CHAPTER 7

QUEUING MANAGEMENT
 CHAPTER 7. QUEUING MANAGEMENT
What is QUEUING MANAGEMENT?

The use of a mathematical method of analyzing the congestions and

delays of queing/waiting in line, on order to reduce or avoid them.

Queuing theory examines every component of waiting in line to be

served, including the arrival process, service process, number of
servers, number of system places and the number of "customers"
(which might be people, data, packets, cars, etc.).
 CHAPTER 7. QUEUING MANAGEMENT

Real-life applications of queuing theory include providing faster

customer service, improving traffic flow, shipping orders
efficiently from a warehouse and designing telecommunications
systems such as call centers or online platforms.
 CHAPTER 7. QUEUING MANAGEMENT

Queues are….. everywhere:

- Manufacturing.

- Services.

- Internet.
 CHAPTER 7. QUEUING MANAGEMENT

Queues are….. everwhere:

When they are well managed, queues give competitive

advantage:

- Customer satisfaction.

- Flexibility.

- Extra revenue.
 CHAPTER 7. QUEUING MANAGEMENT

Queues… Why?

We can “suffer” queues permanently:

- Perpetual queues. Because Demand > Capacity.

 CHAPTER 7. QUEUING MANAGEMENT

Queues…Why?

We can “suffer” queues during some periods:

- Temporary/Seasonal queues. Because…

- Demand > Capacity ➔ queues.

- Demand < Capacity ➔ over-capacity

 CHAPTER 7. QUEUING MANAGEMENT

Queues… Why?

We can “suffer” queues permanently:

- Random: Stochastic Queues.

(Stochastic process: is a collection of random variables, representing the evolution of

some system of random values over time).

Demand < Capacity (on average).

 CHAPTER 7. QUEUING MANAGEMENT
Queues… a situation…

The Pharmacy:

The pharmacy clerk (who is alone).

Clients can be of two types:

- Young client. It takes 1 minute to serve him/her.

- The old gentleman/lady who talks about all his/her pains all the time. It takes 5 minutes to
serve her.

There is a probability of 50% / 50% of a patient being a young one or the old lady. So, How many
clients can the clerk attend in 1 hour?
 CHAPTER 7. QUEUING MANAGEMENT
Queues… a situation…

The Pharmacy:

So, How many clients can the clerk attend in 1 hour?

Y = young client. Z = old lady.

Probability is 50% for each kind of client.

If we suppose that the sequence is:

Y, Z, Y, Z, Y, Z…….

Then, it takes 6 min. to serve client Y + client Z, so

2 clients/6 min.

2cl./6 min. x 60 min./1h = 20 clients/h.

 CHAPTER 7. QUEUING MANAGEMENT
Queues… a situation…

The Pharmacy:

But, what if the sequence is not that? This was a unique realization of
a situation occurring in a queing system during one hour. But….

What is the probability that this exactly same situation occurs in the
next hour? And the next one, and next…
 CHAPTER 7. QUEUING MANAGEMENT
Queues… a situation…

The Pharmacy:

So, we’ll model the queuing system in a way that we can study the
AVERAGE BEHAVIOUR of a stable period, and no every single different
situation or sequence.

(But… we’ll study only the steady-state situation. The opposite of that
is the transient state -time to reach steady-state- is more complex, we
won’t see it).

It’s important to remember this limitation.

 CHAPTER 7. QUEUING MANAGEMENT

Queues… a situation…
The Pharmacy:

If a system is in steady state, then the recently observed

behavior of the system will continue into the future. In stochastic
systems, the probabilities that various states will be repeated will
remain constant.
 CHAPTER 7. QUEUING MANAGEMENT
STRATEGIC DECISIONS:

- What is the cost of waiting?

- What is the cost of capacity?

Example:
A Hospital:
Appointment doctors,
- Cost of waiting: low.
- Very high utilization.
Emergency room,
- Cost of waiting; extremely high, people can die.
- Utilization: high or low, but sometimes like “firefighters”.
 CHAPTER 7. QUEUING MANAGEMENT
STRATEGIC DECISIONS:

Cost Trade-offs. We must find the balance between cost of waiting and cost
of service.

A shop: we can lose some sales for not having a higher number of
employees. But if we choose to have some more employees, even though
we’ll sell more, the higher personnel costs will lead to an economic loss for
the company.

Concepts: cost of service, cost of waiting, and….

ρ = Utilization (utilization rate). 0 <ρ < 1

 CHAPTER 7. QUEUING MANAGEMENT
How can we manage this situation?

We can try:

- Decreasing variability of service times, without changes in the

rest of system.

- Decreasing variability of interarrival times, without changes in

the rest of the system.
 CHAPTER 7. QUEUING MANAGEMENT

How can we manage this situation? We can try:

- Decreasing variability of service times, without changes in the

rest of system. The young client and the old lady. Or: Clinica
Dr. Barraquer. Barcelona.

- https://www.youtube.com/watch?v=wra4tI7kJ1E min 19,10.

- Decreasing variability of interarrival times, without changes in

the rest of the system. Galleria deggli Uffizi, Firenze.

- https://www.uffizi.it/en/tickets
 CHAPTER 7. QUEUING MANAGEMENT

THE QUEUING SYSTEM

The queuing system consists essencially of 3 basic components:

1) The source of population and the way the customers (…) arrive to
the system.

Arrival process:

- Random (distribution) or Deterministic.

- Single Customers or in batches.

- Degree of patience (impatient): Balking or reneging.

 CHAPTER 7. QUEUING MANAGEMENT
THE QUEUING SYSTEM

The queuing system consists essencially of 3 basic components:

1) The source of population and the way the customers (…)

arrive to the system.

Queue:

- Discipline: FCFS, LCFS, Priority.

- Lenght: Finite or Infinite.

- Single Queue of Multiple Queue.

 CHAPTER 7. QUEUING MANAGEMENT

THE QUEUING SYSTEM

The queuing system consists essencially of 3 basic components:

2) The servicing system.

Server process:

- Random (distribution) or deterministic.

- Single Customers or in batches.

- Single Server of Multiple Server.

 CHAPTER 7. QUEUING MANAGEMENT
THE QUEUING SYSTEM

The queuing system consists essencially of 3 basic components:

3) The condition of the customers (…) exiting the system (back to

source population or not?).

Departure:

- Leave system or stay in system and move to next queue.

 CHAPTER 7. QUEUING MANAGEMENT
RANDOMNESS. Different Random Processes…

Deterministic Process (no variability)

- Standard deviation = 0
- Coefficient of variation = 0. “Controlled variability”

Poisson/Markov Process. (“natural” Variability)

- Decreasing probability distribution.

- Average = Standard deviation.
- Coefficient of variation = 1. “Uncontrolled variability”

General Process (everything else)

0 < Coefficient of variation < Infinite

THE COEFFICIENT OF VARIATION: standard deviation/average

 CHAPTER 7. QUEUING MANAGEMENT
POISSON/MARKOV PROCESS.

How do we know that an arrival or service process is a Poisson process?

- We must collect data on arrivals (services) and compute interarrival times.

- Then, we draw a frequency histogram, compute average and standard deviation.
- A Poisson process has the following characteristics:
- Histogram is decreasing exponentially.
- Average is equal (or similar) to the standard deviation.
HISTOGRAM bus stops - passengers

20

15

10

0
1 2 3 4 5 6 7 8 9 10
 CHAPTER 7. QUEUING MANAGEMENT
We’ll see now some different models. This is the simplest one:

M/M/1 System.

This is the meaning of that:

ARRIVALS: M arrivals follow a Poisson process (a random-by-nature process).

SERVICE: M services follow a Poisson process (a random-by-nature process).

NUMER OF SERVERS: 1 (unique server with a possible infinite queue in front).

ARRIVAL QUEUE SERVICE

J KKLLKJ J
 CHAPTER 7. QUEUING MANAGEMENT
But we can also have…

- A single queue, pooled servers:

KKLLKJ

Advantages:
Never wait with one server idle.
Best use of servers (easier sharing)
 CHAPTER 7. QUEUING MANAGEMENT
…and…

- Specialized servers with special queues:

KKLLKJ

KKLLKJ

KKLLKJ

Advantages:
Can provide different level of service to different customers.
Allows specialization of servers
 CHAPTER 7. QUEUING MANAGEMENT
We can have Fast queues and Slow queues:

One fast server: Advantages:

- Shorter processing time.
KKLLKJ -Shorter overall total time in system.

Or Several slower servers: Advantages:

- Lower probability of waiting.

- Lower waiting time.

- Robustness.
KKLLKJ
 CHAPTER 7. QUEUING MANAGEMENT
And we can have some different Waiting Line Models:

Single channel:

KKLLKJ Single phase.

Single channel:
KKLLKJ Multiphase.

Multichannel:
Single phase.

KKLLKJ

Multiphase.
 CHAPTER 7. QUEUING MANAGEMENT

LAYOUT SERVICE SOURCE ARRIVAL QUEUE SERVICE PERMISSIBLE EXAMPLE

PHASE POPULATION PATTERN DISCIPLINE PATTERN QUEUE
LENGHT
MODEL 1 SINGLE SINGLE INFINITE POISSON FCFS EXPONENTIAL UNLIMITED ONE
Human
LINE
ticket
CHANNEL TOLL
seller
MODEL 2 SINGLE SINGLE INFINITE POISSON FCFS CONSTANT UNLIMITED DRAGON
CHANNEL KHAN
MODEL 3 MULTICHANNEL SINGLE INFINITE POISSON FCFS EXPONENTIAL UNLIMITED AUTO
PARTS
COUNTER
MODEL 4 SINGLE SINGLE FINITE POISSON FCFS EXPONENTIAL UNLIMITED MACHINE
CHANNEL REPAIR IN
FACTORY
 CHAPTER 7.
QUEUING MANAGEMENT
You will love this:

Human
ticket
seller
 CHAPTER 7. QUEUING MANAGEMENT
NOTATION FOR MODELS 1-3: (Pnemotechnics are suggested)

λ= Arrival rate.

μ= Service rate.

1/λ = Average time between arrivals.

1/μ = Average service time.

ρ= Ratio of total arrival rate to service rate for a single server; (λ/μ). For single-server queues, this is equivalent to utilization.

Lq = Average number waiting in line.

Ls = Average number in system.(including any being served)

Wq = Average time in waiting line.

Ws = Average time in system. (including any being served)

n = number of units in the system.

S = k = Number of identical service channels.

Pn = Probability of exactly n units in system.

Pw = Probability of waiting in line. (We are not respecting some Process Analysis terminology….)
CHAPTER 7. QUEUING MANAGEMENT

SEE QUEUING EQUATIONS.

CHAPTER 7. QUEUING MANAGEMENT
EXAMPLE: Customers in line.

MacMillans’ is considering opening a drive-through window for

customer service. Customers arrive at a rate of 15 customers per hour.
The waiter can service customers at a rate of 1 every 3 minutes.
Assuming Poisson arrivals and exponential service, find:
1. Utilization of the waiter.
2. Average number in line.
3. Average number in the system.
4. Average waiting time in line.
5. Average waiting time in the system, including service.
 CHAPTER 7. QUEUING MANAGEMENT

EXAMPLE: Customers in line.

MacMillans’ is considering opening a drive-through window for customer service. Customers arrive at te rate of 15 customers per hour. The waiter
can service customers at the rate of 1 every 3 minutes. Assuming Poisson arrivals and exponential service, find:
1. Utilization of the waiter.
2. Average number in line.
3. Average number in the system.
4. Average waiting time in line.
5. Average waiting time in the system, including service.

The first thing we must do is to identify the model. This is model 1

LAYOUT SERVICE SOURCE ARRIVAL QUEUE SERVICE PERMISSIBLE EXAMPLE

PHASE POPULATION PATTERN DISCIPLINE PATTERN QUEUE
LENGHT
MODEL 1 SINGLE SINGLE INFINITE POISSON FCFS EXPONENTIAL UNLIMITED ONE LINE
CHANNEL TOLL
MODEL 2 SINGLE SINGLE INFINITE POISSON FCFS CONSTANT UNLIMITED DRAGON
CHANNEL KHAN
MODEL 3 MULTICHANNEL SINGLE INFINITE POISSON FCFS EXPONENTIAL UNLIMITED AUTO
PARTS
COUNTER
MODEL 4 SINGLE SINGLE FINITE POISSON FCFS EXPONENTIAL UNLIMITED MACHINE
CHANNEL REPAIR IN
FACTORY
CHAPTER 7. QUEUING MANAGEMENT

PSYCHOLOGY OF WAITING LINES.

THE 8 PRINCIPLES OF WAITING TIME PERCEPTION:
1 – Unoccupied time feels longer than occupied time.
2 – Pre-process waits feel longer than in-process waits.
3 – Anxiety makes wait feel longer.
4 – Uncertain waits seem longer than known waits.
5 – Unexplained waits feel longer than explained waits.
6 – Unfair waits feel longer than fair waits.
7 – Waiting alone feels longer than waiting in group.
8 – The more valuable the service (or product), the longer I wait.

Source: The Psychology of Waiting Lines (D. Master, 1984)