•BU275: Business

Decision Models –
Additional Examples

• Instructor – Dr Qiang
Li
Winter 2024
February 7/8, 2024
Queuing Models –
Problem Solving
Queuing Model Examples

In all examples given in the following few

slides,
assume that
(1) arrivals occur according to a Poisson
process;
(2) service times are exponentially
distributed.
Example 1

A huge hardware store has a single counter designated as a counter for

“returns”.
Customer arrivals to the counter can be adequately modelled by a Poisson
process with an average of 1 person arriving every 3 minutes.
There is one customer service rep at the counter who takes back the item,
inspects it and gives customers their money. Service time follows an
exponential distribution, with an average of 2 minutes to serve a customer.
The cost of employing one server is $20 per hour (including EI, benefits etc.).
The store management estimates the cost per minute that a customer is in
the store is 30 cents.
There are 4 possible alternatives that the management team of the hardware
store is considering. Determine which is most cost effective.
Example 1 Contd.
1) Do NOT change current status.
2) Employ one more customer service rep. Assume that this rep has the
same service time distribution as the original rep. A single waiting
line will be formed in front of these two counters and the customer at
the head of the line will be served by the first available server.
3) Employ one more customer service rep. Assume that this rep has the
same service time distribution as the original rep. The management
team doesn’t want other customers entering the store to get the idea
that there were a lot of dissatisfied customers if they see a long line at
“returns”. In order to manage the optics better, a waiting line is
formed in front of each customer service agent.
4) A second person is added who helps the existing customer service rep
process the returns by doing the inspection of the returned items. The
original rep will now just deal with the customer and process the
refund. The two employees working as a team reduces the average
service time to 1.5 minutes. The new person added to help the
Example 1 Contd.

Queue type λ µ L Lq
M/M/1 20 30 2 1.33
M/M/1 10 15 2 1.33
M/M/1 10 30 0.5 0.167
M/M/1 20 40 1 0.5
M/M/2 20 30 0.75 0.083
M/M/2 20 60 0.34 0.01
Analysis Draft
Current system

Customers: Poisson process;

average of 1 person arriving every 3 minutes. =20 per hour
One server: exponential distribution; M/M/1
average of 2 minutes to serve a customer. per hour

Cost of employing one server is $20 per hour

Waiting cost: per minute that a customer is in the store is 30 cents.

Including Waiting cost=(0.30)(60)L

service time Total costs = 20+(0.30)(60)(2)=56
Analysis Draft
Alternative 2

Employ one more customer service rep. Assume

that this rep has the same service time
distribution as the original rep. A single waiting
line will be formed in front of these two
counters and the customer at the head of the
line will be served by the first available server.
Queue type λ µ L
Customers: =20 per hour
M/M/2 M/M/1 20 30 2
Two servers: per hour M/M/1 10 15 2
M/M/1 10 30 0.5
M/M/1 20 40 1
Service cost = (20)(2) M/M/2 20 30 0.75
M/M/2 20 60 0.34
Waiting cost=(0.30)(60)L

Total costs = (20)(2)+(0.30)(60)(0.75)=53.5

Analysis Draft
Alternative 3

Employ one more customer service rep. Queue type λ µ L

M/M/1 20 30 2
Assume that this rep has the same service time
M/M/1 10 15 2
distribution as the original rep. The M/M/1 10 30 0.5
management team doesn’t want other M/M/1 20 40 1
customers entering the store to get the idea M/M/2 20 30 0.75
that there were a lot of dissatisfied customers M/M/2 20 60 0.34
if they see a long line at “returns”. In order to
manage the optics =20 better,
per hour a waiting line is
Customers: Two M/M/1
formed in front of each customer service agent.
Two servers: per hour Each M/M/1: =10 per hour
per hour
Service cost = (20)(2)
Waiting cost=(0.30)(60)L
Total costs = (2)[20+(0.30)(60)(0.5)]=58
Analysis Draft
Alternative 4

A second person is added who helps the existing customer service rep process
the returns by doing the inspection of the returned items. The original rep will
now just deal with the customer and process the refund. The two employees
working as a team reduces the average service time to 1.5 minutes. The new
person added to help the customer service rep is only paid $16 per hour.

Queue type λ µ L
M/M/1 20 30 2
Customers: =20 per hour M/M/1 10 15 2
M/M/1 M/M/1 10 30 0.5
Two servers: per hour
M/M/1 20 40 1
M/M/2 20 30 0.75
M/M/2 20 60 0.34
Service cost = 20+16
Waiting cost=(0.30)(60)L Total costs = 20+16+(0.30)(60)(1)=54
Solution

Alternative Queue type λ µ L Total cost per hour

A1 M/M/1 20 30 2 20+(0.30)(60)(2)=56
A2 M/M/2 20 30 0.75 (20)(2)+(0.30)(60)(0.75)=53.5
A3 M/M/1 10 30 0.5 (2)[20+(0.30)(60)(0.5)]=58
M/M/1 10 30 0.5
A4 M/M/1 20 40 1 20+16+(0.30)(60)(1)=54

The second alternative is most cost effective.

Example 2

Consider a queueing system where there are two types of customers,

high priority(HP) and low priority (LP). HP customers arrive at a rate
of 40 per hour and LP customers arrive at the rate of 60 per hour. The
service time for both types of customers is on an average 3 minutes
per customer.
a. If each type of customer has dedicated servers, what is the
minimum number of servers who should be serving HP and LP
customers so that the queues are stable?
b. If the two types of customers are treated alike and they form a
single line and the person at the head of the line goes to whichever
server is free, what is the minimum number of servers required to have
a stable system?
c. Is the answer to part b, sum of the two numbers which are answers
to part a? Explain.
Analysis Draft

Two types of customers,

high priority(HP); arrive at a rate of 40 per hour 40
low priority (LP); arrive at the rate of 60 per hour 0

Service time: average 3 minutes per customer. 0

Solution

a. If each type of customer has dedicated servers, what is the minimum number of
servers who should be serving HP and LP customers so that the queues are stable?

40 0 0 HP: M/M/?
LP: M/M/?
To have stable queues,
s>2 s>3

Have to be an integer.
The minimum number of servers who should be serving HP
and LP customers are 3, and 4, respectively.
Solution

b. If the two types of customers are treated alike and

they form a single line and the person at the head of the
line goes to whichever server is free, what is the
minimum number of servers required to have a stable
system?
40 0 0
M/M/?
To have stable queues,

s>5

The minimum number of servers required to have a stable system are 6.

Solution

c. Is the answer to part b, sum of the two numbers which

are answers to part a? Explain.

Each type of
customer has a single line
dedicated servers M/M/6
HP: M/M/3
LP: M/M/4

No. In a stable queue, the system is free sometimes. If two

systems are combined, it is possible to reduce capacity waste.
Example 3

Diana’s daughter Tara runs a gas station and a car wash business. Tara
estimates that dirty cars arrive at a rate of 10 per hour. With a single
crew working the wash line, Tara figures that the mean cleaning time
per car is 5 minutes. Cleaning is done one car at a time and in a FCFS
fashion.
a. What is the average number of cars waiting in line at the car wash at any given time?
b. What is the average time that the car spends in the car wash?
c. What is the proportion of time that the crew is idle?
d. What is the probability that there are 3 cars at the car wash facility at any given time?
e. What is the probability that there are 5 cars waiting in line at the car wash?
f. Tara doesn’t want to have more than 4 cars in the car wash on an average. Is she
meeting her goal with the current setup?
Solution
=10 12 M/M/1

a. What is the average number of

cars waiting in line at the car 𝜆 𝜆
𝐿 𝑄= − ¿ 4.1667
wash at any given time? 𝜇− 𝜆 𝜇

b. What is the average time that 1 1

the car spends in the car wash? 𝑊 W  Wq    (    ) =0.5

c. What is the proportion of time 

𝑃 0 =1− =0.1667
that the crew is idle? 𝜇

d. What is the probability that (0.1667)=0.0965

there are 3 cars at the car wash
facility at any given time?
 =10 12

𝑃 0 =1− =0.1667
𝜇

e. What is the probability that

there are 5 cars waiting in (0.1667)=0.0558
line at the car wash?

f. Tara doesn’t want to have

more than 4 cars in the car /(-)=5 No.
wash on an average. Is she
meeting her goal with the
current setup?