how much must arrivals increase in order to justify the
second ATM?

Solution:

The probability that a new customer finds the ATM engaged is

P ( n>0 ) =1−P ( n=0 )

That is

λ
Pn=1−P0=1−1−ρ= ρ=
μ

Mean Arrival Rate λ=12

Since the service follows exponential distribution with the mean

1 2 1
¿ = =
μ 60 30

∴ Mean Service Rate μ=30

i. Hence Probability of finding the ATM machine engaged

λ 12 2
Pn= ρ= = = =0.4
μ 30 5
ii. The length of queue is

Ln =
Average length of queue
=
1
( )
λ
μ μ−λ
=
μ
Probability of non−empty queue
()
2
λ λ ( μ−λ )
μ

¿
μ
λ ( )
×
1
μ−λ
1
= ×
ρ
1−
1 1
= ×
1
λ ρ 1−ρ
μ

14
 1 1 1 1 1
¿ × = × = =4.17 ≅ 4 person
0.4 1−0.4 0.4 0.6 0.24

iii. For the second boot to be justified the mean delay in the queue must be greater than
3
hrs
60

That is,

W q= ( )
λ 1
>
3
μ μ−λ 60

λ 3
⇒ >
30 ( 30−λ ) 60

⇒ 60 λ> 90 ( 30−λ )

⇒ 60 λ> 2700−90 λ

⇒ 150 λ> 2700

2700
∴ λ= =18 /hr
150

The increase therefore is 18−12=6 persons/hr , therefore the second ATM machine is
justified with this increase.

Exercises:

1. In a single pump gas station, the arrival rate is exponential with a mean of 10mins. The
service rate is also exponential with a mean of 6mins. If the waiting space is unlimited,
find
a) The probability that there is no customer in the system
b) Expected number of customers in the system
c) Expected number of customers in the queue
d) Mean waiting time in the system
e) Mean waiting time in the queue
2. At one-man hair saloon, customers arrive according to Poisson distribution with mean
arrival rate of 5persons per hour and the plaiting time was exponentially distributed with
an average plaiting taking 10mins. It is assumed that because of his excellent reputation,
customers were always willing to wait. Calculate the following:
i. Average number of customers in the saloon and the average number of customer
waiting for their hair to be plaited.
ii. The percentage of time arrival can walk in straight without having to wait.

15
 iii. The percentage of customers who have to wait before getting into the stylist’s chair
3. Vehicles passing through Niger bridge at Onitsha are required to pay a toll gate fee of
N200 per entry. At average, 70vehicles pass through the toll gate per hour. The average
time for toll gate attendant to pass a vehicle through the gate is 45secs. The arrival rate
and service rate follow Poisson distribution. There is complain that vehicles wait for
long duration. The authority are willing to install one more gate to reduce the average
time to 35secs if the idle time of the toll gate is less than 9% and the average queue
length at the gate is more than 8vehicles, check whether the installation of the second
gate is justified.

MULTI-CHANNEL QUEUE MODEL (m/m/c):(GD/∞/∞)

We consider here the queue of type m/m/c. This treats the condition in which there are several
servers in parallel. This means that we have c servers where, c >1. The arrival rate λ and service
rate μ are mean value from Poisson distribution and exponential distribution respectively.
Service discipline is first come first serve. Note that arrival rate is not affected by the number of
persons in the system. That is

λ n=λ , Ɐn

However, for the service rate the following conditions apply:

1. When n< c , there is no queue because all arrivals are begin severed, and the arte of
serving will be nμ as only n channels are busy, at the rate of μ. That is
μn=nμ ,Ɐ n<c

2. When n=c , all channels will be working and when n> c , there will be n−c customers in
the queue and rate of service will be cμ as all the When c channels are busy. That is
μn=cμ , Ɐ n ≥ c

There will be three cases in this system. We then try to obtain the distribution for Pn for these
three cases.

Let,

n = number of customers in the system (waiting line + service

facility) at time t.

c = number of parallel servers ( c >1 )

λ = mean arrival rate (number of arrivals per unit time).

μ = mean service rate per busy server (number of customers served

16
 per unit of time).

λdt = probability that an arrival enters the system between t and t + dt

time interval i.e within time interval dt.

1 – λdt = probability that no arrival enters the system within interval dt plus

higher order in dt

μdt = probability of one service completion between t and t + dt

time interval i.e within time interval dt.

1 – μdt = probability that no service rendered during the interval dt plus

higher order in dt

Pn = steady state probability of exactly n customers in the system.

Pn(t) = transient state probability of exactly n customers in the system at

time t, assuming the system started its operation at time zero.

Pn+1(t) = transient state probability of having n + 1 customers in the system

at time t.

Pn–1(t) = transient state probability of having n – 1 customers in the system

at time t.

Pn(t+dt) = probability of having n customers in the system at time t + dt.

CASE 1: When n=0

This event can occur only in two exclusive and exhaustive ways as shown in the table.

17
Event No. of No. of arrivals it No. of services No. of
Customers at time dt in time dt customers at
time t time t + dt

1 0 0 0 0

2 1 0 1 0

Now let’s first find Pn ( t+ dt ). This is the sum of the probability of the two events.

Pn ( t+ dt )=P 0 ( t )( 1−λdt ) + P1 ( t ) ( 1−λdt ) μdt

¿ P0 ( t ) −P 0 ( t ) λdt + P1 ( t ) μdt

Pn ( t+ dt )−P 0 ( t )=P1 ( t ) μdt−P 0 ( t ) λdt

P n ( t +dt )−P0 ( t )
=μ P1 (t )− λ P 0 ( t )
dt

Taking the limit dt → 0

d Pn ( t )
=μ P1 ( t ) −λ P0 ( t )
dt

Considering the steady state system,

0=μ P 1−λ P0

λ
∴ P 1= P0
μ

CASE 2: When 1 ≤n ≤ c−1

When n lies between 1 and c−1 , all customers arriving will be served immediately and n
channels out of c will be busy. The event of having n customers in the system at time t+ dt will
occur in three exclusive and exhaustive ways as shown in the table below.

Event No. of No. of arrivals it No. of services No. of

Customers at time dt in time dt customers at
time t time t + dt

18
1 N 0 0 n

2 n–1 1 0 n

3 n+1 0 1 n

Thus the probability of having n customer in the system is

Pn ( t+ dt )=P n ( t )( 1−λdt ) ( 1−nμdt ) + Pn−1 ( t ) ( λdt ) ( 1−( n−1 ) μdt ) + Pn+1 ( t )( 1− λdt ) [ ( n+1 ) μdt ]

¿ Pn ( t ) ( 1−dt ( λ+nμ ) ) + Pn−1 ( t ) ( λdt ) + Pn+ 1 ( t )( n+1 ) μdt

¿ Pn ( t )−P n ( t ) dt ( λ+ nμ ) + Pn−1 ( t )( λdt ) + Pn+1 ( t )( n+1 ) μdt

Pn ( t+ dt )−P n ( t )=Pn−1 (t ) ( λdt ) + Pn +1 ( t )( n+ 1 ) μdt−Pn ( t ) dt ( λ +nμ )

Divide through by dt

P n ( t +dt )−Pn ( t )
=Pn−1 ( t ) λ+ P n+1 ( t ) ( n+1 ) μ−Pn ( t ) ( λ+nμ )
dt

Considering steady state system

0=Pn−1 λ + Pn+1 ( n+1 ) μ−Pn ( λ+nμ ) (6)

Recall,

λ
P1= P0
μ

Putting n=1 in (6), we have

P0 λ+2 P2 μ−P1 ( λ + μ )=0

λ2
P0 λ+2 P2 μ−P0 −P 0 λ=0
μ
2
λ
2 P2 μ=P 0
μ

∴ P 2= ()
1 λ 2
2 μ
P0

19
Similarly, putting n=2 in (6), we have

P3=
1 λ 3
3! μ ()
P0

In general,

()
n
1 λ
Pn= P0
n! μ

CASE 3: When n ≥ c

Pn=
1
Sn +s S ! μ()
λ n
P0

Hence,

{ ()
n
1 λ
P0
n! μ
Pn=
1
¿ n+c
c c! μ
λ n
P0 ()
While the probability of having zero customer in the system is given as

[ ]
−1

( ) ( )
n c
λ λ

( )
c−1
μ μ 1
P0= ∑ n!
+
c!
.
λ
n=0
1−
cμ

PROPERTIES OF THE MULTI-CHANNEL QUEUE MODEL

1. Expected number of customers in the system

()
c
λ
λμ
μ λ
Ls = P+
2 0
( c −1 ) ! ( cμ−λ ) μ

2. Expected number of customers in the queue

Lq= Ls−average number being served

λ λ
¿ Ls−c . =Ls −
cμ μ

20
 ()
c
λ
λμ
μ
¿ P0
( c−1 ) ! ( cμ− λ )2

3. Average time a customer spend in the system

()
c
λ
μ
Ls μ 1
W s= = P+
2 0
λ ( c−1 ) ! ( cμ− λ ) μ

4. Average waiting time of a customer in the queue

()
c
λ
μ
Lq μ
W q= = P0
λ ( c −1 ) ! ( cμ−λ )2

5. Probability that a customer has to wait.

P ( n ≥ c) =
μ
μ()
λ c
P0
( c −1 ) ! ( cμ−λ )❑

6. Probability that a customer enters the service without waiting

()
c
λ
μ
μ
1−P ( n ≥ c )= P0
( c−1 ) ! ( cμ−λ )❑

7. Utilization rate

λ
ρ=
cμ

Example:

A tax consulting firm has 3 counters in its office to receive people who have problems
concerning their income, wealth and sales taxes. On the average, 48persons arrive in an 8-hour
day. Each tax advisor spends 15 minutes on an average on arrival. If the arrivals are Poisson
distributed and service times are exponentially distributed, find:

i. The average number of customers in the system

21
 ii. Average number of customers waiting to be served
iii. Average time a customer spends in the system
iv. Average waiting time for a customer
v. The number of hours each week a tax advisor spends performing his job
vi. The probability that a customer has to wait before he gets service
vii. The expected number of idle tax advisors at any specified time.

Solution:

Given: c=3

48 6 1 4 λ 6
λ= = , μ= ×60= , = =1.5
8 hr 15 hr μ 4

First we find the probability of having no customer in the system P0.

[ ]
−1

() ()
n c
λ λ

( )
c−1
μ μ 1
P0= ∑ n!
+
c!
.
λ
n=0
1−
cμ

[ )]
−1

(
3−1
( 1.5 )n ( 1.5 )3 1
¿ ∑ + .
n=0 n ! 3! 1
1− ×1.5
3

[ )]
−1

(
2
( 1.5 )n ( 1.5 )3 1
¿ ∑ + .
n=0 n ! 3 ! 1−0.5

[ ]
2 n −1
( 1.5 )
¿ ∑ + 0.5625× 2
n=0 n !

−1
¿ [ 1+ 1.5+1.125+1.125 ]

−1
¿ [ 4.75 ] =0.210

i. Average number of customers in the system

22
 ()
c
λ
λμ
μ λ
Ls = P0 +
( c −1 ) ! ( cμ−λ ) 2
μ
3
6 × 4 × ( 1.5 )
¿ 2
×0.21+1.5
2× ( 3× 4−6 )
17.01
¿ +1.5=0.24+1.5
72
¿ 1.74 ≅ 2

ii. Average number of customers waiting to be served

()
c
λ
λμ
μ
¿ P0
( c−1 ) ! ( cμ− λ )2

3
6 × 4 × ( 1.5 )
¿ 2
×0.21
2× ( 3× 4−6 )

17.01
¿ =0.24 ≅ 0 customers
72

iii. Average time a customer spends in the system

Ls 1.74
W s= = =0.29 hrs=17.4 mins
λ 6

iv. Average waiting time for a customer

L q 0.24
W q= = =0.04 hrs ≅ 0 hrs=0 mins
λ 6

v. The number of hours each week a tax advisor spends performing his job
The question here is to find the utilization factor. Thus
λ 6
ρ= = =0.5
cμ 3 × 4
Number of hours each day a tax advisor spends doing his job
4 hrs
¿ 0.5 ×8=
day
On the average therefore, a tax advisor is busy in a 5working days a week

23
 4 ×5=20 hrs

vi. Probability that a customer has to wait

()
c
λ
μ
μ
P ( n ≥ c) = P0
( c −1 ) ! ( cμ−λ )❑
4 × ( 1.5 )3
¿ ❑ ×0.21=0.236
2× ( 3× 4−6 )

vii. The expected number of idle tax advisors at any specified time.
Here we determine the probability that three tax advisors are idle that is there are no
customers in the system, two tax advisors are idle that is there is only customer in the
system and one tax advisor is idle that is there are two customers in the system.
Thus,
Probability that there are no customers in system

[ ]
−1

() ()
λ n λ c

( )
c−1
μ μ 1
P0= ∑ n!
+
c!
.
λ
=0.21
n=0
1−
cμ

Probability of having n customers in the system

()
n
1 λ
P n= P0
n! μ
Now for
n=1, that is two tax advisors are idle
1
P1= ( 1.5 )1 × 0.21=0.315
1!

n=2, that is one tax advisor is idle

1
P2= ( 1.5 )2 ×0.21=0.236
2!
Therefore, expected number of idle advisors at any specified time
3 P0 +2 P 1+ P2
¿ 3 ×0.21+2 ×0.315+ 0.236=1.496 ≅ 2

24
 Lecture Two
Inventory Theory
An inventory consists of usable but idle resources such as men, machine, material or money.
When the resources involved are material, the inventory is called stock. An inventory problem is
said to exit if either the resources are subject to control or if there is at least one such cost that
decreases as inventory increases. The objective is to minimize total (actual or expected cost.
However, in situation where inventory affects demand, the objective may also be to maximize
profit. A business or an industry usually maintains a reasonable inventory of goods to ensure
smooth operation. Too little of inventory causes costly interruption; too much of it result in idle
capital. The inventory problem therefore determines the inventory level that balances the two
extreme cases.

INVENTORY COST
Because inventory policies affect profitability, the choice among policies depends upon their
relative profitability. Four costs that determine this profitability which are considered in
inventory models are:

1. Purchase cost
2. Inventory carrying or stock holding cost
3. Procurement cost (for bought _outs) or setup costs (for made – ins)and
4. Shortage cost (due to disservice or the customer)

PURCHASE COST
It is the price that is paid for purchasing/producing an item. It may be constant per unit or may
vary with the quantity purchased/produced. If the cost per unit is constant, it does not affect the
inventory control decision. However, the purchase cost is definitely considered when it varies as
in quantity discount situation.

INVENTORY CARRYING COST (OR HOLDING COSTS)

They arise on account of maintaining the stocks and the interest paid on the capital tied up with
the stocks. They vary directly with size of the inventory as well as the time for which the item is
held I stock. Various components of the stockholding cost are:
i. Cost of Money or Capital tied up in Inventories: This is by far, the most important
component. Money borrowed from the banks may cost interest of about 21%. But usually
the problem is viewed in a slightly different way i.e, how much the organization would
have earned, had the capital been invested in an alternative project such as developing

25
 new product, etc. it is generally taken somewhere around 15% to 20% of the value of
inventories.

ii. Cost of Storage space: This consists of rent for space. Besides space expenses, this will
also include heating, lighting and other atmospheric control expenses. Typically values
may vary from 1 to 3%

iii. Depreciation and Deterioration Cost: They are especially important for fashion items
or items undergoing chemical changes during storage. Fragile items such as crockery are
liable to damage, breakage, etc. 0.2% to 1% of the stock value may be lost due to
damages and deterioration.

iv. Pilferage cost: It depends upon the natural of the item. Valuable such as gold and
expensive tools may be more tempting, while there is hardly any possibility of heavy
casting or forging begin stolen. Pilferage cost may be taken as 1% of the stock value.

v. Obsolescence cost: It depends upon the nature of the item in stock. Electronic and
computer component are likely to be fast outdated . change indesign also lead to
obsolescence.it may be possible to quantify the percentage loss due to obsolescence and it
may be taken as 5% of the value.

vi. Handing cost: These include all cost associated with moveemont of stock, such as cost
of labour, overhead cranes, gantries and other machinery used for this purpose.

vii. Record keeping and Administrative cost: There is no use of keeping stock unless one
can easily know whether or not the required item is in stock. This signifies the need of
keeping funds for record keeping and necessary administration.

viii. Taxes and Insurance cost: Most organization have insurance cover against possible loss
from theft, fire, etc. and this may cost 1% to 2% of the invested capital.

Inventory carrying cost C 1 is expressed either as percent/unit time (e.g 20% per year) or in terms
of monetary value/unit/unit time (e.g. N20,000.00, the inventory carry cost, being, say, equal to
20%, amount to
20
20000 × =N 4000
100

PROCUREMENT COST OR SETUP COST

These include the fixed cost associated with placing of an order or setting up machinery before
starting production. They include cost of purchasing, requisition, follow up, receiving the good,
quality control, cost of mailing, telephone calls and other follow up action, salaries of persons
for accounting and auditing etc. also called order costs or replenishment costs, they are assumed
to be independent of the quantity ordered or produced but directly proportional to the number of

26
orders placed. At times, however, these costs may not bear any simple relationship to the number
of orders. More than one stock item may be ordered on one set of the documents; the clerical
staff is not divisible and without the existing staff increasing or decreasing, there may be
considerable scope for changing the number of orders. In such a case, the acquisition cost
relationship may be quadratic or stepped instead of a straight line. They are expressed in terms of
N /Order or N / Setup.

SHORTAGE COSTS OR STOCK-OUT COSTS

These costs are associated with either a delay in meeting demands or the inability to meet it at
all. Therefore, shortage costs are interpreted in two ways. In case the unfilled demand can be
filled at a later stage (backlog case), these cost are proportional to quantity that is short as well as
the delay time and are expressed as N/unit of back ordered/unit time (eg N7/unit/year). They
represent loss of goodwill and cost of idle equipment. In case the unfilled demand is lost (no
backlog case), these cost become proportional to only the quantity that is shot. These result in
cancelled orders, lost sales, profit and even the business itself.
It follows from the above discussion that if the purchase cost is constant and independent of the
quantity purchased, it is not considered in formulating the inventory control policy. The total
variable cost in this case is given by

Total Variable Inventory Cost =Carry Cost +Odering Cost +Shortage Cost .

However, if the unit cost depends upon the quantity purchased i.e, price discount are available
the purchase cost is definitely considered in formulating the inventory control policy. The total
variable inventory cost in this case is then given by

Total Variable Inventory Cost =Purchased cost+ carry cost +Odering cost + Shortage cost

THE INVENTORY GRAPH

Inventory problem can be solved graphically by plotting the relationship between the quantity of
stock held (q) and time (t). The general inventory graph is shown below:

27
 q

Instantaneous Demand
No Activity
s × ×
ou nt
t inu shme
n i
Co plen ×
R e

Replenishment
Instantaneous
Co ema
D
ns nd
tan
t
Shortage

×
Position

×
t
Figure 1: Inventory Graph

An important factor in the formulation and solution of an inventory model is that the demand
(per unit time) of an item may be deterministic (known with certainty) or probabilistic or
stochastic (describe by a probability distribution).
The demand for a product in inventory is the number of units that will need to be withdrawn
from inventory for some use (eg sales) during a specific period. If the demand in future period
can be forecast with considerable precision, it is reasonable to use an inventory policy that
assumes that all forecasts will always be completely accurate. This is the case of known demand
where a deterministic inventory model would be used. However, when demand cannot be
predicted very well, it becomes necessary to use a stochastic or probabilistic inventory model
where the demand in any given period is a random variable rather than a known constant.

SOME INVENTORY TERMINOLOGY

Here we defined some commonly used inventory terminologies.
i. Lead Time: The time between ordering goods and their replenishment.

ii. Economic Order Quantity: This is specifically derived value which gives the
external order quantity that minimizes total inventory costs.

iii. Minimum or Buffer Stock: A term used to describe the stock held to cover possible
deviations in demand or supply during lead time. It is sometimes called safety stock.
iv. Maximum Stock: A level used as an indicator above which stock is too high.

v. Reorder Level: A level of stock which when reached, signals a replenishment order.

28
 vi. Reorder Quantity: The level of a replenishment order. This is often the EOQ.

vii. Inventory Cycle: This is the part of an inventory graph which regularly repeats itself
in a cyclic process.

viii. Length of Inventory Cycle: This is simply the length of time over which an
inventory cycle extends.

ix. Average Inventory Level: This is total area under graph divide by the total time.

Total Area Uder Graph

Average Inventory Level=
Total Time
The best way to calculate the area under graph is to split it up into right-angled triangles (and/or
rectangles) and then apply the appropriate formula in each case.

Example 1:
A particular item of stock has an initial inventory of 600. A particular production line requires
the items to be drawn (continuously) from stores at a steady rate of 200 per day. As soon as a
stockout is reached, a batch of 600 items is moved in overnight from another source to replenish
the inventory. Sketch the inventory graph for a period of 90days.
Solution:

600× × ×

300

0 × × ×
3 6 9
Figure 2: Graph of Inventory

Note:

29
 1. From the graph, there is stock out after every three days. Hence we have cyclic inventory
over the whole 9days period
2. The length of inventory cycle is 3days
3. To calculate the average inventory level, we note that already the graph have right angled
triangle. Thus for each triangle:

Area=0.5 ×3 ×600=900

Total area=3 ×900=2700

Therefore,
Total Area Under Graph
Average Inventory level=
Total Time
2700
¿ =300
9
Example 2:
An inventory situation over a period of 10weeks was as follows:
Beginning of period: No initial inventory
First two weeks: Goods supplied at continuous rate of 500 per week, No withdrawals
Next four weeks: Goods withdrawn at constant rate of 250 per week. No supply
Next four weeks: required at the constant rate of 200 per week and also goods are supplied to
store continuously at 300 per week.
a) Sketch the inventory graph for the 10week period.
b) Determine the amount in store at the end of the period
c) Calculate the average inventory level for the 10weeks period

Solution:
a) The first two weeks and the next four weeks have a steady supply and demand
respectively, the last four weeks movements need combining to give a net value of
300−200=100 per week continuous supply per week. The graph is as shown below

30
Inventory
level

1000

800

600

400

200 A B
C

0 2 6 10 Weeks

b) Amount in store at the end of the 10weeks period is calculated thus:

Total demand for the last four weeks ¿ 200 × 4=800
Total supply for the last four weeks¿ 300 × 4=1200
Therefore, amount in store at the end of the 10weeks period ¿ 1200−800=400

Area of ∆ A+ Area of ∆ B+ Area of ∆C

c ¿ Average inventory level=
Total Time

¿¿¿

1000+ 2000+800 3800

¿ = =380
10 10

Definitions:

i. Reorder Level
LRO =Maximumusage per period × Maximum Lead Time∈ periods

ii. Maximum Level

31
 Lmax =Reorder Level+ EOQ−¿

iii. Minimum Level

Lmin =Reorder Level−¿

Example: Suppose for a particular inventory, we had:

i. The weekly minimum, normal and maximum usage are 600, 1000, and 1400
respectively
ii. The lead time varies between 4 to 8 weeks , an average of 6weeks.
iii. The normal ordering quantity (EOQ) is 20,000

Then calculate

a) Reorder level
b) Maximum stock level
c) Minimum stock level

Solution:

a) Reorder Level
LRO =Maximumusage per period × Maximum Lead Time∈ periods

¿ 1400 ×8=11,200

b) Maximum Level
Lmax =Reorder Level+ EOQ−¿

¿ 11200+20000−( 600× 4 )=28,800

c) Minimum Level
Lmin =Reorder Level−¿

¿ 11200− (1000 × 6 )=5200

INVENTORY MODELS WITH DETERMINISTIC DEMAND

It is extremely difficult to formulate a single general inventory model which takes into account
all variations in real systems. In fact, even if such a model were developed, it may not be
analytically solvable. Thus inventory models are usually developed for some specific situations.

In this lecture we shall deal with situations in which demand is assumed to be fixed and
completely known. Models for such situations are called economic order quantity models or
economic lot size models.

32
THE BASIC ASSUMPTIONS OF EOQ MODELS

While deriving the basic EOQ model the following assumptions were made:

i. Demand is known and uniform (Constant)

ii. Shortages are not allowed; as soon as the stock level becomes zero, it is
instantaneously replenished.
iii. Replenishment of stock is instantaneous or replenishment is infinite
iv. Lead time is zero. The moment the order is placed, the quantity ordered is received.
v. Inventory carrying cost and ordering cost per order remain constant over time. The
former is linearly related to the quantity ordered and the latter to the number of
orders.
vi. Cost of the item remains constant over time. There are no price breaks or quantity
discounts.
vii. The item is purchased and replenished in lots or batches
viii. The inventory system pertains to a single item.

CLASSICAL EOQ MODEL (DEMAND RATE UNIFORM, REPLENISMENT RATE

INFINITE)

In this model, the demand is fixed and known since goods are supplied to customers at a uniform
rate R per unit time. No shortages are allowed, consequently, the cost of shortage, C 2 is infinity.
Now if an order is placed every t time units, where t is fixed, and the ordering cost per order is
C 3. Here the lead time is zero since replenishment rate is infinite, that is instantaneous. The
holding cost is assumed to be proportional to the amount of inventory as well as the time the
inventory is held. Thus the cost of holding inventory I for time T is C 1¿, where C 1 is the cost of
holding one unit in inventory for a unit time. The costs C 1, C 2, and C 3 are assumed to be
constants. The question here are

i. How often should order be placed?

ii. How many units should be ordered in each order?

If orders are placed at interval t , a quantity q=Rt must be ordered in each order. Since the stock
in small time dt is Rtdt , the stock in time period t will be
t

∫ Rtdt= 12 R t 2= 12 qt
0

1 2
∴ Cost of holdinginventory during time t= C 1 R t
2

Ordering cost=C 3

33
 1
∴ Total Cost during time t= C 1 R t 2+C 3
2

1 C
∴ Average total cost per unit time ,C ( t )= C 1 Rt + 3 (2.1)
2 t

C will be minimized if

dC (t) 0∧d 2 C(t )

= >0
dt dt
2

Differentiating (2.1) we have

dC ( t ) 1 C3
= C 1 R− 2 =0
dt 2 t

Thus

t=
√ 2 C3
C1 R

2 C3
t is optimal since 2
>0
t

Thus, C ( t ) is minimum for optimal time interval,

t 0=
√ 2 C3
C1 R
(2.2)

Optimum quantity q 0to be ordered during each order,

q 0=R t 0=
√ 2 C3 R
C1
(2.3)

(2.3) is economic order quantity.

The resulting minimum average cost per unit time

1
C 0 ( q )= C 1
2
2 C3
C1 R √
+C 3
C1 R
2 C3 √
¿ √ 2C 1 C 3 R

Also the total minimum cost per unit time, including the cost of the item
¿ √ 2C 1 C 3 R+ CR

34
Where, C is the cost per unit of the item.

Assignment
Show that the alternative form of eqtn 2.1 is

1 C R
C ( q )= C 1 q+ 3
2 q
Corollary: In the above model if the order cost is C 3+ bq instead of being fixed where b is the
order cost per unit time, we can prove that there is no change in the optimum order quantity due
to the changed order cost.
Proof:
Note
q
q=Rt ⟹ t=
R
Thus the average cost per unit time
1 R
C ( q )= C 1 q+ ( C3 +bq )
2 q
For the minimum cost
d (C q) 0∧d 2 C (q)
= is positive
dq dq
2

ie

1
2
RC
C1 − 2 3 =0∨q=
q
2C 3 R
C1 √
Thus

q 0=
√ 2C 3 R
C1
, which isthe same as equation 2.3

Limitations of EOQ

The EOQ formula has a number of limitations. Number of objections have been raised regarding
its validity. Some of them are:

1. In practice the demand is neither known with certainty nor is it uniform. If the
fluctuation is mild, the formula can be applicable but for large fluctuations it loses its
validity. Dynamic EOQ models, instead, may have to be applied.

35
 2. The ordering cost is difficult to measure. Also it may not be linearly related to number of
orders as assumed in the derivation of the model. The inventory carrying rate is still
more difficult to measure and even to define precisely.
3. It is difficult to predict the demand. Present demand may be quite different from the past
history. Hardly any prediction is possible for a new product to be introduced in the
market.
4. The EOQ model assumes instantaneous replenishment of the entire quantity ordered. In
practice, the total quantity may be supplied in parts. EOQ model is not applicable in such
a situation.
5. Lead time may not be zero unless the supplier is next door and has sufficient stock of the
item, which is rarely so.
6. Price variations, quantity discounts and shortages may further invalidate the use of the
EOQ formula

Example 1:

A stockist has to supply 12000units of a product per year to his customer. The demand is fixed
and known and the shortage cost is assumed to be infinite. The inventory holding cost is N0.20
per unit per month and the ordering cost per order is N350. Determine

i. The economic order quantity

ii. Optimum scheduling period t 0
iii. Minimum total variable yearly cost

Solution:

12000
Supply rate R= =1000units /mont h
12

C 1=N 0.20 per unit per month C3 =N 350 per order

i .q 0=
√ √2C 3 R
C1
=
2× 350 ×1000
0.20
=1,870 units/order

ii . t 0 =
√ √2C 3
C1 R
=
2× 350
0.20× 1000
=1.87 months

iii . C 0=√ 2C 1 C 3 R=√ 2× 0.20 ×12 ×350 ×1000 ×12

¿ N 4,490 per year

Example 2:

36
A particular item has a demand of 9,000 units/year. The cost of one procurement is N100 and
the holding cost per unit is N2.40 per year. The replacement is instantaneous and no shortages
are allowed. Determine

i. The economic order quantity

ii. The number of orders per year
iii. The time between orders
iv. The total cost per year if the cost of one unit is N1

Solution:

R=900 units/ year C1=N 2.40 /unit / year C 3=N 100/ procurement

i .q 0=
√ √
2C 3 R
C1
=
2× 100 ×9000
2.40
=866 units / procurement

ii . n0=
√ √
C1 R
2 C3
=
2.40 × 9000
2 ×100
=√ 108=10.4 orders / year .

1 1
iii . t= = =0.0962 years=1.15 months between procurement
n0 10.4

iv . C0 =√ 2C 1 C 3 R+CR =√2 ×2.40 × 100× 900+9000 ×1

¿ 9000+2080
¿ 11080/ year

Lecture Three
Simulation

37
Simulation is an imitation of reality. A simple illustration is the testing of an aircraft model in a
wind tunnel from which we determine the performance of the actual aircraft under real operating
conditions.

Another example of simulation is planetarium which represents the simulation of the planet
system. The examples above try to imitate the reality to see what might happen under real
operating conditions. This imitation of reality which may be in the physical form or in the form
of mathematical equations may be called simulation.

For the complex and intricate problems of managerial decision making, the analogue simulation
may not be practicable and actual experimentation with the system may be uneconomical. Under
such circumstances, the complex system is formulated into a mathematical model for which a
computer program is developed and the problem is solved by using high speed electronic
computer and hence it is named a computer simulation or system simulation.

Application of Simulation

Simulation is quite versatile and commonly applied technique for solving decision problems. It
has been applied successfully to a wide range of problems of science and technology as given
below:

1. In the field of basic sciences, it has been used to evaluate the area under curve, to
estimate the value of π in matrix inversion and study of particle diffusion.

2. In industrial problems including the shop floor management, design of computer systems,
design of queue system, inventory control, communication networks, chemical processes,
nuclear reactors and scheduling of production processes.

3. In business and economic problems, including customer behaviour, price determination,

economic forecasting, portfolio selection and capital budgeting.

4. In social problems, including population growth, effect of environment on health and

group behaviour.

5. In biomedical system including fluid balance distribution of electrolyte in human body

and brain activities.

6. In the design of weapon systems, war strategies and tactics.

7. In the study of projects involving risky investments

Monte Carlo Simulation

38
The technique employs random numbers and is used to solve problems that involve probability
and wherein physical experimentation is impracticable and formulation of mathematical model is
impossible. It is a method of simulation by sampling technique. The steps involve in carrying
out Monte Carlo simulation are:
1. Select the measure of effectiveness (objective function) of the problem. It is either to be
maximized or minimized. For example, it may be idle time of service facility in a queue
theory or number of shortage or the total inventory cost in an inventory control problem.

2. Identify the variables that affect the measure of effectiveness significantly. For example,
number of service facilities in a queuing problem or demand, lead time and safety stock
in inventory control problem.

3. List all possible outcomes of the experiment

4. Determine the probability of each outcome

5. Setup a correspondence between the outcomes of the experiment and the random
numbers.

6. Repeat the experiment and tally the outcomes

7. Compute any statistics and state the conclusion.

Example:
Customers arrive at a service facility to get the required service. The inter-arrival and service
times are constant and 1.8minutes and 4minutes respectively. Simulate the system for
14minutes. Determine the average waiting time of a customer and idle time of the service
facility.

Solution:

This is the case where simulation technique which involves repetitive experimentation can be
used to analyse a problem where variables are constant and not probabilistic in nature.

The arrival times of customers within 14minutes period will be:

Customer: A B C D E F G H

Arrival Time(Mins): 0 1.8 3.6 5.4 7.2 9.0 10.8 12.6

The time at which the service begins and ends within time period of 14mins is shown below.
Waiting time of customers and idle time of service facility are also calculated.

Customer Service Waiting Time of Idle Time of

39
 Begins Ends Customer Service Facility

A 0 4 0 0

B 4 8 4 – 1.8 = 2.2 0

C 8 12 8 – 3.6 = 4.4 0

D 12 16 12 – 5.4 = 6.6 0

E 14 14 – 7.2 = 6.8 0

F 14 14 – 9.0 = 5.0 0

G 14 14 – 10.8 = 3.2 0

H 14 14 – 12.6 = 1.4 0

0+ 2.2+ 4.4+ 6.6+6.8+5.0+3.2+1.4

∴ Average waiting time of a customer=
8

29.6
¿ =3.7 mins
8

Idle time of facility = nil.

The example that follows involves variables that are probabilistic in nature.

Example:

A town has six wards and they contain 170, 510, 640, 75, 250, and 960 houses respectively.
Make a random selection of 8 houses using the table of random numbers. Example the
procedure adopted by you.

Solution:

Since the total number of houses is 2,605 hence random numbers range 0 – 2604 are allocated in
proportion to the number of houses in each of the six wards shown in the table below:

40
 Allocation of Random Number to The Houses

1 2 3 4 5

Ward No. No. of Houses Cumulative No. Range Random Nos.

of Houses Fitted

1 170 170 0 – 169

2 510 680 170 – 679 0590, 0354

3 640 1320 680 – 1319 1128, 0764, 1292

4 75 1395 1320 – 1394 1340

5 250 1645 1395 – 1644

6 960 2605 1645 – 2604 2181, 1749

The first random number picked up from the random number table is 2181. Since it lies within
the interval 1648 – 2604, it is fitted against ward number 6 in column 5. The next random
number is 1128, which lies in the interval 680 – 1319 and therefore fitted against ward number 3
in column 5. The next random number 7112 is greater than 2604 in column 4 and therefore
dropped from consideration. In this manner the following random number are either fitted in
column 5 or dropped. Thus using D for dropped and F5 for fitted in column 5 we have the
following:

6557(D), 4199(D), 3545(D), 1749(F5), 9103(D), 0764(F5), 3493(D), 1292(F5), 4397(D),

3807(D), 4984(D), 1340(F5), 0590(F5), 9566(D), 7615(D), 8508(D), 6970(D), 5799(D),
6343(D), 4165(D), 0354(F5).

We stop because 8 houses have been selected. The eight houses belong to ward number 6, 3, 6,
3, 3, 4, 2 and 2 respectively.

41