Queueing Theory

1
2
3
4
Some more examples

• Sale of train tickets

• Patients waiting at doctor’s clinic
• Customers at bank
• Aircraft awaiting landing
• Ships awaiting berthing
• Letters on a typist’s desk

5
6
7
8
9
What causes Queues?

Variations in Arrival rate and Service rate causes queues.

• Arrival Variation
• Service Variation

Approaches to Modelling Queues:

• Analytical Model (differential equations)
• These are used in Simple Cases for e.g. Fast food joint queues

• Simulation (Complex Systems with multiple stations)

• Low Observable (LO) Aircraft Maintenance Systems (F22 & F35)

Why Model Queues?

• To Balance Costs
• Service Costs: Cost of Providing Service
• Waiting Costs: Cost of not providing service
• If Waiting Costs become high the following happens
• Bulk – Leave without buying because the line is too long
• Buy and Never return
10
11
Let us take an example of coffee Shop baristas

#. Of Baristas
1 2 3 4
Average wating time 7 4 3 2

Wating Costs = (Average #,. Of Customers X

Average Waiting Cost)
$350 $200 $150 $125
Assume 5 customers @
US$10 per person per minute
Service Cost per Barista $60 #120 $180 $240
Total expected cost $410 $320 $330 $365
My Objective is to reduce cost. So I choose 2 Baristas

A barista is an espresso machine “coffee artist” who has extensive knowledge about coffee and
prepares, decorates and serves drinks to the customer. Barista (m/f) is the Italian word for
barkeeper.
12
What do we need to model a Queue?
• Arrival Rate
• Service Rate
• # of servers
• Maximum Queue length
• Population size

Types of Queues
• Single Server : Single Phase (McDonald’s / ATMs etc.)
• Single Server : Multiphase (Subway / Chipotle)
• Multiple Server : Single Phase (Chick-fil-A / Chick King)
• Multiple Server : Multiphase (Laundromat Washing m/cs)

Queueing Assumptions
• First In First Out (FIFO)
• No Bulking or Reneging
• Arrivals are Independent
• Service times are Independent
• Service systems operate under steady ongoing conditions, Arrival and Service rates
remain stable
13
Single Server : Singlephase Single Server : Multiphase

Multiple Server : Singlephase Multiple Server : Multiphase

14
15
16
17
18
19
Kendall's Notation is a system of notation according to which the various characteristics
of a queuing model are identified. probability distribution of the interarrival time.
probability distribution of the service time. number of servers in the system.

Kendall (Kendall, 1951) has introduced a set of notations which have become standard
in the literature of queuing models. A general queuing system is denoted by (a/b/c):
(d/e/f) where

a|b|c : d|e|f }

a = probability distribution of the inter-arrival time (Arrival time distribution)

b = probability distribution of the service time (Service time distribution)
c = number of servers in the system (Number of Service channels)
d = Queue discipline (Service discipline)
e = Maximum number of customers allowed in the system
f = Calling source or Population

20
Grocery Store

21
22
Traffic Intensity
It is the ratio of average arrival rate to the average service rate

ρ=
Where  arrival rate
 service rate
s  no of servers
i) ρ < 1, customer will be serviced faster than the arrival rate
ii) ρ = 1, customer will be arriving as faster as they can be
serviced
iii) ρ > 1, customer will be arriving faster than they can be
serviced as such as the queue will go on building up

23
24
25
26
27
28
29
Model – 1 (Single Server Queuing model)

Probability that there is no one in the system (no one in the queue and no one being served)
Po = 1-ρ = 1 -
µ
Probability that the customer will have to wait in the queue (Probability of service facility being busy)
Pw = 1- Po = 1- (1- ) =
µ µ
Probability that there are n customers in the queue system is given by
Pn = (1 - ) ( )n
µ µ
Average (Expected) number of customers in the queue, also called queue length
Lq =
µ(µ )
Average (Expected) number of customers in the system, also called system length
Ls =
µ
Average waiting time in the queue is given by
Wq =
µ(µ )
Average time in system is given by
Ws =
µ
Probability that the number of customers in the system is greater than or equal to k.
P(n> k) = ( )k
µ
P(n> k) = ( )k+1
µ
30
Question 1
Service rate is 4min/workmen and arrival rate is 20 workers per
100 minutes. Determine a) Lq b) Wq c) Po
Lq =
µ(µ )
Solution:
 Arrival rate( ) = (20/100) * 60 = 12 Wq =
µ(µ )
 Service Rate( ) = (1/4) * 60 = 15
a) Lq = 3.2 Po = 1-ρ = 1 -
µ
b) Wq = 0.26
c) Po = 1 - = 0.2 = 20%
µ

31
Question 2
The service counter is operated one shift per day of 8 hrs. Service rate is 40 minutes per customer and
arrival rate is 80 minutes per customer. Determine a)Pw b)Lq c)Wq d) Find 𝜆 if utilisation rate is
75%

Solution:
1 shift = 8 hrs
Arrival rate (𝜆) = 1/80*480 = 6 Pw = 1- Po = 1- (1- )
µ
Service rate (µ) = 1/40 * 480 = 12
a) Pw = 0.5 =50% =
µ

b) Lq = 0.5
Lq =
µ(µ )
c)Wq=0.083
Wq =
d) ρ =75% µ(µ )

ρ=
µ

75/100 =
𝜆=9 32
Model – 2 (Multiple Server Queuing model)

 Probability that there is no customer in the system

µ
Po =[ {∑ ) }+ ) ]
! (µ ! (µ µ

 Probability that there are n customers in the system.

Pn = (µ)n . Po
 Average number of customers in the queue is given by
µ(µ)
Lq = !(
µ
)
Po

 Average number of customers in the system is given by

L s = Lq +
µ
 Average waiting time in the queue
µ(µ)
Wq = !(
µ
)
Po

 Average waiting time in the system

Ws = Wq +
µ
33
Question 1
Service rate is 24min/customer and arrival rate is 40 customers per 8 hrs
per day. Number of servers is 3. Determine a)Lq b)Wq c) Ls d)Ws

Solution:
 Arrival rate( ) = 40
 Service Rate( ) = (1/24) * 8*60 = 20
 Number of servers (S) = 3
Po = 0.11
a) Lq = 0.88
b) Wq = 0.022
c) Ls = 2.88
d) Ws = 0.072

34
Question 2
Service rate is 9 customers per hour and arrival rate is 20 customers per hour.
Number of servers is 3. Determine a)Lq b)Wq c) Ls d)Ws

Solution:
 Arrival rate( ) = 20
 Service Rate( ) = 9
 Number of servers (S) = 3
Po = 0.078
a) Lq = 1.567
b) Wq = 0.0783
c) Ls = 3.767
d) Ws = 0.1883

35
36
37
38
39
40