11/26/2024

Airports and Airlines

SPC577

Lecture 07
Network Flows (Cont.) – Flight Scheduling
Dr. Mohannad Draz
Assistant Professor
Zewail City of Science, Technology and Innovation
[email protected]

Lecture o Networks

o Network Flow Models

Contents o Shortest Path (Route) Problem
o Minimum Cost Flow Problem
o Maximum Flow Problem
o Multi-Commodity Problem
o Set-Covering/Partitioning Problems

o Flight Scheduling
o Introduction
o Hub-and-Spoke
o Route Development and Flight-
Scheduling Process
o Load Factor and Frequency
o Case study

1
 11/26/2024

Network Flow Models

Multi-Commodity Problem
All the network models explained so far assume that a single commodity or type of
entity is sent through a network.

Sometimes a network can transport different types of commodities.

The multi-commodity problem seeks to minimize the total cost when different
types of goods are sent through the same network.

The multi-commodity problem is extensively used in the transportation industry.

In the airline industry, the multi-commodity model is adopted to formulate crew

pairing and fleet assignment models.

Network Flow Models

Multi-Commodity Problem
Example

We modify the example that was presented for the Minimum Cost Flow
problem discussed earlier to address the multi-commodity model formulation.

The only difference is that instead of having only one type of cargo,
we have two types (two commodities).

Node 1 supplies:
40 tons of cargo 1
35 tons of cargo 2

2
 11/26/2024

Network Flow Models

Multi-Commodity Problem

The transportation costs per ton are similar.

Determine how much from each cargo should be routed on each arc so
that the total transportation cost is minimized.

Decision variable:

The indices i and j represent the nodes (i,j = 1,.,7) and k represents the
type of commodity (k = 1,2).

The objective function is:

Network Flow Models

Multi-Commodity Problem

We need to write one constraint for each node.

For example, for node 1

Write a similar constraint

for the other six nodes.

All the flights to and from node 4 can carry a maximum of 50 tons

3
 11/26/2024

Network Flow Models

Multi-Commodity Problem

Solving this problem using software generates a total minimum cost of $1,150

Network Flow Models

Set-Covering/Partitioning Problems
Set-covering problems relate to cases where each member of one set should
be assigned/matched to member(s) of another set.

Examples include the assignment of crew members to flights,

aircraft to routes, and so on.

The objective of a set covering problem is to minimize the total cost of this
assignment.

4
 11/26/2024

Network Flow Models

Set-Covering/Partitioning Problems
Example

An airline wants to design its ‘hub’ system. Each hub will be used for
connecting flights to and from cities within 1,000 miles of the hub.

The airline wants to serve the following cities:

Atlanta, Boston, Chicago, Denver, Houston, Los Angeles, New Orleans, New
York, Pittsburgh, Salt Lake City, San Francisco, and Seattle.

The airline wants to determine the smallest number of hubs it will need in
order to cover all of these cities.

(Each city should be within 1,000 miles of at least one hub. Table 2.2 lists
the distances between the cities.

Network Flow Models

Set-Covering/Partitioning Problems

10

5
 11/26/2024

Network Flow Models

Set-Covering/Partitioning Problems
Identify which cities are covered by each hub.
Simply replace all the distances in the table by 1 if the distance is less than
1,000 miles (covered) and 0 otherwise.

11

Network Flow Models

Set-Covering/Partitioning Problems

binary decision variable

minimize the number of hubs, therefore the objective function is:

Each city must be covered by at least one hub. Atlanta (Index 1), for example,
is covered by cities 1, 3, 5, 7, 8, and 9

Therefore, the constraint for Atlanta is:

write similar constraints

for all the other 10 cities.

12

6
 11/26/2024

Network Flow Models

Set-Covering/Partitioning Problems

Solving this binary integer program using software generates three

hubs as follows:
Atlanta covers Chicago, Houston, New Orleans, New York, and Pittsburgh;
Pittsburgh covers Atlanta, Chicago, Boston, and New York;
Salt Lake City covers Denver, Los Angeles, San Francisco, and Seattle.

Some cities are covered by more than one hub

To cover each city by exactly one hub, all the inequalities in the above model become
equal signs. This special case where each member of one set is covered exactly once is
called set-partitioning.

If we run the above program with this restriction, that is,

The hubs are:

Boston covers New York and Pittsburgh

New Orleans covers Atlanta, Chicago, and Houston
Salt Lake City covers Denver, Los Angeles, San Francisco, and Seattle

13

Network Flow Models

Set-Covering/Partitioning Problems

The general model for set-covering is as follows:

14

7
 11/26/2024

Network Flow Models

Set-Covering/Partitioning Problems

The set of constraints imposes that each member of set 1 is

covered by at least one member of set 2.

The set-partitioning formulation problem is similar, except

that is now rewritten with a strictly equal to sign

15

Flight Scheduling
Introduction
• Flight scheduling is the starting point for all other airline planning and
operations .
• The flight schedule is a timetable consisting of what cities to fly to and at what
times.
• An airline’s decision to offer certain flights will mainly depend on market
demand forecasts, available aircraft operating characteristics, available
manpower, regulations, and the behavior of competing airlines.

• Table 3.1 shows a small portion of the daily flight schedule for Delta Air Lines.

• The level of detail in constructing the flight schedule varies among the airlines,
but it will be a complete schedule for a full cycle
• A cycle is normally one day for domestic and one week for international
services.

16

8
 11/26/2024

Flight Scheduling

17

Flight Scheduling
Hub-and-Spoke

• Most airlines adopt some variation of a hub-and-spoke system.

• Major carriers operate up to five hubs, while smaller ones typically have one
hub located at the center of the region they serve.
• Each hub has a set of cities that it serves, normally referred to as spokes.
• Figure 3.1 shows an airline network with Chicago O’Hare and Washington
Dulles as hubs.

18

9
 11/26/2024

Flight Scheduling
• Air carriers normally assign large capacity non-stop flights between their
hubs.

• Smaller airplanes are assigned to hub-and-spoke flights.

• Major advantages for the airlines adopting hub-and-spoke operations

include higher revenues, higher efficiency, and lower number of aircraft
needed as compared with point-to-point operations.

• Disadvantages of these operations include discomfort to the passengers, as

they may require multiple connecting flights at different hubs, congestions
and delays at hub airports, and higher personnel and operational costs for the
airlines.

19

Flight Scheduling
Route Development and Flight-Scheduling Process

There are two types of route development activities:

1- strategic
2- tactical.

• Strategic development focuses on future schedules which may range

from a few months to ten years depending on the air carriers’ policies.
• Strategic developments respond to major changes in both business and
operational environments.
• Tactical strategies, on the other hand, focus on short-term changes to
the schedule and routes, sometimes on a daily basis.
• This is done by constantly monitoring markets, competitors, and
operations.
• The tactical strategy includes adding, dropping flights, and making
changes to city-pair markets and their frequencies.

20

10
 11/26/2024

Flight Scheduling
Load Factor and Frequency

• Average load-factor plays an important role in determining the

frequency of flights between city pairs.
• Load factor is the average percentage of aircraft seats which are filled
with passengers.
• The parameters affecting load factors include flight times, frequency,
type of service and, of course, fare levels. It should be noted that a
higher load-factor does not necessarily translate into higher revenues
for the airlines.

21

Flight Scheduling
• The following table shows the fares, expected demands and load factors for a
150-seat Airbus A-320. According to this table, an 85% load factor generates
higher revenues of more than 100% for the airline!

22

11
 11/26/2024

Flight Scheduling
• The load factor is utilized to determine the frequency between city pairs.
• Let the forecasted daily number of passengers between two cities be PAX
• The airline’s policy on average load-factor be LF.
• The average aircraft capacity is CAP.
• The frequency (FREQ) of flights between these two cities is determined
by:

As implied by the above equation, the load factor and frequency have an
inverse relationship.

It is up to the marketing and scheduling departments to actually assign

these frequencies between city pairs to different times of the day/week.

23

Flight Scheduling
Case Study
In this section a fictitious airline is presented. We will use this airline in the
following chapters to introduce the various phases of planning within the airlines.
Ultimate Air is a new airline that provides service to the most important domestic
business destinations within the United States from its hub at JFK in New York.
The cities serviced from JFK are Boston (BOS), Los Angeles (LAX),
San Francisco (SFO), Miami (MIA), Atlanta (ATL), Washington D.C. (IAD), and
Chicago (ORD).

24

12
 11/26/2024

Flight Scheduling

25

Flight Scheduling
Table 3.4 presents the demand distribution for each flight as well as
distances between cities. It is assumed that demand for each flight is
normally distributed with the given means and standard deviations.

26

13
 11/26/2024

THANK
YOU
27

14