VIDEO LECTURE ON

QUEUeING THEORY
Part I
(Introduction, single server
model)

By
Dr.P.Ilamathi,
GCT, Coimbatore
 Introduction

 The basis of waiting line analysis is the trade-off between the cost of
improving service and the costs associated with making customers
wait.

 Queuing analysis is a probabilistic form of analysis.

 The results are referred to as operating characteristics.

 Results are used by managers of queuing operations to make

decisions.
 History of queueing theory
• The theory of queueing systems was developed to provide models for
forecasting behaviors of systems subject to random demand.

• In 1909, Agner Krarup Erlang, a Danish telecommunication engineer

started applying principles of queuing theory in the area of
telecommunications. He addressed the problem concerned with
congestion of telephone traffic.

• Erlang observed that a telephone system can be modeled by Poisson

customer arrivals and exponentially distributed service times.
 Definition of a queueing system
Departure of
Customer served
arrivals customers

Departure of
impatient customers

• A queueing system can be described as follows:

“Customers arrive for a given service, wait if the service cannot start
immediately and leave after being served“
• The term "customer" can be men, products, machines, ...
 Major Constituents of a queuing system

 Customer: The arriving unit that requires some service to be performed.

 Queue (Waiting line) : The number of customers waiting to be served.
The queue does not include the customer(s) being serviced.
 Service Channel: The process or facility which is performing the services
to the customer. This may be single or multi channel.
 Elements of a Queueing system
1. Arrival 2. Service time
4. Service
distribution distribution
discipline

5. Max no. of
6. Customer Customers 7. Customer’s
Population behaviour 3. Number of
servers
 Elements of a Queuing System
(Structure of a Queueing System)
 Arrival distribution – Poisson distribution – Mean arrival rate λ.
 Service distribution – Exponential distribution – Mean service rate μ.
 Service channel – Single server model or multi server model.
 Service discipline – FCFS (FIFO), LCLS (LIFO), SIRO, Priority
 Maximum number of customers allowed in the system – Finite or infinite.
 Calling source or population - Finite or infinite.
 Customer’s behaviour – Balked, Reneged, and Jockeying customers.
 Balked – If a customer decides not to enter the queue since it is too long.
 Reneged – If a customer enters the queue, but after sometime loses his
patience and leave it.
 Jockeying – When there are two or more parallel queues and the customers
move from one queue to the other.
 Characteristics of Poisson distribution
 Probability of n arrivals in time t =

 The probability density function of the inter-arrival time (time interval

between two consecutive arrivals)
 Operating Characteristics
 Queue Length (Lq)
 System Length (Ls)
 Waiting time in the Queue (Wq)
 Total time in the system (Ws)
 Utilization factor (ρ ) - Traffic intensity
CLASSIFICATION OF QUEUEING
SYSTEMS
Kendall’s Notation for Queuing Models
 ( a / b /c ) : (d / e / f)
 where,
 a - arrival (or inter-arrival) distribution (M /Ek / GI / D)
 b - departure (or service time) distribution (M /Ek / G / D)
 c - no. of parallel service channels ( 1/C)
 d - service discipline (FIFO, LIFO, SIRO, GD)
 e - max. no. of customers allowed in the system (N/ ∞)
 f - calling source or population (N/ ∞)
 Queueing Models - classifications
Model I : (M/M/1) : (FIFO / ∞ / ∞)
Model II : (M/D/1) : (FIFO / ∞ / ∞)
Model III : (M/M/1) : (FIFO / N / ∞)
Model IV : (M/M/C) : (FIFO / ∞ / ∞)
Model I: (M/M/1) : (FIFO / ∞ / ∞ )
 Single-Server Model

 Assumptions of the basic single-server model:

 An infinite calling population
 A first-come, first-served queue discipline
 Poisson arrival rate
 Exponential service times
 Symbols:
 = the arrival rate (average number of arrivals/time period)
 = the service rate (average number served/time period)
 The arrival rate is less than the service rate ( < ) or an infinitely
large queue will build up.
 Little’s Law

where,
 L – the average number of items in a queuing system
 λ – the average number of items arriving at the system per unit of time
 W – the average waiting time an item spends in a queuing system
 (M/M/1) : (FIFO / ∞ / ∞ ) Formulae

1. Probability that no customers are in the queuing system / Probability of

an empty or idle system:
P0   1 
 
 
 
2. Probability that there are n customers are in the system:
n n
Pn    
    



 P

0
  





1  
     

3. Average (expected) number of customers in system:

4. Average (expected) number of customer in the queue:

Lq    2

     
 (M/M/1) : (FIFO / ∞ / ∞ ) Formulae
5. Average (expected) number in non-empty queue (length of queue that is
formed from time to time)

Lq
6. Average time customer spends waiting in the queue: Wq 

Wq  
     
7. Average time customer spends in the system (waiting and being served):

8. Average (expected) waiting time of a customer in non-empty queue

(average waiting time of a customer who has to wait)
 (M/M/1) : (FIFO / ∞ / ∞ ) Formulae
9. Probability of queue length being greater than or equal to k.
10. Probability of queue length being greater than k.

11. Variance of the queue length:

12. Probability density function of waiting time distribution:

13. Probability density function of (waiting + service) time distribution:

 (M/M/1) : (FIFO / ∞ / ∞ ) Formulae
14. Probability that a service channel is busy (traffic intensity or utilization
factor)

15. Probability that server is idle:

16. Probability that the queue is non-empty:

 GATE Qn
 In a single-channel queuing model, the customer arrival rate is 12 per
hour and the serving rate is 24 per hour. The expected time that a
customer is in queue is _______ minutes.
 λ = 12 per hour,
 μ = 24 per hour,
= (0.5/12) hour = (30/12) min = 2.5 min

Lq  2 = (12 x 12)/24(24-12) = ½ = 0.5

     

 Answer : 2.5 minutes

 GATE Qn
 In the notation (a/b/c) : (d/e/f) for summarizing the characteristics of
queueing situation, the letters ‘b’ and ‘d’ stand respectively for

(A) service time distribution and queue discipline

(B) number of servers and size of calling source

(C) number of servers and queue discipline

(D) service time distribution and maximum number allowed in system

 Answer : (A) service time distribution and queue discipline

 GATE Qn
 The jobs arrive at a facility, for service, in a random manner. The
probability distribution of number of arrivals of jobs in a fixed time
interval is
(A) Normal
(B) Poisson
(C) Erlang
(D) Beta
 Answer : (B) Poisson
 GATE Qn
 Jobs arrive at a facility at an average rate of 5 in an 8 hour shift. The
arrival of the jobs follows Poisson distribution. The average service time
of a job on the facility is 40 minutes. The service time follows
exponential distribution. Idle time (in hours) at the facility per shift will
be
λ = 5 per 8 hour,
(A) 5/7
μ = 1/40 per min = 60/40 per hour
(B) 14/3 = (3/2) x 8 per 8 hour = 12 per 8 hour
(C) 7/5
(D) 10/3 Proportion of time facility is idle =
= 7/12
Idle time (in hours) at the facility per shift = 7/12 x 8
= 14/3
 Answer : (B) 14/3
 GATE Qn
 At a work station, 5 jobs arrive every minute. The mean time spent on
each job in the work station is 1/8 minute. The mean steady state
number of jobs in the system is _______

 λ = 5 per min,

 μ = 1/8 min = 8 per min

= 5/3 = 1.67 jobs

 Answer : 1.67
 GATE Qn
 Customers arrive at a ticket counter at a rate of 50 per hr and tickets are
issued in the order of their arrival. The average time taken for issuing a
ticket is 1 min. Assuming that customer arrivals form a Poisson process
and service times are exponentially distributed, the average waiting
time in queue in min is

(A) 3 λ = 50 per hour,

(B) 4 μ = 1 per min = 60 per hour,

(C) 5 = (25/6) x (1/50) hr = (1/12) hr = 5 min

(D) 6
Lq    2
= (50 x 50)/60(60-50) = 25/6
     

 Answer : (C) 5
 GATE Qn

 Little’s law is relationship between

(A) stock level and lead time in an inventory system
(B) waiting time and length of the queue in a queuing system
(C) number of machines and job due dates in a scheduling problem
(D) uncertainty in the activity time and project completion time

 Answer : (B) waiting time and length of the queue in a queuing system
 GATE Qn
 In an M/M/1 queuing system, the number of arrivals in an interval of
length T is a Poisson random variable (i.e. the probability of there
being n arrivals in an interval of length T is

 The probability density function f(t) of the inter-arrival time is given by

 Ans:
 GATE Qn
 Cars arrive at a service station according to Poisson’s distribution with a
mean rate of 5 per hour. The service time per car is exponential with a
mean of 10 minutes. At steady state, the average waiting time in the
queue is
λ = 5 per hour,
(A) 10 minutes
μ = 1/10 per min = 6 per hour,
(B) 20 minutes
(C) 25 minutes = (25/6) x (1/5) hr = (5/6) hr = 50 min
(D) 50 minutes
Lq    2
= (5 x 5)/6(6-5) = 25/6
     

 Answer : (D) 50 minutes

 Tutorial

 A person requires 10, 12, and 12 units chemicals

A, B and C respectively for his garden. A liquid
product contains 5, 2 and 1 units of A,B and C
respectively per jar. A dry product contains 1,2
and 4 units of A,B and C per carton.
 If the liquid product sells for Rs.3 per jar and the
dry product sells for Rs.2 per carton, formulate
this as an LPP to determine how many of each
should be purchased, in order to minimize the
cost and meet the requirements?
VIDEO LECTURE ON

QUEUeING THEORY
Part II
(Problems - single server
model)

By
Dr.P.Ilamathi,
GCT, Coimbatore
 Example 1
 Arrival at a telephone booth are considered to be Poisson, with an
average of 9 minutes between one arrival and the next. The length of a
phone call is assumed to be distributed exponentially, with mean 3
minutes.
i) What is the probability that a person arriving at the booth will have
to wait?
ii) What is the average length of the queues that form from time to
time?
iii) The telephone company will install a second booth when convinced
that an arrival would expect to have to wait at least four minutes for the
phone. By how much must the flow of arrivals be increased in order to
justify a second booth?
iv) What is the probability that an arrival will have to wait for more than
10 mins before the phone is free?
v) What is the probability that an arrival will have to wait for more than
10 mins before the phone is available and call is also complete?
 Solution
 Arrival at a telephone booth are considered to be Poisson, with an
average of 9 minutes between one arrival and the next. The length of a
phone call is assumed to be distributed exponentially, with mean 3
minutes.
 λ = 1/9 per min
 μ = 1/3 per min
 Probability that a person arriving at the booth will have to wait:

 Average length of the queues that form from time to time:

 Increase in arrival rate to justify the installation of second telephone

booth:
 Increase in arrival rate of arrivals =
 Probability that an arrival will have to wait for more than 10 mins before
the phone is free:
 Probability that an arrival will have to wait for more than 10 mins before
the phone is available and call is also complete:
 Example 2
 Workers come to tool room to receive special tools for accomplishing a
particular project assigned to them. The average time between two
arrivals is 60 seconds and the arrivals are assumed to be in Poisson
distribution. The average service time of the tool room attendant is 40
seconds. Determine,
i) average queue length,
ii) average length of non empty queues,
iii) average number of workers in system including the worker being
attended,
iv) mean waiting time of an arrival,
v) average waiting time of an arrival who waits,
vi) whether to go in for an additional tool room attendant which will
minimize the combined cost of attendant’s idle time and the cost of
worker’s waiting time. Assume the wages of a skilled worker as Rs.400/-
per hour and that of the tool room attendant as Rs.75/- per hour.
 Solution
 λ = 1/60 per sec = 1 per min
 μ = 1/40 per sec = 1.5 per min
i) Average queue length:
= 4/3 workers
ii) Average length of non empty queues:
= 3 workers
iii) Average number of workers in system including the worker being
attended:
= 2 workers
iv) Mean waiting time of an arrival:
= 4/3 minutes
v) Average waiting time of an arrival who waits:
= 2 minutes
vi) Decision on additional tool room attendant:
Probability that the tool room attendant remains idle =
= 1/3
Idle time cost of one tool room attendant = 1/3 x 8 x 75 = Rs. 200/-
Waiting time cost of workers = Wq x no. of workers arriving per day x
wages of worker
= (4/3 x 1/60) x (8 x 60) x 400 = Rs. 4266.67/-
As the waiting time cost is much higher than the idle time cost, it is justified
to employ an additional tool room attendant.
 Example 3
 Customers arrive at a one window drive in bank according to a Poisson
distribution with mean 10 per hour. Service time per customer is
exponential with mean 5 mins. The space in front of the window,
including that for the serviced car can accommodate a maximum of
three cars. Other cars can wait outside this space.
 What is the probability that an arriving customer can drive directly to
the space in front of the window?
 What is the probability that an arriving customer will have to wait
outside the indicated space?
 How long an arriving customer expected to wait before starting service?
 Solution
 λ = 10 per hour
 μ = 1/5 per min = 12 per hour

Probability that an arriving customer can drive directly to the space in front
of the window:
= P0 + P1 + P2

= 0.42
Probability that an arriving customer will have to wait outside the indicated
space:
= 1- 0.42 = 0.58
Arriving customer expected to wait before starting service:

= 0.417 hour = 25 mins

 Example 4
 Customers arrive at a bank counter handled by a single cashier according
to Poisson distribution with mean arrival rate 6 customers/hour. The
customer attends the customer on first come first served basis at an
average rate of 10 customers/hour with the service time exponential
distribution. Find
 the probability of the number of arrivals (0 through 5) during i) 15
minute interval ii) 30 minute interval.
 the probability that the queueing system is idle.
 the probability associated with the number of customers (0 through 5) in
the queueing system.
 the time the customer should expect to spend in the queue.
 the time the customer spends before leaving the bank counter.
 Solution
 λ = 6 customers per hour
 μ = 10 customers per hour
 Probability of the number of arrivals (0 through 5) during i) 15 minute
interval ii) 30 minute interval:

i) t = 15 min = ¼ hour

ii) t = 30 min = ½ hour

 Probability that the queueing system is idle:
= 0.4
 Probability associated with the number of customers (0 through 5) in the
queueing system:
Time the customer should expect to spend in the queue:

= 0.15 hour = 9 mins

Time the customer spends before leaving the bank counter:

= 0.25 hour = 15 mins

 Assignment Qn
 Goods truck arrive randomly at a stockyard with a mean of 8
trucks/hour. A crew of four operators can unload a truck in 6 minutes.
Trucks waiting time in queue to be uploaded are paid a waiting charge at
the rate of Rs. 600 per hour. Operators are paid a wage rate of Rs. 200
per hour. It is possible to augment the crew strength to 2 or 3 (of four
operatives per crew) when the unloading time will be 4 minutes or 3
minutes respectively per truck. Find the optimal crew strength.
 Tutorial

 A person requires 10, 12, and 12 units chemicals

A, B and C respectively for his garden. A liquid
product contains 5, 2 and 1 units of A,B and C
respectively per jar. A dry product contains 1,2
and 4 units of A,B and C per carton.
 If the liquid product sells for Rs.3 per jar and the
dry product sells for Rs.2 per carton, formulate
this as an LPP to determine how many of each
should be purchased, in order to minimize the
cost and meet the requirements?
VIDEO LECTURE ON

QUEUeING THEORY
Part III
(single server with constant
service model)

By
Dr.P.Ilamathi,
GCT, Coimbatore
Model II: Single server with
Constant Service Times
(M/D/1):(FCFS/ ∞ / ∞ )
 Single server with
Undefined and Constant Service Times
 Constant, rather than exponentially distributed service times, occur
with machinery and automated equipment.
 Constant service times are a special case of the single-server model
with undefined service times.
 Queuing formulas for the undefined service time model:
2
    /
2 2 



Lq    
2 1  /  

Lq
Wq 

 Single server with
Undefined Service Times - Example
 Data: Single fax machine; arrival rate of 20 users per hour, Poisson
distributed; undefined service time with mean of 2 minutes, standard
deviation of 4 minutes.
 Operating characteristics:
λ = 20 users per hour,
μ = ½ per min = 30 per hour,
 = 4 min = 1/15 hour
 Single server with
Undefined Service Times - Example
 Single server with
Constant Service Times - Formulae
 In the constant service time model there is no variability in service
times;  = 0.
 Substituting  = 0 into equations:

2 2 2
 2 2    /    202    /    /  
 
 2
Lq      
2 1  /   2 1  /   21  /   2     

 All of the remaining formulas are the same as the single-server

formulas.

Lq
Wq 

 Single server with
Constant Service Times - Formulae
 Car wash servicing one car at a time; constant service time of 4.5
minutes; arrival rate of customers of 10 per hour (Poisson distributed).
 Determine average length of waiting line and average waiting time.
 = 10 cars per hour,
 = 60/4.5 = 13.3 cars per hour

Lq   2
 (10)2
 1.14 cars waiting
2(   ) 2(13.3)(13.3 10)
Lq 1.14
Wq    0.114 hour or 6.84 minutes waiting time
 10
 Tutorial

 A person requires 10, 12, and 12 units chemicals

A, B and C respectively for his garden. A liquid
product contains 5, 2 and 1 units of A,B and C
respectively per jar. A dry product contains 1,2
and 4 units of A,B and C per carton.
 If the liquid product sells for Rs.3 per jar and the
dry product sells for Rs.2 per carton, formulate
this as an LPP to determine how many of each
should be purchased, in order to minimize the
cost and meet the requirements?
 VIDEO LECTURE ON

QUEUeING THEORY
Part IV
(single server with finite capacity
model, Multi server model)

By
Dr.P.Ilamathi,
GCT, Coimbatore
Model III: (M/M/1) : (FIFO / N / ∞ )
 (M/M/1) : (FIFO / N / ∞ ) Formulae
 In a finite queue, the length of the queue is limited.
 N is the maximum number of customers in the system.
1. Probability that there is no customer in the system:

2. Probability that there are ‘N’ customers in the system:

3. Average number of customers in the system:

4. Average number of customers in the queue:

 (M/M/1) : (FIFO / N / ∞ ) Formulae
5. Average time a customer spends in the system:

6. Average waiting time in the queue:

 Example
 At a railway station only one train is handled at a time. The railway is
sufficient only for two trains to wait while the other is given signal to
leave the station. Trains arrive at the station at an average rate of 6 per
hour and the railway station can handle them on an average of 12 per
hour.
 Assuming Poisson arrivals and exponential service distribution, find the
steady state probabilities for the various number of trains in the system.
Also find the average number of trains at the railway station and the
average waiting time of a new train coming into the yard.
 Solution:
 λ = 6 per hour
 μ = 12 per hour
 N=3
Probability that there is no customer in the system:

= 0.53

Steady state probabilities for the various number of trains in the system:

Average no. of trains in the system:

= 0.723

Average waiting time of a new train coming into the yard:

= 0P1 + 1P2+2P3 = 0.262

Average waiting time of a new train coming into the yard:

Wq = 0.262 / 5.604 = 0.0467 hours = 2.81 minutes

Model IV: (M/M/C) : (FIFO / ∞ / ∞ )
 Multiple-Server model
 Assumptions:
 First-come first-served queue discipline
 Poisson arrivals, exponential service times
 Infinite calling population.
 Parameter definitions:
  = arrival rate
  = the service rate per server (channel)
 c = number of servers
 c  = mean effective service rate for the system (must exceed arrival
rate)
 (M/M/C) : (FIFO / ∞ / ∞ ) Formulae
 Probability that there is no customers in the system:

 Probability that there are ‘n’ customers in the system:

 Average number of customers waiting in the queue:

 (M/M/C) : (FIFO / ∞ / ∞ ) Formulae
 Average (expected) number of customers in system:

 Average time customer spends waiting in the queue:

 Average time customer spends in the system (waiting and being served):

 Probability that a customer has to wait:

 Probability that a customer enters the service without waiting:

 (M/M/C) : (FIFO / ∞ / ∞ ) Formulae
 Average number of idle servers = c – average number of customers served.
 Utilization rate:

 Efficiency of M/M/C model:

 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 48 persons arrive in an 8-hr day. Each tax adviser spends 15
mins on an average on an arrival. If the arrivals are Poissonly distributed
and service times are according to exponential distribution, Find
i) average number of customers waiting to be served
ii) average number of customers in the system
iii) average time a customer spends in the system
iv) average waiting time for a customer
v) number of hours each week a tax adviser spends performing his job
vi) probability that a customer has to wait before he gets service
vii) expected number of idle tax advisers at any specified time.
 Solution
 λ = 48/8 = 6 per hour
 μ = 1/15 per min = 4 per hour
 C=3

= 0.21
 Average number of customers waiting to be served:

= 0.24 customers

 Average number of customers in the system:

= 1.74 customers
 Average time a customer spends in the system:
= 17.4 mins
 Average waiting time for a customer:
= 2.4 mins
 Number of hours each week a tax adviser spends performing his job:
Utilization factor =
= 1/2
Number of hours each day a tax adviser spends performing his job
= ½ x 8 = 4 hours/day
Number of hours each week a tax adviser spends performing his job
= 4 x 5 = 20 hours / week
 Probability that a customer has to wait before he gets service

= 0.236
 Expected number of idle tax advisers at any specified time:

Expected number of idle tax advisers at any specified time: 1.5 tax advisors
Assignment Qn
 Tutorial

 A person requires 10, 12, and 12 units chemicals

A, B and C respectively for his garden. A liquid
product contains 5, 2 and 1 units of A,B and C
respectively per jar. A dry product contains 1,2
and 4 units of A,B and C per carton.
 If the liquid product sells for Rs.3 per jar and the
dry product sells for Rs.2 per carton, formulate
this as an LPP to determine how many of each
should be purchased, in order to minimize the
cost and meet the requirements?