Embed Size (px)
Citation preview
Runway/Taxiway System Capacity Analysis at the LisbonAirport (LPPT-Lisboa)
Goncalo Soares Roque
Thesis to obtain the Master of Science Degree in
Aerospace Engineering
Supervisors: Prof. Pedro da Graca Tavares Alvares SerraoM.Sc. Max Georg Schultze Schwienhorst
Examination Committee
Chairperson: Prof. Jose Fernando Alves da SilvaSupervisor: Prof. Pedro da Graca Tavares Alvares Serrao
Member of the Committee: Sr. Manuel Antonio de Magalhaes Alberto de Araujo
June 2017
Acknowledgments
First and foremost I would like to thank both my supervisors, Prof. Pedro Serrao and M.Sc. Max
Schwienhorst, for the incredible opportunity they gave me to perform this work in Aachen, Germany.
Prof. Pedro introduced me to NAV Portugal, where the topic and focus of this work was discussed and
determined, and always showed extreme interest in the work being developed, providing invaluable help
and guidance throughout this work. Max introduced me to the world of airport simulation, and was
extremely kind in providing access to a work station and the simulation tool Simmod PLUS!, as well as
providing me with the opportunity to travel to Manchester, United Kingdom, for the Airside Simulation and
Performance Assessment Group (ASPAG) and the European Simmod User Group (ESUG) meetings. I
would also like to thank Max for the help provided with all the logistical requirements in order to be able
to perform this work in Aachen.
I would like to thank Jesus Conde, Manuel Araujo and Vanda Cruz, from NAV Portugal, for always
being helpful and available, and providing crucial insight and information for the realization of this work,
as well as for receiving me in NAV Portugal, and clarifying the many doubts relatively to the operational
reality of the air traffic control in the Lisbon Airport.
Finally, I would like to thank my family for giving me the opportunity to study in a university, and being
there for me every time I needed. I would also like to thank the friends with whom I shared this journey
through university in Lisbon, as well as the ones with whom I shared the ’Hiwi Raum’, place where I
spent countless hours developing this work.
Abstract
This work makes an assessment of the capacity of the Lisbon Airport through computer simulation, with
Simmod PLUS!. The current state of the airport, with special focus on the runway and taxiway systems,
is fully modeled. Results are obtained from two traffic samples with the duration of one day. One is from
the current year, 2017, representing the current demand of the airport and used to calibrate the model,
and one with an increase in the number of flights relatively to the current traffic sample, representing a
medium term future demand. Given the results obtained in simulation by the current state of the airport
when tested with the medium term future demand, capacity enhancements to the airport, specially
in the form of layout modifications, are proposed and further modeled. With the modifications made,
the simulated airport is able to comfortably accommodate the medium term future demand and show
considerable performance improvements when compared to the current state of the airport.
Keywords: Airport; Runway; Taxiway; Capacity; Simulation; Simmod PLUS!
iii
Resumo
Este trabalho faz uma avaliacao da capacidade do Aeroporto de Lisboa atraves de simulacao por com-
putador, com Simmod PLUS!. O estado atual do aeroporto, com especial incidencia nos sistemas de
pista e taxiway, e totalmente modelado. Os resultados sao obtidos a partir de duas amostras de trafego
com duracao de um dia. Uma provem do ano corrente, 2017, representando a procura atual do aero-
porto e utilizada para calibrar o modelo, e outra com um aumento no numero de voos relativamente a
amostra de trafego atual, representando uma procura futura a medio prazo. Dado os resultados obtidos
em simulacao pelo estado atual do aeroporto, quando testado com a procura futura a medio prazo, mel-
horias de capacidade do aeroporto, especialmente sob a forma de alteracoes ao layout do aerodromo,
sao propostas e modeladas. Com as modificacoes feitas, o aeroporto simulado e capaz de acomodar
confortavelmente a procura a medio prazo e mostrar melhorias consideraveis de desempenho quando
comparado com o estado atual do aeroporto.
Palavras-chave: Aeroporto; Pista; Taxiway; Capacidade; Simulacao; Simmod PLUS!
v
Contents
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 LPPT Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Airspace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Airfield . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Background 7
2.1 Airport Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.1 Airspace Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.2 Runway Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1.3 Taxiway Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.4 Gate Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1.5 Terminal Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2 Measuring Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3 Methodology 13
3.1 Simmod PLUS! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Runway in Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Meteorological Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.4 Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5 Traffic Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4 Implementation 19
4.1 LPPT Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.1.1 Airspace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.1.2 Runway . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.1.2.A Runway Distances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.1.2.B Departure Queues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.1.2.C Line-up and Takeoff . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
vii
4.1.2.D Landing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.1.2.E Runway Crossing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.1.2.F Runway Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4.1.3 Taxiways . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.1.4 Gates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2 Flights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
5 Results 47
5.1 Number of Iterations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2 Current vs. Future Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.3 Runway 03 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.3.1 Current Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.3.2 Future Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.4 Runway 21 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.4.1 Current Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.4.2 Future Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.5 New Airport Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.5.1 Runway 03 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.5.2 Runway 21 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
6 Conclusion 75
A LPPT Charts 81
A.1 LPPT Aerodrome Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
A.2 LPPT Ground Movement Chart RWY03 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
A.3 LPPT Ground Movement Chart RWY21 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
B Flights Timetable 87
B.1 Current Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
B.2 Future Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
viii
List of Figures
1.1 LPPT evolution throughout history [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 LPPT visual approach ICAO chart [7]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.1 Beta distribution probability density function examples. . . . . . . . . . . . . . . . . . . . . 17
4.1 Simmod PLUS! world map. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2 Model airfield. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.3 Model airspace. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.4 Separation increment probability density function. . . . . . . . . . . . . . . . . . . . . . . . 26
4.5 Model runway. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.6 Aircraft queuing for departure example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.7 FRTT probability density function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.8 Takeoff roll base probability density function. . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.9 Runway crossing example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.10 Runway crossing extra time probability density function. . . . . . . . . . . . . . . . . . . . 35
4.11 Model taxiway speeds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.12 Blocked links due to taxiing aircraft example. . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.13 Blocked links due to dynamic path example. . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.14 Model gates, pushback and dwelling example. . . . . . . . . . . . . . . . . . . . . . . . . 43
4.15 Dwell time probability density function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.16 Arrival lateness probability density function. . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.17 Departure lateness probability density function. . . . . . . . . . . . . . . . . . . . . . . . . 45
5.1 Maximum throughput convergence study. . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.2 LPPT current demand. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.3 LPPT medium term future demand. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.4 Arrivals hourly percentage of aircraft by International Civil Aviation Organization (ICAO)
wake vortex category. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.5 Departures hourly percentage of aircraft by performance after takeoff group. . . . . . . . . 52
5.6 Runway 03 maximum throughput for different wind conditions. . . . . . . . . . . . . . . . . 53
5.7 Runway 03 throughput vs current demand. . . . . . . . . . . . . . . . . . . . . . . . . . . 54
ix
5.8 Runway 03 average arrival and departure delay for the current demand. . . . . . . . . . . 54
5.9 Runway 03 average arrival delay breakdown. . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.10 Runway 03 average departure delay breakdown. . . . . . . . . . . . . . . . . . . . . . . . 55
5.11 Runway 03 throughput for no wind and 30 knots headwind conditions. . . . . . . . . . . . 56
5.12 Runway 03 average arrival and departure delay for no wind and 30 knots headwind con-
ditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.13 Runway 03 throughput vs future demand. . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.14 Runway 03 average arrival and departure delay for the future demand. . . . . . . . . . . . 58
5.15 Runway 21 maximum throughput for different wind conditions. . . . . . . . . . . . . . . . . 59
5.16 Runway 21 throughput vs current demand. . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.17 Runway 21 average arrival and departure delay for the current demand. . . . . . . . . . . 60
5.18 Runway 21 average arrival delay breakdown. . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.19 Runway 21 average departure delay breakdown. . . . . . . . . . . . . . . . . . . . . . . . 61
5.20 Runway 21 average departure delay breakdown. . . . . . . . . . . . . . . . . . . . . . . . 62
5.21 Runway 21 throughput for no wind, 15 knots and 30 knots headwind conditions. . . . . . . 63
5.22 Runway 21 average arrivals and departures delay for different wind conditions. . . . . . . 63
5.23 Runway 21 throughput vs future demand. . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.24 Runway 21 average arrival and departure delay for the future demand. . . . . . . . . . . . 64
5.25 Airfield layout modifications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.26 New runway 03 maximum throughput for different wind conditions. . . . . . . . . . . . . . 67
5.27 New runway 03 throughput vs future demand. . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.28 New runway 03 average arrival and departure delay for the future demand. . . . . . . . . 68
5.29 New runway 03 throughput for no wind, 15 knots and 30 knots headwind conditions. . . . 69
5.30 New runway 03 average total delay for no wind, 15 knots and 30 knots headwind conditions. 69
5.31 New runway 21 throughput vs future demand. . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.32 New runway 21 average arrival and departure delay for the future demand. . . . . . . . . 72
5.33 New runway 21 throughput for no wind, 15 knots and 30 knots headwind conditions. . . . 72
5.34 New runway 21 average total delay for no wind, 15 knots and 30 knots headwind conditions. 73
x
List of Tables
4.1 Aircraft separation minimum. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
4.2 True Air Speed (TAS) of aircraft arriving and departing from the airport. . . . . . . . . . . 24
4.3 Runways declared distances [7]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.4 Departure queues and their sizes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.5 Departure queues usage percentage for airspace groups. . . . . . . . . . . . . . . . . . . 29
4.6 Takeoff rolls duration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.7 Runway 03 exits and usage percentage for airspace groups. . . . . . . . . . . . . . . . . 32
4.8 Runway 21 exits and usage percentage for airspace groups. . . . . . . . . . . . . . . . . 32
4.9 Runway 03 Runway Occupancy Time for Arrivals (ROTA) for each runway exit and airspace
group combination. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
4.10 Runway 21 ROTA for each runway exit and airspace group combination. . . . . . . . . . . 33
4.11 Aircraft groups according to performance after takeoff. . . . . . . . . . . . . . . . . . . . . 36
4.12 Lisbon Airport aprons and related use description. . . . . . . . . . . . . . . . . . . . . . . 42
4.13 Aprons attributed to each type of flight. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.1 New runway 03 exits and usage percentage for airspace groups. . . . . . . . . . . . . . . 66
5.2 New runway 21 exits and usage percentage for airspace groups. . . . . . . . . . . . . . . 70
xi
xii
Acronyms
ADS-B Automatic Dependent Surveillance-Broadcast
ANSP Air Navigation Service Provider
ASDA Accelerate-Stop Distance Available
ATA Actual Time of Arrival
ATC Air Traffic Control
ATD Actual Time of Departure
ATM Air Traffic Management
eAIP electronic Aeronautical Information Publication
ETA Estimated Time of Arrival
ETD Estimated Time of Departure
FIFO First In First Out
FRLC Flight Crew Reaction to Line-up Clearance
FRTT Flight Crew Reaction to Takeoff Clearance
GS Ground Speed
ICAO International Civil Aviation Organization
ILS Instrument Landing System
LDA Landing Distance Available
LOS Level of Service
LUPT Line-up Time
PDF Probability Density Function
ROT Runway Occupancy Time
ROTA Runway Occupancy Time for Arrivals
xiii
ROTD Runway Occupancy Time for Departures
TAS True Air Speed
TMA Terminal Control Area
TODA Takeoff Distance Available
TORA Takeoff Run Available
xiv
1Introduction
Contents
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 LPPT Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1
2
1.1 Motivation
The Lisbon Airport (with ICAO code LPPT), since May 2016 formally known as Lisbon Humberto Del-
gado Airport [1], was open to the public in October 1942, with the main purpose of serving as a transfer
airport for transatlantic flights, at the time done by seaplanes that landed on Cabo Ruivo Seaplane Base.
Passengers would then be transfered by car to the Lisbon Airport from which they would board other
flights in order to reach European destinations [2]. Throughout history, both the increase in passenger
demand and the innovation of commercial airplanes resulted in changes to the airport layout, going from
the initially built four runways and single terminal, in 1942, to the current two runways and two terminals
configuration. Different configurations of the airport in its first 50 years of existence can be seen in Figure
1.1.
Figure 1.1: LPPT evolution throughout history [3].
The increasing passenger demand of the Lisbon Airport and subsequent growth of the same has
caused not only changes to the airport layout, but also a debate for the construction of a new airport
in the vicinity of Lisbon, since the city itself has grown so much that it now severely limits the ability to
further expand the current airport. This debate started in 1969, and several locations were suggested
and considered throughout the years such as Rio Frio, Ota and Alcochete [4]. Most recently, the Montijo
Air Base has been introduced as the most favorable location for a new Lisbon Airport, although studies
are still being conducted to further verify the viability of this solution, with the necessary modifications
of the air base in order to transform it to a secondary airport to Lisbon not being planned to start before
2019 [5] [6]. The Lisbon Humberto Delgado Airport is now in terms of yearly passenger numbers one
of the 30 biggest airports in Europe, hitting new records in every single month of 2016, with a total of
22.4 million passengers, a 11.7% growth relatively to the previous year [5]. Facing the current growth of
the Lisbon Humberto Delgado Airport, it is imperative to find practical solutions to increase the airport
capacity in the short term in order to accommodate the predicted traffic demand for the upcoming years.
3
1.2 LPPT Overview
LPPT is the biggest airport in Portugal and is located seven kilometers north of the Lisbon city center,
with an elevation of 114 meters [7]. It serves both national and international commercial flights as well as
charter, private and cargo flights [8]. It also contains the Transit Aerodrome No. 1 from the Portuguese
Air Force, with the 504 Squadron Linces permanently placed in it [9] [10].
1.2.1 Airspace
The airspace surrounding the airport is severely limited by military operations. In the vicinity of Lisbon
there are two air bases, Sintra and Montijo (the later being the one to operate in the future as a secondary
airport to Lisbon), and a field firing range in Alcochete. This military areas can be seen delimited in red
in Figure 1.2.
Figure 1.2: LPPT visual approach ICAO chart [7].
Besides military operations, Cascais Municipal Aerodrome (ICAO code LPCS) also imposes some
limitations in the use of the airspace around the LPPT, although most of its flights are of private or
business nature, which can’t be correctly predicted . Constraints made by such a limited airspace is
itself a topic of study and a limitation to the capacity of the LPPT [11] [12]. However, for the purpose
of this work, it is of very small importance as the airspace area within scope of this study is but a few
kilometers away from the airport, where arriving airplanes are already lined with the runway and given
the correct separation, and departing airplanes are still on their initial climb.
4
1.2.2 Airfield
The airfield of the LPPT has two runways, 03/21 and 17/35 (the name of the runways is given according
to their orientation relatively to the magnetic north, rounded to the tens, and taking into account both
orientations in which a single runway can be used). Runway 03/21 is the runway used in a vast majority
of the time and is the only runway configuration for LPPT. Runway 17/35 may be requested by pilots
when 03/21 is unsuitable for a particular operation (normally due to weather conditions) but it requires
Air Traffic Control (ATC) clearance as coordination with the military operations is needed. Runway 03
has Instrument Landing System (ILS) CAT I while runway 21 has ILS CAT II/III. Runway 17/35 does not
have any type of ILS [7]. The airfield is served by two terminals, terminal 1 being the one used for the
majority of operations and terminal 2 being used for departures by certain low-cost commercial airlines.
Besides the commercial terminals, that do most of the operations at the airport, there is a cargo terminal,
a small terminal for private flights and the military base mentioned in Section 1.2.
1.3 Objectives
The goal of this study is to evaluate the current airport capacity of the LPPT and identify possible en-
hancements so that the predicted higher traffic demands in the upcoming years can be accommodated.
Specifically, focus is given to the runway and taxiway system performance of the airport, having gate
capacity as well in consideration. To perform the study, a large scale airport simulation tool, Simmod
PLUS!, is used.
1.4 Thesis Outline
Chapter 2 of this work gives a background to accessing and determining the capacity of an airport and
introduces the different methods that can be used to do so. In Chapter 3, the simulation tool Simmod
PLUS! is briefly described, an analysis is made of the factors that are taken into account and not taken
into account in this study and the data acquiring process is discussed. In Chapter 4, a full detailed
description of every step made in the creation of the simulation model is given. Chapter 5 provides the
results obtained by the airport under the different simulated conditions and proposes a set of possible
capacity enhancements, necessary due to the results obtained. A further analysis of the airport with
the proposed enhancements is also made in this chapter. Finally, Chapter 6 gives the main conclusions
obtained in this work and proposes ideas for future work in the topic.
5
6
2Background
Contents
2.1 Airport Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Measuring Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
7
8
2.1 Airport Capacity
The capacity of an airport, although being a concept widely used, has no generally accepted definition.
ICAO defines it as the number of movements per unit of time than an airport is able to achieve under
different meteorological conditions [13]. However, some other definitions take into consideration the
achievable Level of Service (LOS) when defining capacity, normally in the form of accepted average
delay, while others take into consideration the number of achievable hourly movements over a period
longer than one hour.
An airport has within it a number of interconnected systems, as the airspace, runways, taxiways,
gates and terminals. The capacity of an airport is the capacity of its least effective system, therefore all
systems of an airport should be taken into account when assessing its capacity.
2.1.1 Airspace Capacity
The airspace within the vicinity of an airport is normally referred to as the Terminal Control Area (TMA).
Its capacity is defined by the physical characteristics of the airspace, such as terrain elevations, popu-
lated areas that impose noise constraints, other airports in the area that also require use of the same
airspace and military operations that impose restrictions on the available airspace. The physical charac-
teristics of the airspace determine the routes that might be taken when approaching or departing from an
airport (taking into account its runways orientation). One factor that influences the airspace capacity on
top of its physical characteristics is air traffic controller workload. Since the airspace in the vicinity of an
airport requires Air Traffic Management (ATM), which is done by air traffic controllers, it must be ensured
that the controllers workload does not surpass a certain accepted level, which imposes limitations on
the airspace capacity. This factor is specially relevant in airports with high traffic density.
2.1.2 Runway Capacity
The runway system is of the utmost importance in an airport and its capacity is often the focus of capacity
enhancements. Runway capacity directly represents the number of airplanes that are able to arrive and
depart from an airport over a certain period of time. It is influenced by the number of runways and its
geometric layout, the Runway Occupancy Time (ROT), the separation requirements between aircraft,
the traffic mix using the runway, the different type of operations being performed ( No. of arrivals vs No.
of departures) and the weather conditions. Besides these factors there are several others that indirectly
influence the runway system performance (e.g. the location of runway exits directly influences the ROT
which in turn directly influences runway capacity). It is worth mentioning that optimizing the runway
capacity is a task of great complexity influenced by all the mentioned factors where every second that
can be gained in an operation is important in the optimization process.
9
2.1.3 Taxiway Capacity
Taxiways determine the ability and the necessary time for airplanes to move between the runways and
the gates. Being an adjacent system to the runway, it is critical that taxiways are able to effectively feed
departures and alleviate arrivals from the runway, diminishing the runway occupancy time as much as
possible. Taxiway capacity is not often referred to as it is rarely a limiting factor for an airport capacity.
However, bottlenecks on the taxiway system can occur and its capacity should not be disregarded.
One particular example where taxiways heavily influence an airport capacity is runway crossings. If
an airplane in order to reach its gate or to reach the runway needs to cross a runway, it will need to
occupy that runway for a certain period of time which degrades the performance of the whole system.
Some smaller airports have also the particularity of having a single taxiway connecting the stands with
the runway. When faced with larger demands, this taxiway is often a bottleneck and limits the airport
capacity.
2.1.4 Gate Capacity
Gates are the locations in an airport in which aircraft can park for a certain amount of time, commonly
referred to as stands, and gate capacity is of extreme relevance. If an aircraft that arrives to an airport
has no available gate to go to, it needs to wait in a taxiway, blocking it from any further movements which
can provoke huge delays for several other aircraft. While the number of aircraft that can be parked in
an airport at any single time can be determined simply by counting the number of stands, gate capacity
needs to take into account the availability of those stands, as different gates are often attributed to
different airlines and the time that an airplane occupies certain gate is not constant. Other factor to take
into account is the service that aircraft require when parked in order to prepare for their following flight.
The availability of service vehicles and staff to perform this service is a factor that often limits the gate
capacity of an airport.
2.1.5 Terminal Capacity
In order to access the aircraft, passengers and staff need to go through a certain infrastructure area
designated terminal. When studying the ability of this infrastructure to deliver people to their target
aircraft, the passenger terminals are the ones most often taken into account, as they are by far the
ones through which more people need to go through in an airport. The terminal itself can be subdivided
in different systems, as check-in area and security control, and when analyzing the terminal capacity
all this subsystems need to be taken into account. The optimization of a terminal has gain relevance
throughout the years, as it is the system that influences passenger experience the most, a factor used
to directly analyze an airport quality.
10
2.2 Measuring Capacity
Measuring the capacity of an airport is a task driven by different factors. If the current demand is
causing unacceptable delays, a capacity analysis to identify possible enhancements is needed in case
such demand is predicted to be maintained for a longer period of time. However, this is done reactively
in response to the current state of an airport. A better and more proactive approach is to perform a
forecast of the future demand of the airport, verify if a significant change relatively to the current demand
is predicted, and perform a capacity assessment accordingly. Besides demand related reasons, the
capacity of an airport may also be analyzed due to regulation by different entities on airport performance
as well as by occasional changes on airport operations (e.g. closing a taxiway or runway, a shortage
of staff, applying different procedures due to environmental reasons or a specific event occurring in the
area) [14].
Different approaches can be taken in order to calculate the capacity of an airport, each having a
different level of complexity. Historical approaches look at data collected in the past and perform an
analysis to determine the evolving performance of the airport. If a degradation in performance is seen
and sufficient data has been collected, conclusions can be drawn regarding which factors contributed
to the performance degradation, providing a basis to where to focus possible future capacity enhance-
ments. Analytical models and look up tables provide another method of measuring the capacity of an
airport based on previously calculated capacities taking into account changes in the different factors
influencing capacity mentioned in Section 2.1. However, this models and tables normally look at the run-
way capacity alone, disregarding all the other systems an airport has within itself. Simulation models,
on the other hand, can provide a full gate-to-gate analysis and take into account all the different factors
that influence the capacity of an airport, providing the analyst a full overview of the entire airport system.
Simulation tools are usually a paid software owned by private companies. Some examples are:
• AirTOp, from Airtopsoft [15]
• CAST, from Airport Research Center GmbH [16]
• PIATA, from EUROCONTROL [17]
• RAMS Plus, from ISA Software [18]
• Simmod PLUS!, from ATAC [19]
11
12
3Methodology
Contents
3.1 Simmod PLUS! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Runway in Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Meteorological Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.4 Random Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5 Traffic Samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
13
14
In order to perform the analysis of the Lisbon Airport, Simmod PLUS! is used as the airport simulation
tool to create a model of the airport. In a first stage, the current state and layout of the airport is modeled,
which provides a good calibration factor for the model, by comparing it to the current results achieved in
real time by the airport. The results are obtained by testing two different traffic samples in the model.
One from an average week day of the current year, 2017, used not only to obtain results but also as
calibration factor for the model, and other representing a medium term future demand.
3.1 Simmod PLUS!
Simmod PLUS! is a fast time simulation tool that describes the evolution of an airport over time by a
mathematical model. The state of the model changes at discrete points in time, determined by the
occurrence of an event that changes the variables of the model. Therefore the simulation schedules
events and analyzes the state of the model only at the points in time in which an event occurs. After
processing an event, the simulation advances to the next point in time in which the next event occurs,
ignoring the time interval between two consecutive events [20].
In order to model an airport, Simmod PLUS! uses nodes and links. Nodes are the points in space in
which an aircraft position is evaluated in the system, and links define the paths that aircraft are allowed
to travel between nodes. A difference is made between ground and airspace nodes as well as links
due to the different characteristics that each have. The point in space at which an aircraft changes from
airspace to ground is called an interface node.
As the purpose of the simulation is to produce realistic results, not any ideal or specific case scenario,
Simmod PLUS! takes a certain amount of variables as user defined probability distributions, such as:
• Lateness of flights
• ROTA
• Runway Occupancy Time for Departures (ROTD)
• Separation requirements
• Start-up times
• Dwell time after push-back
In order to obtain realistic results, considering the amount of detail given to the model and to its
random variables, several iterations of the same data set should be done in order to obtain statistically
significant tendencies. Results should then be obtained by averaging the individual results from each of
the iterations made.
15
3.2 Runway in Use
As mentioned is Section 1.2.2, runway 03/21 is the only runway configuration used in the LPPT. Since
runway 17/35 has no ILS, it is very rarely used and only on pilot request, due to certain weather condi-
tions (normally strong crosswinds in runway 03/21), usually by experienced pilots that fly frequently to
the Lisbon Airport. Considering this factors, runway 17/35 is not modeled as a runway, and is instead
treated as a normal taxiway. Since the operational procedures and aircraft movements are extremely
different depending on which runway is in use, 03 or 21, two different models are created in order to
simulate the two different runways in use.
3.3 Meteorological Conditions
Unlike other airports in the world, the Lisbon Airport is not constantly affected by bad meteorological
conditions, which can drastically influence the capacity of an airport and provoke a huge amount of
delays. The most frequent adverse weather condition affecting the Lisbon Airport is fog, which produces
low visibility and increases separation requirements between aircraft throughout the airport. It happens
in about 30 days per year, with a total of 140 hours. For the sake of simplicity, this work does not take
into consideration the impact of weather conditions on the airport capacity, although the simulation tool
used allows to do so, mainly in the form of runway visual range and airport minimum ceiling. Instead,
the capacity of the airport is analyzed for conditions of clear sky and total visibility. Wind conditions, on
the other hand, constantly affect the Lisbon Airport. In consequence, wind is taken into consideration by
analyzing the effects on the capacity of the airport by winds between 0◦ and 90◦, with a 30◦ increment
and for speeds between 0 and 30 knots, with 15 knots increments. Note that the orientations given
are relative to the aircraft movement orientation and not to the north. As a result, 0◦ wind corresponds
to wind opposite to the aircraft movement, full headwind, and 90◦ wind corresponds to full cross wind.
Note that tailwind does not need to be considered as the runway in use is selected and if necessary
changed so that aircraft have headwind and not tailwind. The side from which the wind comes (e.g. 30◦
from the left or 30◦ from the right) is not taken into consideration not only for the sake of simplicity, but
also because in simulation there is no difference between the two. The simulation variable impacted
by the wind strength and direction is aircraft Ground Speed (GS). In simulation, the wind direction
is not introduced as headwind but relatively to the magnetic north. The impact of the wind on GS as
internally calculated by the simulator is given in Equations 3.1-3.3, which presents a good approximation
for situations where the wind speed is significantly lower than the aircraft TAS [20].
Headwind = cos(Wind Heading −Aircraft Heading) ×Wind Speed (3.1)
Crab Factor = 1 − Wind Speed2 −Headwind2
TAS2(3.2)
Ground Speed = (TAS × Crab Factor) −Headwind (3.3)
16
3.4 Random Variables
As mentioned in Section 3.1, some of the variables of the simulation have some randomness introduced
into them. As such, an appropriate probability distribution must be defined to describe such variables.
The most accurate approach is to do a case by case analysis of such variables and extrapolate from
a big enough sample of real data a probability distribution for each variable. However, for the sake of
simplicity, this work takes a broader approach. As all variables that need to be defined by a probability
distribution have limited intervals (e.g. a flight has no probability of arriving 10 hours late or 10 hours
early), a broad limited distribution, the Beta distribution, is chosen as the distribution for all random
variables of the simulation.
The Beta distribution is defined in the interval [0 1], and takes two positive shaping parameters, α
and β. The Probability Density Function (PDF) of the Beta distribution can be seen in Equation 3.4 and
some examples can be observed in Figure 3.1 [21].
f(x, α, β) = k · xα−1 · (1 − x)β−1 =xα−1 · (1 − x)β−1∫ 1
0uα−1 · (1 − u)β−1 du
(3.4)
Figure 3.1: Beta distribution probability density function examples.
Although the Beta distribution is defined only in the interval [0 1], the different random variables on
the simulation have different ranges, therefore a scaling of the defined interval is done in a case by case
basis in order to define a probability distribution in the correct interval for each of the random variables.
As in computer simulation it is not possible to define a continuous function, approximately 100 equally
spaced points are given to every variable defined by a probability distribution.
17
3.5 Traffic Samples
The traffic sample for the current year, 2017, is from an average weekday, Wednesday the 12th of April,
and can be seen in Appendix B.1. It is obtained from two sources. NAV Portugal, the Portuguese Air
Navigation Service Provider (ANSP), kindly provided data relative to the day with all flights call sign,
origin, destination, airline, Estimated Time of Arrival (ETA), Actual Time of Arrival (ATA), Estimated
Time of Departure (ETD) and Actual Time of Departure (ATD). In this sample, both commercial and
cargo flights are included, and three private flights are added as an estimate of the private movements
in an average day. However, with this information alone, it is not possible to extrapolate exactly the
route of every aircraft, specially for the Portuguese airline TAP Portugal which constantly has aircraft
arriving from an airport and departing a certain time later to a different destination. This information is
crucial to correctly simulate aircraft movement throughout the day in the airport. To obtain it, an online
database which keeps data relatively to the movements of every aircraft separated by airline is used. The
data acquired by the database comes from different sources, the main one being Automatic Dependent
Surveillance-Broadcast (ADS-B) surveillance data [22].
A medium term future traffic sample for the Lisbon Airport is extremely hard to predict as it has
immense factors contributing to it, many of which are dependent on the economical and political situation
of Portugal and the world overall. As an approximation, the future traffic sample is obtained by adding
a certain amount of commercial flights that occurred on different days of 2017 to the base 12th of April
traffic sample. Flights are added throughout the day, with special focus on the morning traffic peak. The
added flights can be seen in Appendix B.2.
NAV Portugal also kindly provided a data sheet with 1078 departures and 1020 arrivals collected
in 2010, with important information to the simulation such as ROTA, ROTD, Flight Crew Reaction to
Line-up Clearance (FRLC) and Flight Crew Reaction to Takeoff Clearance (FRTT).
18
4Implementation
Contents
4.1 LPPT Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2 Flights . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
19
20
4.1 LPPT Model
The content present in this section is relative to the modeling of the current state of the LPPT, which
serves as a base to obtain results and identify further possible enhancements to the airport. The
changes done to the Lisbon Airport in the end of 2016 and beginning of 2017, published in the latest
electronic Aeronautical Information Publication (eAIP) package, are included and modeled. Although
two different models are created, one for each runway in use, some of the creation steps are common
between the two models. When not, emphasis is given to the differences between models.
In a first contact with the simulation tool, a 2D world map is presented, as seen in Figure 4.1. This is
because Simmod PLUS! uses the geographical coordinate system, and all of the simulation nodes are
geographically defined by longitude, latitude and altitude. As such, the first step in the creation of the
model is to introduce the known geographical points of the airport.
Figure 4.1: Simmod PLUS! world map.
By consulting the Portuguese eAIP [7], the geographical coordinates of the LPPT runway thresholds
and ends, as well as of the stands checkpoint locations are obtained and introduced into the model. In
order to fully design the rest of the nodes and links in the airfield, the LPPT ICAO Aerodrome Chart,
that can be seen in Appendix A.1, is integrated in the model as a background picture and scaled taken
into account the reference nodes introduced with geographical coordinates. After this step, the rest of
the nodes and links of the airfield can simply be drawn over the scaled chart of the aerodrome. Even
though the airfield has slight variations in altitude, these are not taken into account and all airfield nodes
are given the aerodrome reference altitude of 114 meters (374 feet). The final stage of the model airfield
can be seen in Figure 4.2 (note that visually the airfields for the two different models are similar).
21
Figure 4.2: Model airfield.
As for the airspace, the initial and final seven nautical miles traveled by departing and arriving aircraft,
respectively, are modeled. To do so, seven nautical mile length air links are drawn, aligned with the
runway 03/21 and with starting points on this runway thresholds. The altitude of the outer nodes is
dependent on whether arriving or departing aircraft travel through them, which is dependent on the
runway in use (and consequently on the model, since there are two different models in order to simulate
each of the runways in use). For departing aircraft, the altitude of the node seven nautical miles away
from the runway threshold, that from now is addressed as departure node, is 5000 feet. This is a rough
approximation since the altitude of an aircraft after take-off depends strongly on the aircraft performance.
For arriving aircraft, the altitude of the node seven nautical miles away from the runway threshold, that
from now is addressed as arrival node, is 2600 feet. Since at seven nautical miles from the runway,
aircraft are already established on ILS approach, which as a predetermined slope, the altitude of arriving
aircraft within seven nautical miles from the runway threshold is approximately constant and can be
consulted on the LPPT ICAO Instrument Approach Charts [7]. Besides the air links for arriving and
departing aircraft, air links are added for aircraft that need to execute a missed approach procedure,
although this links do not represent the exact route of such procedure. Instead, they simply form an
oval route which ensures that aircraft that missed their approach circle back to the approach initial node
(seven nautical miles away from the runway threshold) and perform another landing try. The modeled
airspace can be seen in Figure 4.3, where all air routes are represented in blue (since directionality is
not drawn in the figure, the two different models are visually similar).
22
Figure 4.3: Model airspace.
4.1.1 Airspace
The airspace is where aircraft are injected and ejected from the simulation. Arriving aircraft are injected
in the arrival node and departing aircraft are ejected in the departure node. If due to minimum separation
requirements on the air links arriving aircraft need to make a queue at the arrival node, the queuing logic
used is First In First Out (FIFO). This logic is valid not only for the injected aircraft at the arrival node
but also to aircraft that arrive to the arrival node after having performed a missed approach. Aircraft
that in the simulation are in a queue at the arrival node do not accurately represent the movement
or delay of aircraft in reality. Instead, they represent the necessity for air traffic controllers to delay
aircraft, somewhere in the TMA zone, prior to the last seven nautical miles before the runway, in order
to respect minimum separation requirements. As the exact sources of this delay are not simulated and
can’t therefore be calculated, conclusions are not drawn relative to the delay of aircraft while in the air.
23
In order to simulate the minimum separation requirements and speeds used by the aircraft on the
simulated airspace, aircraft are separated into airspace groups. In the simulation, the airspace groups
are created according to the ICAO wake vortex categories as well as aircraft performance, which results
in the following groups: Heavy, Medium-Jet, Medium-Prop and Light. Since the entirety of the simulated
airspace is in the vicinity of the airport and at a relatively low flight level, the minimum separation re-
quirements are the same in every air link and given according to the ICAO regulations [23]. Table 4.1
presents the separation minimum applied for every par of leading/trailing aircraft group, where the values
are in nautical miles.
Leading
TrailingHeavy Medium-Jet Medium-Prop Light
Heavy 4NM 5NM 5NM 6NM
Medium-Jet 3NM 3NM 3NM 4NM
Medium-Prop 3NM 3NM 3NM 4NM
Light 3NM 3NM 3NM 3NM
Table 4.1: Aircraft separation minimum.
As described in Section 3.1, the variables of the model are only changed at discrete points in time,
when events occur. This means that an aircraft traveling through a link can’t change its current speed
until it arrives at the next node, which is where the next event related to this specific aircraft takes
place. As a result, departing and arriving aircraft have constant speeds when traveling through the
simulated initial and final seven nautical miles of the airspace, respectively. After consulting experienced
air traffic controllers that operate in the Lisbon tower, the values for the TAS of aircraft on their initial
and final seven nautical miles are estimated and can be seen in Table 4.2. Note that this values are
a gross approximation, not only because the speeds of arriving and departing aircraft are in reality not
constant on the initial and final seven nautical miles, but also because they change within aircraft group,
depending on factors such as aircraft type, aircraft weight, pilot and airline procedures.
Airspace Group Arrival Departure
Heavy 140KT 175KT
Medium-Jet 140KT 175KT
Medium-Prop 130KT 160KT
Light 120KT 130KT
Table 4.2: TAS of aircraft arriving and departing from the airport.
24
At recommendation from the air traffic controllers, the ICAO code A343 aircraft is given its own
airspace group, due to its performance after take-off being significantly lower than the other jet aircraft
(note that both heavy and medium-jet aircraft have attributed the same speeds both for arrivals and
departures). The departure speed given to the A343 is the same as for a medium-prop, 160 knots.
Since the purpose of the simulated missed approach route is not to accurately represent a real missed
approach procedure, but simply to redeliver aircraft to the approach route in order to perform another
landing try, aircraft traveling through the missed approach route are given the same speed as if they
were in the arrival route.
It is important to notice that internally the simulation does not deal with distance based separations.
Instead, the simulator transforms the given distance separations to time separations, taking into account
the speed of the aircraft trailing the leading aircraft. The equation used by the simulator is given as:
∆t =∆x
Vt(4.1)
where:
∆t − Time based separation
∆x− Distance based separation
Vt − Speed of the trailing aircraft
As an example, if two medium-jets are arriving separated by the minimum separation value of three
nautical miles given in Table 4.1, and both traveling at the speed of 140 knots given in Table 4.2, the time
between the first aircraft lands, and the second aircraft lands is given exactly by:
3NM
140KT× 3600s ≈ 77.14 seconds (4.2)
This would mean that the same pair of aircraft arriving at any air node separated by a minimum
separation would always have the same time interval, which is indeed a big approximation. In reality,
although the purpose of accurately spacing aircraft is carefully done by air traffic controllers, the time
intervals between two succeeding aircraft are certainly not constant and have some sort of randomness
introduced into them. In order to model such randomness, the minimum separations presented in Ta-
ble 4.1, which are the ∆x variable in Equation 4.1, are added to a value according to a scaled Beta
distribution, with PDF represented in Figure 4.4.
25
Figure 4.4: Separation increment probability density function.
This separation increment, defined in the interval [−0.25 0.25] nautical miles, is given to every min-
imum separation found in Table 4.1. This means that every time the simulation needs to impose a
minimum separation, it first draws a value from Table 4.1, according to the type of aircraft being sepa-
rated (deterministic), and then draws a value from the defined probability distribution (random) and add
it to the table determined value.
4.1.2 Runway
The runway is one of the most complicated modeled systems with a big amount of procedures and
variables associated with it. It is also where the runway in use has a bigger influence, and a majority of
the differences between the two models are located in this part of the airport. As mentioned in Section
3.1, aircraft in the simulation travel from node to node, through links. As such, the runway is simulated as
a group of links, with nodes being located at the runway thresholds, ends, touchdown zones, which are
located approximately 300 meters (984 feet) from the runway thresholds and every intersection between
the runway and taxiways (note that RWY17/35 is treated as a normal taxiway). The modeled runway
with all nodes highlighted in yellow can be seen in Figure 4.5.
Figure 4.5: Model runway.
26
4.1.2.A Runway Distances
The runway distances are summarized in Table 4.3. The positions represent the locations in the runways
from which an aircraft is allowed to start its takeoff roll, and their name is given according to the name of
the taxiway that leads to them. The threshold column gives the distance the threshold is displaced from
the runway end. The other columns in the table are the Landing Distance Available (LDA), Takeoff Run
Available (TORA), Takeoff Distance Available (TODA) and Accelerate-Stop Distance Available (ASDA).
The TODA is calculated by adding the TORA to the clearway, an area beyond the paved runway but
still in control of airport authorities. In the LPPT, there is a clearway of 100 meters on both sides of
the runway. The ASDA is calculated by adding the TORA to the stopway, an area beyond the runway,
normally paved, which can be used to decelerate in the event of a rejected or failed takeoff. The LPPT
has no declared stopway and therefore the ASDA and the TORA have the same values.
Runway Threshold LDA TORA TODA ASDA
03 90m 3617m - - -
03 (Position M) - - 3705m 3805m 3705m
03 (Position N) - - 3631m 3731m 3631m
03 (Position P) - - 3007m 3107m 3007m
21 500m 3205m - - -
21 (Position S) - - 3805m 3905m 3805m
21 (Position U) - - 2410m 2510m 2410m
Table 4.3: Runways declared distances [7].
By analyzing Table 4.3 it is clear that position U in runway 21 has significantly less runway available
for performing the takeoff roll than any other departure position in any of the runways. This obligates
aircraft that need a longer distance to takeoff, normally aircraft of type heavy, to choose position S as
their departure point when runway 21 is in use. However, by analyzing the airport layout in Figure 4.2
and in Appendix A.1, it is clear that in order to reach the departure position S aircraft must cross the
runway. This limits the capacity of the airport when runway 21 is the runway in use.
4.1.2.B Departure Queues
Departure queues are nodes in the simulation that serve as an entry point to the runway. If the runway is
occupied when an aircraft arrives at one of this nodes, it will wait in the queue until clearance for takeoff
is given. Each departure queue can hold a different number of aircraft depending on the layout of the
airfield, as an excessively big line of aircraft waiting to depart can block taxiways and provoke delays to
other aircraft. Table 4.4 presents the size of each departure queue in the LPPT, where the name of the
departure queue is given according to the runway in use and taxiway that it is located in.
27
Departure Queue Queue Size
03-M 3
03-N 3
03-P 1
21-U 6
21-S 6
Table 4.4: Departure queues and their sizes.
In the simulation, before aircraft begin to taxi to their assigned departure queue, they check if the
number of aircraft waiting in this departure queue plus the number of aircraft taxiing to this departure
queue does not exceed the departure queue size. If it does, the aircraft will instead wait at the gate
until the departure queue is no longer full. This simulates the tower air traffic controllers purpose of not
having big lines of aircraft waiting to depart, blocking necessary taxiways.
Figure 4.6 gives an example of aircraft queuing for departure in departure queues 03-M and 03-N
obtained with the current traffic sample, described in Section 3.5. Since construction work was done in
this area of the airport in late 2016 and beginning of 2017, departure queues 03-M and 03-N are treated
by air traffic controllers as a single dynamic departure queue. In order to simulate this, aircraft assigned
to any of this two departure queues check the occupation of both departure queues, and proceed to
the emptiest one. Only if both departure queues are full, with a total of six aircraft occupying them, an
aircraft waits at the gate until one of the two departure queues is no longer full.
Figure 4.6: Aircraft queuing for departure example.
28
Airspace Group
Departure Queue03-M/N 03-P 21-U 21-S
Heavy 100% 0% 5% 95%
Medium-Jet 100% 0% 80% 20%
Medium-Prop 100% 0% 100% 0%
Light 0% 100% 100% 0%
Table 4.5: Departure queues usage percentage for airspace groups.
From the data relative to the 2010 traffic sample provided by NAV Portugal, it is possible to determine
the usage percentage of each departure queue for every airspace group, presented in Table 4.5. Note
that the majority of aircraft from the airspace group Heavy, when runway 21 is the runway in use, use
departure position S (departure queue 21-S). As discussed in Section 4.1.2.A, in order to reach this
departure queue aircraft need to cross the runway, which not only delays aircraft that need to cross the
runway but also imposes limitations on the runway capacity.
4.1.2.C Line-up and Takeoff
In operational reality, the process of an aircraft going from holding in its departure queue to having his
wheels of the ground is done in a series of successive steps, each of them being crucial to do the process
as fast and effective as possible. First, a line-up clearance is given, meaning that the aircraft is allowed
to line-up with the runway. After, the pilot needs to react to the line-up clearance, a duration defined
as the time between the line-up clearance is received, and the aircraft starts its movement, known as
FRLC. Note that in conditions of clear weather the pilot is normally given a conditional clearance to
line-up, meaning that as soon as the previous arrival passes the departure queue at which the aircraft is
holding, the pilot is cleared to start the line-up, without the need of any further communication between
the pilot and the air traffic controllers. Then the aircraft needs to line-up with the runway. The time
between an aircraft starts moving from its departure queue until it comes to a full stop, lined-up with
the runway, is called Line-up Time (LUPT). After the runway is considered to be clear to execute the
takeoff, air traffic controllers give a takeoff clearance to the pilot, which then reacts to it and proceeds
to perform the takeoff roll. The time between the takeoff clearance is given and the aircraft starts doing
its takeoff roll is commonly referred to as FRTT. Note that some of this times are conditional to the type
and frequency of operations being performed in the airport. If an aircraft wants to depart and there are
no constraints on the runway by any incoming arrival or previous departure, as well as no aircraft waiting
to depart, a pilot might receive a take-off clearance while taxiing to the runway in which case there is no
need to stop the movement of the aircraft at all. In the same way, if there are no aircraft at a departure
queue waiting to depart, and a taxiing aircraft wants to depart but the runway is currently blocked only
due to a recent departure, a pilot might receive a line-up clearance while taxiing to the runway.
29
Simmod PLUS! does not deal with communications between pilots and air traffic controllers, which
complicates the task of simulating the described procedure. However, it does allow to define a time
interval that aircraft remain stopped after being cleared to leave a departure queue. Note that although
possible through complex simulation procedures, the created model does not simulate line-up, meaning
that if it is necessary to hold at a departure queue due to the runway being currently occupied, aircraft
hold in the departure queue until the runway is clear (which happens either when the previous arrival
vacates the runway, when long enough time has passed since the previous departure did its takeoff
roll, or when an aircraft finishes crossing the runway). In order to accurately reproduce reality, the time
in the simulation between an aircraft is cleared to leave its departure queue, and starts performing its
takeoff roll needs to equal the FRTT. In simulation, this is given by the time an aircraft takes to move
from a departure queue to the runway, which in the simulation is deterministic, plus the defined extra
time interval that aircraft hold at a departure queue after being released from it, that can be given as a
probability distribution. This is described in Equation 4.3.
FRTT = Taxi to Runway + Extra T ime at Departure Queue (4.3)
Figure 4.7: FRTT probability density function.
The FRTT is extracted from the data provided by NAV Portugal and approximated to a Beta distribu-
tion, which can be seen in Figure 4.7. Note that in order to reproduce this probability density function
in the simulation for every departure queue, different probability density functions have to be defined for
the Extra T ime at Departure Queue variable in Equation 4.3, as for each different departure queue, the
Taxi to Runway variable is different, although the result, FRTT, needs to be the same.
30
The takeoff rolls, more specifically the time aircraft take from starting the takeoff roll until the wheels
are off the ground, is also in the data provided by NAV Portugal. This variable is highly dependent
on the aircraft model and weight as well as on runway conditions. In the simulation, this time can be
defined as a probability distribution for each airspace group. For every departure, the simulation verifies
which airspace group the departing aircraft belongs to, and draws a value from the probability density
function defined for that airspace group. It does not take into consideration the aircraft weight or runway
conditions, but it serves as a good approximation by taking into account the data provided by NAV
Portugal. Aircraft models that no longer operate in the Lisbon Airport are excluded. All the takeoff roll
times are defined as Beta distributions, with α = β = 2. The scaling of the distribution for each airspace
group can be seen in Table 4.6 and the base probability density function in Figure 4.8.
Airspace Group Minimum Maximum
Heavy 30s 50s
Medium-Jet 25s 45s
Medium-Prop 20s 40s
Light 20s 40s
Table 4.6: Takeoff rolls duration.
Figure 4.8: Takeoff roll base probability density function.
Note that since α and β are equal, the average of every probability density function defined for each
airspace group is the average of the minimum and maximum values presented in Table 4.6. By observing
Figure 4.8, it is easily seen that the probability density function for the takeoff rolls has a bigger variation
than the one for FRTT, presented in Figure 4.7 (lower α and β). This is an expected result due to the
higher amount of factors that influence the takeoff roll of an aircraft.
31
4.1.2.D Landing
A landing aircraft, specially in high traffic hours, is informed to vacate the runway as expeditiously as
possible [24]. This happens because air traffic controllers are only allowed to give clearance to the
following departure to takeoff after the arriving aircraft has vacated the runway. As such, minimizing
ROTA is crucial to improve the capacity of an airport. ROTA depends on the weather conditions, aircraft
model, location of runway exits, runway conditions and even on the pilot. In the model, an arriving
aircraft chooses a certain runway exit taking into account the airspace group that it belongs to. Then,
the simulator draws a ROTA from a user defined probability distribution attributed to that airspace group,
for that runway exit. All runway exits usage probabilities for each airspace group and respective ROTA
are extracted from the traffic sample provided by NAV Portugal. Once again, the aircraft models that no
longer operate in the Lisbon Airport are excluded from the sample. The runway exits usage probabilities
are shown in Table 4.7 for runway 03 and in Table 4.8 for runway 21, where the runway exit name is given
according to the name of the taxiway to which aircraft vacate. Note that in runway 03, no aircraft vacates
through runway exit S4. This is an important data as any aircraft that vacates the runway through runway
exit S4 needs to cross the runway in order to reach its gate, which imposes limitations on the runway
capacity.
Airspace Group
Runway Exit17/35 S1 HN U5
Heavy 0% 5% 80% 15%
Medium-Jet 25% 5% 65% 5%
Medium-Prop 70% 20% 10% 0%
Light 50% 40% 10% 0%
Table 4.7: Runway 03 exits and usage percentage for airspace groups.
Airspace Group
Runway Exit17/35 HS P N2
Heavy 3% 90% 2% 5%
Medium-Jet 3% 96% 0.5% 0.5%
Medium-Prop 60% 40% 0% 0%
Light 90% 10% 0% 0%
Table 4.8: Runway 21 exits and usage percentage for airspace groups.
32
The minimums and maximums of every defined probability distribution for every combination of run-
way exit and airspace group can be seen in Tables 4.9 and 4.10 for runway 03 and runway 21, respec-
tively. Runway exits are ordered from left to right according to their distance to the runway threshold
(in ascending order). The base probability density function used is the same as for the takeoff roll, in
Figure 4.8, as there are as well several factors affecting ROTA. Once again, since α and β are equal,
the average of each probability density function is equal to the average of the minimum and maximum
values. Note that in Table 4.9, where ROTA for runway 03 are represented, even though runway exit S1
is physically located at a shorter distance from the runway threshold than runway exit HN, the average
ROTA are smaller for runway exit HN for every airspace group except Light. By analyzing Figure 4.2 and
Appendix A.1 it is visible that not only both runway exits are extremely close to one another, but also that
in order for an aircraft to vacate the runway via taxiway S1, it needs to significantly have lower speed
when compared to runway exit HN, since the turn to taxiway S1 has an angle superior to 90◦, while the
turn to taxiway HN has an angle inferior to 45◦. Aircraft of the airspace group Light do not have this
tendency applied to them since normally when they reach this part of the runway their speed is already
low, independent of which runway exit they intend to take. Besides that, it is also easier for a lighter
aircraft to make such a sharp turn than for an heavier aircraft.
Airspace Group
Runway Exit 17/35 S1 HN U5
Min Max Min Max Min Max Min Max
Heavy - - 50s 70s 40s 70s 50s 80s
Medium-Jet 35s 65s 40s 70s 40s 60s 50s 65s
Medium-Prop 45s 55s 55s 75s 40s 70s - -
Light 40s 60s 50s 70s 55s 75s - -
Table 4.9: Runway 03 ROTA for each runway exit and airspace group combination.
Airspace Group
Runway Exit 17/35 HS P N2
Min Max Min Max Min Max Min Max
Heavy 40s 45s 40s 60s 60s 70s 75s 85s
Medium-Jet 35s 45s 40s 60s 55s 65s 75s 85s
Medium-Prop 30s 45s 50s 70s - - - -
Light 30s 45s 40s 60s - - - -
Table 4.10: Runway 21 ROTA for each runway exit and airspace group combination.
33
4.1.2.E Runway Crossing
With the analysis of the traffic sample done in Sections 4.1.2.C and 4.1.2.D, it is possible to conclude
that aircraft only need to cross the runway when runway 21 is the runway in use and position S is
chosen by the pilot as a departure position (except for very rare occasions, which are not considered
in the simulation). This happens frequently for aircraft belonging to the airspace group Heavy (95% of
the times) and not so frequently for aircraft from the airspace group Medium-Jet (20% of the times).
Obtaining clearance to cross a runway is in itself very similar to obtaining clearance to line-up. If the
runway is totally free of any operation, an aircraft taxiing to a runway cross might obtain clearance to
cross the runway in which case there is no necessity to stop the aircraft at all in order for it to enter and
then cross the runway. On the other hand, if when approaching a runway cross an aircraft is informed by
the air traffic controllers to hold short of the runway due to it being currently occupied by other operation
(normally an aircraft performing its landing or takeoff roll), the pilot then needs to wait for the air traffic
controllers to give clearance to cross the runway, react to it, and then proceed. Note that as in line-up
clearance, conditional clearance to cross the runway can be given, meaning that as soon as the pilot
of the holding aircraft sees that the landing or departing aircraft has passed the runway point to be
crossed, it has clearance to cross the runway, without the need of any further communication with air
traffic controllers.
In the simulation, runway crossings are simulated and, as in reality, take two different behaviors. If an
aircraft when approaching a runway finds it free of any operation, it proceeds to cross it without making
any stop. If the runway is being used to perform any other operation, the aircraft holds short of the
runway and wait until it is no longer occupied in order to cross. The node at which aircraft evaluate if
the runway is occupied and hold in case necessary is visible in Figure 4.9 by the node at which the red
aircraft (which is stopped) is located (for the other points at which it is possible to cross the runway, the
holding nodes are at a similar distance from the runway). In this case, another aircraft, in green color, is
performing its landing, therefore the red aircraft needs to hold.
Figure 4.9: Runway crossing example.
34
An aircraft holding short of a runway waiting to cross it, such as the red aircraft in Figure 4.9, con-
siders the runway to be cleared not when no operation is blocking the runway anymore (as happens
with aircraft holding in departure queues), but as soon as the landing or departure that is blocking the
runway passes the intersection that the holding aircraft intends to cross. After the holding aircraft con-
siders the runway to be cleared, it waits a certain amount of extra time before it starts crossing the
runway, in order to simulate both the time a pilot takes to react after the aircraft occupying the runway
has passed its crossing point (in case conditional clearance has been given) as well as the time that
air traffic controllers require to give the pilot crossing clearance and the time the pilot needs to react to
it. After discussing with experienced air traffic controllers from the Lisbon tower, and analyzing the data
from the traffic sample provided by NAV Portugal for FRLC, which has a similar duration to the time it
takes for an aircraft to start its movement after the runway is clear to be crossed, the probability density
function to select an extra holding time for every time an aircraft needs to hold short of a runway in order
to cross it is shown in Figure 4.10. In case after a runway is cleared from a previous operation, both
an aircraft waiting to cross and an aircraft waiting to depart request the simulator to use the runway,
priority is given to the crossing aircraft. In operational reality, when required, aircraft waiting to depart
are given clearance to line-up and at the same time aircraft waiting to cross are given clearance to cross
the runway. However, due to the way the simulated aircraft do their line-up and takeoff, explained in
Section 4.1.2.C, giving priority to the crossing aircraft when both operations request the runway is the
accurate way to simulate what happens in reality.
Figure 4.10: Runway crossing extra time probability density function.
35
4.1.2.F Runway Procedures
Two different kinds of procedures are performed in the runway, departures and arrivals. Besides this
procedures, aircraft that cross the runway also occupy it for a certain period of time. In order to efficiently
operate the runway and achieve its maximum capacity, accurate separations need to be given between
the different combinations of successive procedures. For the Arrival-Arrival case, separation minimum
on final approach is given in distance as represented in Table 4.1. This is because ICAO regulations
regarding minimum separation due to wake turbulence provide enough separation for the leading aircraft
to vacate the runway before the trailing aircraft lands.
In a Departure-Departure situation, separation is given in time, according to aircraft performance
after takeoff. The performance groups for the specific case of aircraft models included in the two traffic
samples considered in the simulation, the current and the medium term future one, can be seen in Table
4.11.
Group 2 Group 1
A310, A319, A320, A321, A332 A343
B734, B737, B738, B739 AT43, AT75, AT76
B752, B762, B763, B77W
C25B, E190, E195, GLF5, LJ35
Table 4.11: Aircraft groups according to performance after takeoff.
According to ICAO documentation, the Departure-Departure separation between aircraft of similar
performance after takeoff should have a minimum separation of one minute, as long as the routes that
aircraft follow after takeoff diverge by at least 45◦ [23]. As in Lisbon the airspace is extremely constrained
by military operations, as described in Section 1.2.1, aircraft take a straight departure route after takeoff
and in line with the runway. As a result, two aircraft from the same group in Table 4.11 departing one after
another are given a separation of two minutes. In case an aircraft with better performance departs in
front of one with less performance (in Table 4.11, an aircraft of group 2 departing in front of an aircraft of
group 1), one minute is subtracted from the base separation of two minutes, and therefore the separation
given is one minute. In the opposite case, an aircraft with worst performance (group 1) departs in front of
an aircraft with better performance (group 2), one minute is added to the base separation of two minutes,
and therefore the applied separation is three minutes.
36
For a Departure-Arrival case, no time or distance separation can be given, as it is unrealistic to stop
an arriving aircraft on final approach. Therefore, in the simulation, a departure procedure does not have
the ability to signal an arrival to stop in order for it to take place. However, if at the time an arrival crosses
the runway threshold node, a previous departure still occupies the runway, the simulator considers that
the arrival had to execute a missed approach, and transfers it to the established missed approach route,
seen in Figure 4.3.
The Arrival-Departure case has two different types of procedure blocking (the departure is the
blocked procedure). After landing, an arrival blocks a departure not for a determined time interval,
but until it vacates the runway. As soon as the arrival has vacated the runway (in operational reality it is
considered an aircraft has vacated the runway when its tail is off the runway), it no longer blocks the fol-
lowing departure. If a departing aircraft at a departure queue requests to start its departure procedure,
but an arrival is already within three nautical miles of the runway threshold, the departure is blocked
from leaving the departure queue and performing its departure procedure. In reality, the distance from
the runway threshold at which an arrival blocks a departure from lining up and taking off is not a fixed
value, but in consultation with the Lisbon tower air traffic controllers three nautical miles is considered to
be the best approximation.
It is understandable from the separation requirements described, that a special case occurs when
the procedure sequence to be performed is Arrival-Departure-Arrival. If the Arrival-Arrival minimum
separations shown in Table 4.1 are kept, it is not possible to perform a departure procedure between
arrivals. This leads to a need of increasing separation requirements between arrivals in order to perform
a departure procedure in between them. This task is done in the simulation every time an aircraft needs
to hold at a departure queue or at a runway crossing point. In general, the new minimum separation
between every pair of airspace groups arriving is six nautical miles, where the separation increment
probability distribution shown in 4.4 still applies. Note that the maximum runway capacity is achieved
when the sequence of procedures is always Arrival-Departure-Arrival. However, from the Lisbon Ca-
pacity Enhancement Exercise done in 2011 by NAV Portugal, it is seen that the maximum number of
movements per hour achieved in runway 03 is bigger than in runway 21 [24]. This is because, ac-
cording to air traffic controllers from the Lisbon tower, the indicated six nautical miles separation in an
Arrival-Departure-Arrival situation needs more often than not to be increased, not only due to the runway
exits configuration, which leads to the average ROTA in runway 21 being bigger than in runway 03, but
also because of the runway crossings that constantly take place when runway 21 is the runway in use
for aircraft models belonging to the airspace group Heavy to reach departure position S. As a result,
in the simulation model regarding runway 21, the target separation of six nautical miles in a Arrival-
Departure-Arrival case is increased to 6.25 nautical miles (the separation increment shown in Figure 4.4
still applies).
37
4.1.3 Taxiways
The movement of aircraft through taxiways between the runway and the gates is fully simulated, through
ground nodes and links (all represented in color green in Figure 4.2). Each ground link has attributed a
speed with which aircraft travel through, according to its location and physical characteristics. Although
it is possible to define in the simulator different speeds for different aircraft groups while traveling through
the same link, such level of detail is not considered and all aircraft have the same speed according to
what link they are in. Figure 4.11 shows the different speeds attributed to different taxiways (in each
image the taxiways highlighted in magenta have attributed the indicated speed). The criterion used is
as follows:
• Turns with more that 90◦ are attributed 10 knots.
• Straight links close to stands and turns with less than 90◦ are attributed 15 knots.
• Straight links away from stands and with medium length or in the vicinity of the runway are at-
tributed 20 knots.
• Straight links away from stands and with high length are attributed 25 knots.
Besides the criterion above and shown in Figure 4.11, aircraft have the speed of five knots in the links
immediately leading to the standing position at the gates (represented in orange color in Figure 4.11).
The runway 03 high speed exit, taxiway HN, has 30 knots as attributed speed and runway 21 high speed
exit, taxiway HS, has 40 knots as attributed speed. The reason for the difference in speed between the
two different runway high speed exits is the high degree turn immediately after the high speed exit HN
of runway 03. This turn requires aircraft to be at a lower speed than usual in high speed exits, which
increases the average ROTA of runway 03 and therefore decreases the capacity of the runway. Besides
speeds, airfield links also have attributed to them directionality, meaning that except for rare exceptions,
aircraft are only allowed to travel a link in a certain direction. This is done in order to model the flow
of aircraft on the ground which can be seen in Appendix A.2 and A.3 for runway 03 and runway 21
respectively.
Every time an aircraft needs to move from any gate to any departure queue, or from any runway exit
to any gate, the simulator internally determines what is the shortest path in time between the two points,
and moves the aircraft through that path. Note that the shortest path in time is not always the same as
the shortest path in distance, due to the differences in speed between the ground links in the simulation
(as shown in Figure 4.11).
38
(a) 10KT (b) 15KT
(c) 20KT (d) 25KT
Figure 4.11: Model taxiway speeds.
39
During the taxiing process, aircraft may come to a conflict with other aircraft also taxiing in the airfield.
This events are also simulated in two ways. First, certain ground links block other ground links from being
used. This occurs mainly where two taxiways cross each other or merge together, and differs from one
model to the other. An example of this situation is shown in Figure 4.12, obtained with the current traffic
sample for the runway 03 model. Both the aircraft in yellow and red are traveling to a departure queue
in runway 03 and come to a conflict. As the aircraft in yellow is the first to arrive to the links that are
defined to block each other, it is given priority and does not stop its movement during the taxiing process
(yellow color indicates that the aircraft is taxiing). On the other hand, the aircraft in red after arriving
to the position where it is shown detects the yellow aircraft moving through the blocking links and must
therefore stop, until the links are cleared (red color indicates that an aircraft is stopped). The second
way in which aircraft conflicts are simulated happens in links where aircraft are allowed to travel in both
directions (occurs mainly in taxiways close to gates). In order to never have a head to head situation
where neither of the aircraft can move forward, certain groups of links are attributed to dynamic paths,
where aircraft are only allowed to move in one direction. If an aircraft needs to stop when entering
a dynamic path, due to it being currently used in the opposite direction to which the aircraft wants to
travel, it holds until the path is free of aircraft traveling in the opposite direction before it proceeds (after
holding 60 seconds, the aircraft signals the simulation to stop allowing aircraft to enter the dynamic
path in the direction opposite to which it wants to travel, so that it can proceed without having a huge
delay). An example of this situation is provided in Figure 4.13, where the small yellow aircraft is traveling
downwards, and therefore the red aircraft, that wants to turn left and travel upwards, needs to hold in
order to avoid a head to head situation (note that both aircraft will travel the same taxiway in opposite
directions).
Figure 4.12: Blocked links due to taxiing aircraft example.
40
Figure 4.13: Blocked links due to dynamic path example.
In any of the two described situations, after being no longer blocked from moving, the aircraft does
not start its movement immediately. Instead, it waits a certain amount of extra time, according to a
probability distribution. The extra time aircraft hold after being cleared to move is drawn from the same
probability distribution obtained for extra time to cross the runway, represented in Figure 4.10. Note that
although for different reasons, extra times after aircraft being blocked exist in order to simulate the same
situation, pilot reaction time either to visually observing that the aircraft is free to proceed its movement
or to air traffic controllers clearance for the aircraft to move.
4.1.4 Gates
In the model, all 83 stands of the LPPT are individually simulated. Every aircraft that lands has a stand
as destination, and every aircraft that wants to depart starts its movement from a stand. The military
airbase within the airport is also simulated but as a single stand, with capacity for more than one aircraft.
Every other 83 stands are set to be able to host a single aircraft. In reality, stands are not able to host
every type of aircraft model, and usually have a critical aircraft model assigned to them (meaning that
aircraft models bigger than the critical one are not able to use the stand). In the model, this is done by
separating the aircraft models into ground groups according to the ICAO Aerodrome Reference Code
(A - F) and then assigning the groups that are allowed to use each gate. Note that certain pairs of
stands can not be used simultaneously, due to space constrictions. This is taken into consideration in
the simulation by blocking the use of certain stands, when other stand is being used.
Gates are also grouped into aprons, with each apron usually having a single type of flight or airline
attributed to it (e.g. cargo flights usually use the same apron, commercial flights aprons are usually
separated by flights with destination Schengen and non Schengen area). Table 4.12 shows the number
of stands in each apron and the general purpose for which each apron is normally used. Based on the
description of each apron, airlines are attributed to the aprons, meaning that only the defined airlines
are allowed to use each apron.
41
Apron Stands Description
10 5 Non-contact apron for Schengen area commercial flights
11 4 Terminal 1 contact apron for Schengen area commercial flights
12 5 Terminal 1 contact apron for Schengen area commercial flights
14 7 Terminal 1 contact apron for non-Schengen area commercial flights
20 10 Terminal 2 contact apron for low cost commercial flights
22 5 Non-contact apron for all commercial flights
30 2 Non-contact apron for all commercial flights
40 5 Non-contact apron for all commercial flights
41 6 Non-contact apron for all commercial flights
42 6 Non-contact apron for all commercial flights
50 6 Non-contact apron for all commercial flights
60 10 Non-contact apron for all commercial flights
70 6 Private flights apron and non-contact Schengen area commercial flights
80 6 Cargo flights apron and non-contact Schengen area commercial flights
Table 4.12: Lisbon Airport aprons and related use description.
What happens to aircraft while parked in the stands can be fully simulated, by defining gate service
times for arrivals and departures (i.e. the time that it takes after an aircraft is parked until passengers are
allowed to disembark or the time it takes to prepare a parked aircraft to be able to receive passengers)
and defining passenger boarding and unloading times. However, although the level of detail would
provide a better understanding of the gate capacity in the Lisbon Airport, this is out of the scope of this
work and is not modeled. As a result, all gates have their service times set to zero. As for boarding and
unloading times, all gates have 15 minutes boarding and 15 minutes unloading time, except for low-cost
airlines, which have both times set to 12 minutes and 30 seconds. In the simulation, this prevents a flight
that arrives later than scheduled from leaving the gate immediately after it has arrived (i.e. every flight
stays at least 30 minutes at the gate, except for low cost flights that stay at least 25 minutes at the gate,
independently of how late the flight is).
The process of pushing back an aircraft from its stand to a taxiway with the help of a pushback tug
in order for the aircraft to be able to start its forward movement is also simulated (in the Lisbon Airport,
all stands require aircraft to be pushed back with a tug, except for stands 701, 702 and 703 in apron
70, where aircraft are allowed to move forward directly from the standing position). The speed at which
aircraft are pushed back is three knots. Figure 4.14 shows an example of an aircraft pushing back, in
yellow color, as well as a parked aircraft in purple color.
42
(a) Pushing back (b) Dwelling
Figure 4.14: Model gates, pushback and dwelling example.
In Figure 4.14, note as well that the gates being used are blocking all the other shown gates that
have approaching link non parallel to the ones being used. In brown color is shown an aircraft after
the pushback movement is completed. Here, the pushback tug needs to detach from the aircraft and
clear the way for it to move, as well as final preparations and checks for the flight are made by the pilot.
Minimizing this dwell time is a task of great importance as aircraft stand still after being pushed back,
blocking a taxiway, which can delay other aircraft. In order to simulate this, every time an aircraft finishes
its pushback movement, the simulator draws a duration from the probability density function represented
in Figure 4.15 which corresponds to the extra time aircraft stand stopped after the pushback movement
is finished. Although the simulator allows for different probability distributions to be defined for different
aircraft models and different gates, such level of detail is not considered, and the same probability
distribution is attributed independently of gate or aircraft model.
Figure 4.15: Dwell time probability density function.
43
4.2 Flights
In the simulation, every flight is defined as an arrival, that then proceeds to its gate, waits for its depart-
ing time, taxis to the runway and then depart. In order to have a correct flow of aircraft throughout the
simulated day, flights that stay overnight in the airport from the previous day are set to arrive in the simu-
lation at 00:00 hours. In the same way, flights that stay overnight in the airport to the next day are set to
depart and leave the simulation at 26:00 hours, meaning 03:00 hours of the following day. Each flight is
attributed an airline and a flight number, as well as an aircraft model (the same aircraft flight number can
and in most cases will differ from arrival to departure, and can differ as well in airline, although it rarely
happens). The time at which aircraft are injected in the simulation needs to be calculated, as it is not
the same as the scheduled ETA or ETD. For arrivals, the time aircraft take to go through the modeled
final approach seven nautical miles needs to be subtracted from the ETA. Although it would be more ac-
curate to take into account the speed at which each different aircraft model goes through its final seven
nautical miles, such level of detail is not considered and all arrival injection times are obtained by sub-
tracting three minutes to the ETA. Departures are also attributed injection times, although the departing
aircraft are already in the simulation (note that every departure in the simulation must previously arrive
to the airport). The departure injection time refers to the time aircraft start performing the necessary
procedures in order to depart (in the model, as the gate service time is defined to be zero minutes, the
injection time of a departure is the time an aircraft starts boarding passengers, as explained in Section
4.1.4). Departure injection times need then to take into account the time needed to board passengers
and the time to taxi to the runway. As described in Section 4.1.4, all boarding times are defined to be
15 minutes, except for low cost flights. As an approximation, all ETD are subtracted 15 minutes when
calculating departure injection times. To take into account taxi times, it would be accurate to individually
calculate each flight taxi time, according to runway in use, runway entry point chosen and flight gate.
However, for the sake of simplicity, the same taxi time of five minutes is considered for every gate and
runway in use. As a result, in order to calculate the departure injection times from the scheduled ETD,
20 minutes are subtracted, 15 for the boarding time and five for the taxi time.
In order to better simulate reality, calculated arrival and departure injection times are added to a
time drawn from a defined probability distribution. This is done in order to simulate the probability of
an aircraft arriving or departing late (or early). Note that a more accurate approach is to gather one
year data from the airport and define probability distributions according to the gathered data (which can
be done for each airline or flight). In the absence of such data, general probability distributions are
attributed to arrivals and departures, which can be seen in Figures 4.16 and 4.17, respectively. Note
that the lateness ranges differ, for they intend to simulate different things. While arrivals lateness intends
to simulate delays that happened in other airports or even during a flight, departure lateness intends to
simulate delays that happened while boarding the passengers or preparing the aircraft to depart. Note
as well that if a departure injection time is set to take place before the corresponding arrival finishes
the unloading of passengers, the aircraft starts boarding passengers only after it finishes unloading
passengers.
44
Figure 4.16: Arrival lateness probability density function.
Figure 4.17: Departure lateness probability density function.
In order to correctly simulate aircraft ground movements, each flight should be attributed a specific
gate. The same aircraft has the possibility to arrive to a certain gate and depart from a different gate.
The ground movement of an aircraft between gates is called towing, and is usually done with the aid
from towing vehicles, without the necessity of starting the aircraft engines. Note that the most common
situation is to tow an aircraft from a contact apron to a non-contact apron after unloading the passengers,
and then, shortly before departure, tow the aircraft back to a contact apron in order to perform the
boarding of the passengers. This is normally done in order to free contact gates to be used by other
flights, when a certain aircraft has a big time interval between its arrival and its departure.
45
The ground movement of aircraft between gates as well as the assignment of a specific gate to each
flight is not modeled, as the information needed to do so is not available. As a result, in the simulation,
an aircraft that arrives to a certain gate departs from that same gate, without performing any ground
movement. However, the attribution of gates to flights is not done randomly, but takes into account the
airline, the type of flight, and origin or destination of the flight. Instead of being attributed a specific
gate, a flight is attributed an apron, and proceeds to any free gate within that apron. If there is no
available gate in the apron attributed to a flight, the simulator searches for any other available gate for
that particular flight airline and aircraft model and direct the aircraft to it. Table 4.13 presents the aprons
that are attributed to each type of flight, shown under Primary Aprons, and the aprons that each type
of flight is redirected to, in case the attributed apron has no available gate, shown under Secondary
Aprons. It should be noted that in case a certain flight has no available gate in any of the corresponding
primary or secondary aprons, the simulation stops and provide an error.
Type of Flight Primary Aprons Secondary Aprons
Commercial from Schengen area 10,11,12 22,30,70(partially),80(partially)
Commercial from non-Schengen area 14 40(partially),41,42,50,60
Low cost 20 22,41(partially)
Cargo 80 -
Private 70 -
Table 4.13: Aprons attributed to each type of flight.
46
5Results
Contents
5.1 Number of Iterations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2 Current vs. Future Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.3 Runway 03 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.4 Runway 21 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.5 New Airport Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
47
48
The results are obtained for a time interval of one day, according to the collected traffic samples
being simulated. The very first hours of the simulation, as well as the last ones, are excluded from the
results, due to this being the hours where the overnight flights arrive and depart from the simulation,
therefore having no significant meaning. The data points are obtained as a 15 minutes moving hourly
average, meaning that the analyzed variables are calculated for and averaged through a one hour time
interval, with a 15 minutes increment (e.g. data is collected and averaged between 07:00 and 08:00,
07:15 and 08:15, etc.). The data points have per basis the flights that happened within the defined time
intervals, and not the time intervals themselves. For example, when calculating the average departure
ground delay between 07:00 and 08:00 hours, only aircraft that completed their takeoff roll (wheels
of the ground) in that time interval are considered. Therefore, ground delay that actually happened
before 07:00 hours, but the related aircraft took off between 07:00 and 08:00 hours, is included in the
departure ground delay data point between 07:00 and 08:00 hours. In the same way, ground delay that
actually happened between 07:00 and 08:00 hours, but the corresponding aircraft departed after 08:00
hours, is not included in the departure ground delay data point between 07:00 and 08:00 hours. Data
points represented are not centered, with data relatively to a one hour interval being represented at the
beginning of the interval (e.g. the number of movements between 07:00 and 08:00 hours is represented
at 07:00 hours, not 07:30 hours). The focus of the results is put on movements per hour throughout the
day, average delay and breakdown of the origins of the delay. Note that in this work delay does not refer
to the difference between ETD and ATD or ETA and ATA, but to the amount of time an aircraft has to be
stopped, due to constraint reasons from different sources.
5.1 Number of Iterations
The first step to make, taking into account the randomness and probability distributions introduced in the
model, is to do a convergence study, in order to learn exactly how many iterations of the same data set
are needed in order to obtain meaningful results. To do so, the analyzed variable is the runway maximum
throughput. In order to obtain the maximum throughput, the simulation tool provides a cloning feature,
where several flights can be cloned, creating an extremely high demand, that keeps the runway in con-
stant use. The traffic sample used to test the convergence is the current one, with every flight between
06:00 and 18:00 hours being cloned once. The convergence study is done for the runway 03 model.
However, since the probability distributions introduced to the two different models are very similar and
often the same, there is no need to perform individual convergence studies for each of the models. The
introduced random variable that directly influences the runway throughput is the separation increment,
with probability density function shown in Figure 4.4. Note that given high enough demand, the runway
is always being requested both by departures and arrivals, which makes it being used in a constant
Arrival-Departure-Arrival sequence, at which point the separation increment to the minimum separation
is the only variable that influences the runway maximum throughput and, given enough iterations, the
maximum throughput value should be constant.
49
Figure 5.1: Maximum throughput convergence study.
Figure 5.1, shows the obtained maximum throughput having performed 1, 5, 10 and 20 iterations.
Considering that every flight is cloned between 06:00 and 18:00 hours, the time interval shown and
analyzed is between 10:00 and 15:00 hours, as it is considered to be the time interval where the system
is stable (which in this case means the time interval at which the runway is in constant use and constantly
being requested both by arrivals and departures). By analyzing the figure, a tendency can be seen in the
number of movements per hour as the number of iterations increases. With one iteration, the number
of movements per hour has some oscillation and his naturally always a integer value. As the number
of iterations increases, the line tends to reduce its oscillatory behavior, and converge into an horizontal
line, representing a constant value. The difference between five iterations and 10 or 20 iterations is still
visible, especially in the two low peaks observed between 10:00 and 11:00 hours and between 13:00 and
14:00 hours. The two other lines represented with a higher number of iterations, 10 or 20, are extremely
closer to a horizontal line. Although it is still observable that the line with 10 iterations has a higher
oscillatory behavior than the line with 20 iterations, the difference is not significant when compared to
the difference between 10 iterations and one or five iterations. Therefore, the number of iterations from
the same data set chosen to obtain significant results is 10 iterations. A significant factor also to take
into account when choosing the number of iterations required is the computational time needed to obtain
such number of iterations. As explained, the difference between 10 iterations and 20 iterations, although
existing, is not extremely significant. However, the computational time needed to compute 20 iterations
is the double of the time needed to compute 10 iterations. This is a factor that strengthens the choice
for 10 iterations.
50
5.2 Current vs. Future Demand
The current demand of the airport is approximated to the traffic sample obtained for Wednesday, the
12th of April, which is considered to be representative of the airport average demand. Figure 5.2 plots
the evolution of the airport demand throughout the day, by considering the flights ETA and ETD. Note
that throughout the middle of the day, departures always peak with lag relatively to arrivals. It is also
observable the morning arrival peak, represented between 07:00 and 08:00 hours and the late afternoon
departure peak, between 18:00 and 20:00 hours. Figure 5.3 shows the medium term future demand
obtained with the method described in Section 3.5. The pattern is extremely similar to the current
demand, but now the number of movements, specially in the morning peak, is significantly higher.
Figure 5.2: LPPT current demand.
Figure 5.3: LPPT medium term future demand.
51
Figure 5.4, shows the percentage of aircraft that belong to each wake vortex category, from 05:00
until 24:00 hours, for the current demand (the differences in wake vortex category between both de-
mands are insignificant, therefore only the current demand percentages are shown and analyzed). It is
clear that medium jet aircraft are predominant in the Lisbon Airport, with about 80% of the movements
throughout the day. However, early in the morning, there is a big arrival flow of heavy aircraft, close
to the morning arrival peak in Figure 5.2. Figure 5.5 presents the same breakdown but for departures
(the aircraft belonging to each group can be seen in Table 4.11). Here, no peak of aircraft from Group 1
or Group 2 is observed, with the mix being approximately constant throughout the day. Predominantly,
aircraft of Group 2 use the Lisbon Airport, with the average percentage throughout the day being slightly
over 80%.
Figure 5.4: Arrivals hourly percentage of aircraft by ICAO wake vortex category.
Figure 5.5: Departures hourly percentage of aircraft by performance after takeoff group.
52
5.3 Runway 03
The first parameter to analyze is the runway maximum throughput. As in Section 5.1, this is done
by cloning several flights throughout the day in order to have the runway constantly requested both
by arrivals and departures. The current demand is used to perform this study, although it should be
noted that in this situation there is no difference between using the current or future demand (specially
since the arrivals and departures traffic mix, represented for the current demand in Figures 5.4 and
5.5 respectively, is kept approximately constant). Figure 5.6 shows the runway maximum throughput
achieved under the different simulated conditions of wind intensity and direction. In the figure colors
are set by wind direction. The constant values are obtained by averaging 12 hours of simulation where
the runway is constantly in use. Under no wind conditions, the maximum runway throughput is 44
movements, which is one more than the maximum 43 movements achieved in reality by the airport [24].
This result comes as no surprise taking into account the operational reality in which air traffic controllers
need to work. It is clear that wind strength and direction has a strong impact in the runway achievable
throughput. The worst case scenario is a full headwind with intensity of 30 knots, where the runway
maximum throughput drops to slightly over 36 movements. Full crosswinds, represented in yellow, have
little impact in the runway throughput. However, it should be noted that crosswinds greatly increase the
difficulty of a pilot performing a landing, factor which in simulation is not taken into account and should
not be disregarded [25].
Figure 5.6: Runway 03 maximum throughput for different wind conditions.
53
5.3.1 Current Demand
Figure 5.7 compares the simulation achieved runway 03 throughput, meaning the actual number of total
movements, landings and takeoffs in each hour, with the current demand of the airport, as represented
in Figure 5.2. The throughput of the runway meets the current demand, although when the demand
peaks, the throughput does not peak with it. By analyzing Figure 5.8, representing the average arrival
and departure delay, it can be seen that when the runway throughput does not accompany the demand
peaks, there is an increase in the departures average delay. Note as well that the departure average
delay is constantly and significantly higher than the arrival average delay.
Figure 5.7: Runway 03 throughput vs current demand.
Figure 5.8: Runway 03 average arrival and departure delay for the current demand.
54
Figures 5.9 and 5.10 breakdown the average arrival and departure delay, respectively. It is observ-
able in both figures that the taxi delays, which refers to delays that occurred while taxiing to or from
the runway, are minimal and almost insignificant. This is an expected result taking into account the
strict unidirectional ground movement flow imposed in the Lisbon Airport, shown in Appendix A.2 for
runway 03. It is then clear that the average departure delay origin is runway constraints. The arrivals
air delay origin is not simulated and therefore no conclusions can be given regarding it. However, the
maximum average arrival air delay is one and half minutes, which can be considered acceptable (note
that arrival air delay refers to delaying arrivals in order to respect minimum separation requirements,
which can only be done by air traffic controllers to a certain extent). The difference between average
arrival and departure delay is expected, since arrivals are not blocked due to runway occupation, unlike
departures. Besides that, the Departure-Departure separation requirements are in time more strict than
the Arrival-Arrival separation requirements, producing bigger waiting times at the departure queues.
Figure 5.9: Runway 03 average arrival delay breakdown.
Figure 5.10: Runway 03 average departure delay breakdown.
55
As the maximum achieved runway throughput for the current demand shown in Figure 5.7 is 35
movements, one movement below the maximum runway throughput for the worst wind conditions, which
is 36 movements for full headwinds with 30 knots wind speed, the wind impact on the airport performance
for the current demand is only analyzed for the worst wind conditions. However, this should be taken
as an approximation, as both in simulation and in reality any wind conditions have a certain degree of
impact in the airport performance. That being said, by analyzing the worst case scenario of 30 knots full
headwind it can be assumed that this is the airport poorest performance relatively to any of the other
simulated wind conditions, meaning that the results achieved with the other wind conditions are certainly
between the no wind results and the 30 knots full headwind results.
Figure 5.11 shows the difference between the achieved runway throughput with the current demand
for no wind and 30 knots headwind conditions. The differences are minimal which should come as no
surprise as the maximum throughput of the runway with 30 knots full headwind is still superior to the
maximum achieved throughput for the current demand with no wind. The differences in the average
arrivals and departures total delay under the two wind conditions is shown in Figure 5.12. Here, the
average total delays are significantly higher in 30 knots headwind conditions, with average departure
delay peaking at 10 minutes in the late afternoon departure peak and the average arrival delay peaking
at four minutes in two instances during the day. It should be noted that 30 knots headwind is not
frequent in the Lisbon Airport, and certainly not during the whole day (in the simulation, wind conditions
are considered constant throughout the day). However, Figure 5.12 shows that such situation is not
sustainable by the airport due to the high amount of delay that it provokes in it. Note that even though
the runway throughput is practically unaffected by the strong headwinds as shown in Figure 5.11, this
does not mean that the operations at the airport are not affected. In fact, although over the course of
one hour the number of movements achieved by the runway slightly changes, the time intervals between
operations change significantly, which produces the increase in delays shown in Figure 5.12.
Figure 5.11: Runway 03 throughput for no wind and 30 knots headwind conditions.
56
Figure 5.12: Runway 03 average arrival and departure delay for no wind and 30 knots headwind conditions.
5.3.2 Future Demand
For the future demand, the achieved runway throughput is represented in Figure 5.13. The analysis
focus is the morning demand peak, as this is where the main differences between the current and future
demand are found. In this case, the demand is higher than the maximum runway throughput, repre-
sented in Figure 5.6. This naturally represents an inconsistency with reality, where given the current
runway maximum throughput, no such demand is permitted to operate in the airport. However, in simu-
lation, it is possible to simulate the airport response to such a high demand. Even though the demand
highest point is bigger that the runway maximum throughput, note that in Figure 5.13 the maximum
runway throughput is not achieved at any part of the day, not even in the morning peak (the maximum
throughput achieved with the future demand is slightly below 43 movements, more than one bellow
the 44 movements of the maximum runway throughput in Figure 5.6). This is an expected result as in
order to achieve the maximum runway throughput, the runway needs to be in constant use and con-
stantly being requested by both departures and arrivals, so that a sequence Arrival-Departure-Arrival is
always maintained, which not even with the morning peak demand can be achieved (note that in order
to calculate the maximum runway throughput in Figure 5.6, almost the double of the current demand
represented in Figure 5.2 is simulated).
In the morning demand peak in Figure 5.13, it is observable that neither the arrivals or the departures
throughput accompanies the respective demand, by a difference of circa five movements at its highest
peak. This is an important factor to take into consideration and strengthens the fact that this high demand
is unsustainable by the airport current state.
57
Figure 5.13: Runway 03 throughput vs future demand.
Figure 5.14: Runway 03 average arrival and departure delay for the future demand.
The average arrival and departure delay for the future demand can be seen in Figure 5.14. The
breakdown of the origins of each delay for the future demand is extremely similar to the one for the
current demand shown in Figures 5.9 and 5.10, where taxi delays are minimal and the vast majority of
arrivals and departures delay occur due to runway constraints, and is therefore not shown. The biggest
differences relatively to the current demand average delays occur naturally at the morning demand peak,
where departure average delay peaks to eight minutes and arrival average delay peaks to two minutes.
This further proves the inviability of the airport sustaining the simulated future demand, as the amount
of average total delay is significantly higher.
For the case of future demand a wind analysis is not performed as even with no wind the airport is
already incapable of acceptably dealing with such a high demand. As a remark, the tendency should
be for runway throughput to lower as headwind increases, in coherence with the maximum runway
throughputs found in Figure 5.6. As a result, longer waiting times can be expected at the departure
queues and aircraft will need to hold longer in the air in order to respect separations.
58
5.4 Runway 21
In the same way the runway 03 maximum throughput calculation is done and described in Section 5.3,
the runway 21 maximum throughput for the different wind conditions is shown in Figure 5.15. The
maximum runway throughput achieved for no wind conditions is slightly below 42 movements pour (note
that as the constant value presented is the average of a total of 12 tested hours, each being the average
of 10 iterations, the constant values achieved for the runway throughput do not have to be integers).
The obtained value is again superior to the achieved in reality at the airport, which is 40 movements
per hour [24]. The difference between maximum throughput achieved in simulation and in reality is
slightly bigger for runway 21, which comes as no surprise as when this runway is in use, air traffic
controllers have an extra amount of work due to all the runway crossings performed together with line-
ups, and it is therefore normal to introduce a bigger buffer in the minimum separations. In simulation,
the difference between the two achieved maximum runway throughputs, for runway 03 and 21, comes
from the increase in separation implemented between arrivals in an Arrival-Departure-Arrival sequence
when runway 21 is in use. Note that in order to calculate the maximum runway throughput, the optimal
sequence is Arrival-Departure-Arrival. The wind impact on the runway maximum throughput is strong
and noticeable, specially for headings with 30 knots speed coming from 0◦ and 30◦, where the maximum
runway throughput drops below the 35 and 36 movements per hour, respectively.
Figure 5.15: Runway 21 maximum throughput for different wind conditions.
59
5.4.1 Current Demand
Runway 21 throughput when tested with the current demand can be seen in Figure 5.16. The achieved
results are similar to the ones achieved by runway 03, an expected result as the maximum demand,
during the morning peak, is still significantly below the maximum throughput of the runway seen in Figure
5.15. On the contrary, the average arrivals and departures total delay, which can be seen in Figure 5.17,
are significantly bigger than the ones found in runway 03, Figure 5.8. The biggest difference is found
in the late afternoon departure peak, where the average delay for runway 21 (Figure 5.17) is circa four
minutes higher than the one for runway 03 (Figure 5.8).
Figure 5.16: Runway 21 throughput vs current demand.
Figure 5.17: Runway 21 average arrival and departure delay for the current demand.
60
Figure 5.18: Runway 21 average arrival delay breakdown.
Figure 5.19: Runway 21 average departure delay breakdown.
Figures 5.18 and 5.19 breakdown the average arrival and departure total delays respectively, seen
in Figure 5.17. It is clear that the origins of the delay are still overwhelmingly located on the runway
constraints for both arrivals and departures. An interesting result is the runway 21 slight increase in
departures average taxi delay when compared to runway 03 (Figure 5.10). Figure 5.20 shows where the
delays occurred in the airfield, according to the size and color of the circles. The circles in orange rep-
resent departure queue delays and the one in red represents taxi delays (note that delays at departure
queues, next to the runway, are changed to the aircraft respective origin gate). It is clear that the slight
increase in taxi delay is due to runway crossings, where aircraft need to wait for the runway to be clear
in order to cross it, which is where the red circle is located.
61
Figure 5.20: Runway 21 average departure delay breakdown.
The runway maximum throughput with wind conditions of 30 knots, at 0◦,is lower than the achieved
runway throughput for the current demand in no wind conditions. The wind analysis in this case is
done for headwinds at 0◦, for speeds of 15 knots and 30 knots, taking into account the maximum runway
throughputs distribution found in Figure 5.15. The runway throughput for the two different wind conditions
and for no wind is show in Figure 5.21. Note how for the 30 knots headwind, with which the runway
has a maximum throughput of circa 34 movements, throughput does not accompany the other two wind
conditions throughput specially at the time of the day where the demand goes above 34 movements. The
average arrival and departure total delay for the different wind conditions can be seen in Figure 5.22. For
the 30 knots headwind, with which the maximum runway throughput is lower than the achieved runway
throughput with the current demand for no wind conditions, the average delay for arrivals is constantly
extremely higher than the other two conditions, reaching almost seven minutes. It should be noted that
retaining aircraft with an average of seven minutes in order to comply with separation requirements is not
feasible. However, a 30 knots headwind scenario is not common at the Lisbon Airport, and the results
are obtained with a full day of 30 knots headwind. The average departure delay is also significantly
higher when compared to the other two wind conditions, specially in the morning demand peak.
62
Figure 5.21: Runway 21 throughput for no wind, 15 knots and 30 knots headwind conditions.
(a) Arrivals total delay. (b) Departures total delay.
Figure 5.22: Runway 21 average arrivals and departures delay for different wind conditions.
5.4.2 Future Demand
The runway 21 achieved throughput when tested with the future demand is represented in Figure 5.23.
Again, the maximum throughput of the runway is not achieved, which is considered to be normal as
explained in Section 5.3.2. The performance of runway 21 when tested with the future demand is
lower than runway 03, as expected, due to the increase in separation for the Arrival-Departure-Arrival
sequence. In particular, the departures difference between demand and throughput is considerably high
during the morning demand peak, reaching between six and seven movements. It is clear that the
current state of the airport is not able to sustain such a high demand.
63
Figure 5.23: Runway 21 throughput vs future demand.
Figure 5.24: Runway 21 average arrival and departure delay for the future demand.
When looking at the average arrival and departure total delay in Figure 5.24, it is immediately no-
ticeable the extremely high average departure delay peak, which surpasses the 20 minutes. Note that
this delay peak happens after the morning demand peak. This is expectable, as the airplanes that
accumulate at the departure queues due to the morning demand peak and have high delays, actually
depart after the morning peak, bringing the average delay at that time up. The average departure delay
is brought down only because the demand comes below the maximum runway throughput, allowing the
airport to relieve the departure queues and consequently bring the average departure delay down. This
further strengthens the fact that such a high demand is not feasible with the current state of the airport
and can not be accommodated.
64
5.5 New Airport Model
Given the results obtained in Sections 5.3.2 and 5.4.2, it is clear that the current layout of the air-
port is not able to accommodate the medium term future demand defined for the airport. Therefore,
the need arises to make changes to the current layout and procedures, in order to improve its capac-
ity. As identified in Sections 5.3 and 5.4, the main constraint to the airports capacity is currently the
runway for both runway 03 and runway 21, where the runway throughput is incapable of following the
high demand. In order to improve runway throughput, taking into account the inviability of adding new
runways to the airport, the target optimization should be ROT, so that the Arrival-Arrival separation in
an Arrival-Departure-Arrival sequence can be lowered (note that a constant Arrival-Departure-Arrival
sequence is the one with which maximum runway throughput is achieved, and that Arrival-Arrival se-
quence separation is currently imposed due to wake vortex minimum separation, not ROT, due to which
it is not viable to consider lowering Arrival-Arrival sequence minimum separation). Besides this, the re-
organization of the Lisbon Airspace expects to allow aircraft to diverge by more than 45◦ after takeoff,
allowing for Departure-Departure minimum separations, explained in Section 4.1.2.F, to be reduced by
one minute (to a minimum of one minute) [11]. Note that this is an optimistic approximation. In real-
ity, some Departure-Departure sequences will not diverge by more than 45◦ after takeoff, and therefore
the separation can’t be reduced. However, as a single departure route is simulated, it is optimistically
considered that all Departure-Departure separations can be reduced by one minute.
Figure 5.25: Airfield layout modifications.
65
The proposed changes to the current airport layout in order to increase its capacity can be seen
in Figure 5.25. The extra taxiways added to the current layout are highlighted in red. Note that the
main target optimizations are done for ROTA and minimization of runway crossings, as ROTD is less
dependent on airport layout and more dependent on flight crew reaction times and aircraft type and
weight (bringing awareness to flight crews in order to optimize FRLC and FRTT is a task already carried
out by NAV Portugal throughout the years, for which the values extracted from the provided traffic sample
are considered to be maintained [24]).
5.5.1 Runway 03
The runway 03 optimizations are focused on ROTA. The current existing runway high speed exit for
runway 03, taxiway HN, has a sharp turn immediately after the runway exit, which reduces average
ROTA. In order to avoid this, an extension to taxiway HN is added, with an accompanying second
taxiway parallel to the runway so that aircraft can turn to the gates area. Besides this, a second runway
high speed exit, named HO, is added before the current one, in order to target the aircraft that currently
exit with a sharp turn to runway 17/35 or taxiway S1. This runway high speed exit ends in the same
second parallel taxiway as the extension of taxiway HN (note that a sharp turn is still done at the end of
both runway high speed exits, but now further away from the runway, giving the aircraft more distance to
reduce speed). The link speed given to both the high speed exits is 40 knots, and the link speeds to the
other added taxiways is given according to the rules specified in Section 4.1.3.
The new runway exits usage percentage for each airspace group can be seen in Table 5.1. As
an assumption, the new layout allows for all aircraft to exit in one of the two high speed exits. The
usage percentages are given in coherence with the ones for the current layout, represented in Table 4.7,
considering the modifications done. Regarding ROTA, the criterion used is to reduced the average ROTA
for runway high speed exit HN by five seconds, due to the added extension, and reduce an extra five
seconds in order to obtain the average ROTA for runway high speed exit HO, due to its location, closer
to runway 03 threshold than runway exit HN. This reduction in average ROTA allows the Arrival-Arrival
separation in an Arrival-Departure-Arrival sequence to be reduced to 5.5 nautical miles (note that the
probability distribution in Figure 4.4 relative to separation increments still applies).
Airspace Group
Runway ExitHO HN
Heavy 25% 75%
Medium-Jet 60% 40%
Medium-Prop 90% 10%
Light 100% 0%
Table 5.1: New runway 03 exits and usage percentage for airspace groups.
66
Figure 5.26 shows the runway maximum throughput, calculated with the same method described in
Section 5.3. The maximum throughput achieved under no wind conditions with the new airport layout
is slightly superior to 50 movements per hour. This represents a six movements improvement relative
to the current layout, even though Arrival-Arrival minimum separation in an Arrival-Departure-Arrival
sequence is only reduced by 0.5 nautical miles. This further demonstrate the need of optimizing every
single process related to ROT, as the smallest reduction possible to make in minimum separations has
a big impact on the runway maximum throughput. Note that the influence of the wind is now stronger on
the maximum throughput, as the minimum separation for Arrival-Arrival in an Arrival-Departure-Arrival
sequence has reduced. With the six nautical miles separation from the current layout, the difference
between no wind conditions and worst wind conditions is close to eight movements per hour (Figure
5.6). With the new separation, this difference increases to more than 10 movements per hour between
no wind conditions and worst wind conditions.
Figure 5.26: New runway 03 maximum throughput for different wind conditions.
As the reason to make changes to the airport current layout is the incapability of it dealing with the
medium term future demand, the analysis to the new layout is done only for the medium term future
demand. The runway throughput and demand comparison can be seen in Figure 5.27. Similar to the
current layout runway 03 throughput when tested with the current demand (Figure 5.7), the new layout
runway 03 does not accompany the demand when it peaks, even thought its maximum throughput is
higher than the highest demand point. This is due to the impossibility of perfectly sequencing movements
in an Arrival-Departure-Arrival sequence, given that the departure demand and the arrival demand are
not constantly equal throughout the day. When compared with the current runway 03 response to the
future demand (Figure 5.13), the new runway 03 performs significantly better, specially in reaction to the
morning demand peak where throughput is able to be kept closer to the required demand.
67
Figure 5.27: New runway 03 throughput vs future demand.
Figure 5.28: New runway 03 average arrival and departure delay for the future demand.
The evidence of the new layout better performance is further amplified in Figure 5.28, where the
average arrival and departure total delay throughout the day is shown. Here, the new runway 03 average
delay for departures peaks at slightly above three minutes, which is almost five minutes lower than the
average departure delay for the current layout runway 03 (Figure 5.14). The average arrival total delay
is also lower although not to the same extent, with the difference to the current layout being almost a
minute in the morning demand peak.
A wind analysis is done for 15 knots and 30 knots full headwind, considering the distribution of
maximum runway throughputs in Figure 5.26. The runway throughput under no wind and 15 knots
headwind conditions is practically the same, while with 30 knots headwind a decrease in performance
occurs, as shown in Figure 5.29. Note how the runway throughput for 30 knots headwind flats at its
maximum value (Figure 5.26), which causes the lower performance of the runway.
68
Figure 5.29: New runway 03 throughput for no wind, 15 knots and 30 knots headwind conditions.
Figure 5.30: New runway 03 average total delay for no wind, 15 knots and 30 knots headwind conditions.
The average total delay for the three different wind conditions can be seen in Figure 5.30. Although
there is a slight increase between no wind and 15 knots headwind conditions, the major difference
occurs for the 30 knots headwind, where the average total delay peaks at slightly above six minutes. This
comes in consequence of the maximum runway throughput being reached, which significantly increases
average waiting times at the departures queues. However, as a side note, the new layout runway 03
performance under the worst tested wind conditions, 30 knots headwinds, is still better that the current
layout runway 03 performance for the future demand under no wind conditions.
69
5.5.2 Runway 21
In order to increase runway 21 performance, for it to be able to accommodate the medium term future
demand, not only ROT, but also runway crossings are target of optimization. Regarding runway cross-
ings, an ideal scenario is to eliminate them completely. This is what the new runway 21 entry point,
seen in Figure 5.25, tries to achieve. In the simulation, it is considered that all aircraft that previously
selected taxiway S4 as their runway entry point are now changed to the new runway entry point, named
X1. This is an optimistic assumption, as despite the fact that runway entry point X1 gives significantly
more distance to takeoff when compared to runway entry point U5, it gives equally less distance when
compared to runway entry point S4. The second taxiway parallel to the runway, that in runway 03 allows
aircraft that vacate the runway through taxiway HN to make the sharp turn later, is in this instance also
useful, as even a small departure queue in runway entry point U5 blocks the taxiway giving access to
the entry point X1, which can cause big delays. With a second parallel taxiway, aircraft intending to use
runway entry point X1 have a way of reaching it even with big queues in departure queue U5. In order
to decrease ROTA, a new runway high speed exit for runway 21 is designed and named HT, which is
located closer to runway 21 threshold. The runway exits usage percentage can be seen in Table 5.2.
Optimistically, it is considered that the low percentages of aircraft previously exiting in runway exits P
and N2 , almost at the end of the runway, are now able to exit in runway high speed exit HS. The ROTA
are kept constant for the already existing runway exits HS and runway 17/35, as there are no layout
changes that impact them. Runway exit HT ROTA is obtained by subtracting five seconds to runway exit
HS ROTA, due to its location, closer to runway 21 threshold. The changes in the airport layout allow
for runway 21 Arrival-Arrival separation in an Arrival-Departure-Arrival sequence to be decreased from
6.25 nautical miles to 5.5 nautical miles, the same as the new separation for runway 03. Note that by
eliminating runway crossings, the two runways are able to operate in a very similar manner, with the
differences in ROT being minimal and mostly due to runway exit locations.
Airspace Group
Runway Exit17/35 HT HS
Heavy 0% 30% 70%
Medium-Jet 0% 60% 40%
Medium-Prop 60% 40% 0%
Light 90% 10% 0%
Table 5.2: New runway 21 exits and usage percentage for airspace groups.
70
Since all the minimum separations of the new runway 21 are equal to the minimum separations of
runway 03, in any procedure sequence, the new runway 21 maximum throughput for all the tested wind
conditions is the same as the maximum throughputs shown for runway 03, in Figure 5.26. This comes
due to the fact that in simulation, when testing the runway maximum throughput as described in Section
5.3, the only influencing factor is Arrival-Arrival separation in an Arrival-Departure-Arrival sequence, and
on the traffic mix (more specifically, the approach speed of each arriving aircraft).
By comparing Figure 5.26, which represents both the new runway 03 and the new runway 21 max-
imum throughput, to Figure 5.15, representing the current runway 21 throughput, it is observable that
runway 21 maximum throughput under no wind conditions has increased by slightly more than eight
movements per hour. This is a bigger increase in performance than the one achieved for runway 03,
which is expected as not only ROTA but also runway crossings are optimized in the new runway 21,
allowing for the minimum Arrival-Arrival separation in an Arrival-Departure-Arrival sequence to be de-
creased for the new runway 21 more than for the new runway 03.
The performance of the new runway 21 when tested with the future demand can be seen in Figure
5.31. Although similar to the new runway 03, the performance is not exactly the same, expected due
to the differences in ground movement between both runways as well as all the random variables intro-
duced in each model. When compared to the current runway 21 response to the future demand (Figure
5.23), the performance of the new runway 21 is significantly better, specially in the departures where it
is able to better accompany the demand, even in the morning peak.
Figure 5.31: New runway 21 throughput vs future demand.
The average arrival and departure delay for the new runway 21 can be seen in Figure 5.32. With the
current runway 21, the departure average delay greatly increases in the morning demand peak, to an
average of more than 20 minutes. With the new runway 21, the average departure delay is kept below
three minutes, representing more than a 15 minutes decrease. This demonstrates that the new runway
21 is able to accommodate the medium term future demand, with the average arrival delay being kept
below 1.5 minutes, and the average departure delay peaking at less than three minutes.
71
Figure 5.32: New runway 21 average arrival and departure delay for the future demand.
The wind analysis is done for the new runway 21 in the same manner as for the new runway 03, as
their maximum throughputs, represented in Figure 5.26, are the same. The new runway 21 through-
put under different wind conditions is represented in Figure 5.33. Very similar to runway 03, only 30
knots headwind has a noticeable impact on the runway throughput, as the runway reaches its maximum
throughput under 30 knots headwind conditions during the morning demand peak. The average total
delay for the three different wind conditions is represented in Figure 5.34. The values are extremely
similar to the new runway 03, and also being extremely more affected by the 30 knots headwind due to
the maximum runway throughput being reached.
Figure 5.33: New runway 21 throughput for no wind, 15 knots and 30 knots headwind conditions.
72
Figure 5.34: New runway 21 average total delay for no wind, 15 knots and 30 knots headwind conditions.
73
74
6Conclusion
75
76
In this work, the current state of the Lisbon Airport is analyzed in detail through the use of computer
simulation. Simmod PLUS! is used as the simulation tool, and a full model of the airport airfield and
surrounding airspace is made. Two different traffic samples are tested in the model. One from the
12th of April, 2017, representing a current average day demand, and the other representing a medium
term future demand, obtained by adding a certain amount of flights to the current demand, in order to
represent the predicted increase in traffic at the Lisbon Airport.
The current state of the Lisbon Airport is not able to cope with the defined medium term future
demand. The runway throughput for both runways in use is not able to accompany the requested
demand, specially for runway 21 where the Arrival-Arrival separation in an Arrival-Departure-Arrival
sequence is 6.25 nautical miles, 0.25 nautical miles higher than for runway 03. The average departure
delays peak at approximately eight and 22 minutes for runway 03 and 21, respectively. Given this results,
enhancements to the current airport layout and operational procedures are identified and modeled. With
the new layout, Arrival-Arrival separation in an Arrival-Departure-Arrival sequence is reduced to 5.5
nautical miles for both runways, increasing the runway maximum throughput to a value slightly superior
to 50 movements, which represents an improvement of approximately six movements for runway 03 and
eight movements for runway 21. The average departure delay is greatly reduced for both runways, to
values surrounding three minutes, which strengthens the better performance of the new airport model.
The results of this work should be analyzed taking into account the things that are not modeled or not
considered by the simulator, such as specific gate attribution to each flight and air traffic controllers work-
load both in the approach and in the tower sectors. Besides that, the layout enhancements proposed
represent a simple practical way of reducing some current constraints to the airport capacity. They do
not take into account all the needed factors in order to actually perform changes to the airport layout,
such as the physical characteristics of the terrain or the economical factors.
For further work, it is crucial to further deepen the detail of the model in order to take more influencing
factors into consideration. The modeling of the surrounding airspace should be further increased, in
order to take into account all the different approach and departure routes and simulate in detail the
movement and sequencing of arriving and departing aircraft. The ground movement of aircraft between
gates as well as specific gate and slot allocation can as well be further modeled.
In conclusion, the current state of the Lisbon Airport is already working at its maximum capacity and
can not cope with the medium term future demand. The proposed changes to the airport layout and
operational procedures bring improvements to the airport capacity and comfortably accommodate the
medium term future demand. However, several factors are not considered by the simulation and this
should be taken into account when analyzing the results.
77
78
Bibliography
[1] “Resolucao do conselho de ministros,” Diario da Republica, no. 5-B/2016, pp. 468–(2), Feb. 2016.
[2] “Aeroporto da portela,” in Artigos de apoio Infopedia, Feb. 2003-2017, accessed: 27-03-2017,
https://www.infopedia.pt/$aeroporto-da-portela.
[3] Revista Militar, vol. N. 2425/2426, pp. 0–0, Feb./Mar. 2004.
[4] M. R. Partidario and M. Coutinho, “Historia de um processo de decisao: o novo aeroporto de
lisboa,” in 3a Conferencia Nacional de Avaliacao de Impactes, Oct. 2008.
[5] ANA Aeroportos de Portugal, “2016 best growth ever for Portuguese airports,” Ana
Press, Feb. 2017, accessed: 27-03-2017, www.ana.pt/en/corporate/press/2017/02/15/
2016-best-growth-ever-for-portuguese-airports.
[6] M. C. Freire, “Montijo e a solucao para aeroporto complementar de lisboa,”
Diario de Notıcias, Jan. 2017, accessed 27-03-2017, www.dn.pt/portugal/interior/
montijo-e-a-solucao-para-aeroporto-complementar-de-lisboa-5615523.html.
[7] “LPPT AD2,” in eAIP Portugal. NAV Portugal E.P.E., Feb. 2016.
[8] “Lisbon airport,” ANA Aeroportos de Portugal, accessed 28-03-2017, https://www.aeroportolisboa.
pt/en/lis/flights-destinations/airlines/airlines-and-destinations.
[9] “Aerodromo de Transito No 1,” Forca Aerea Portuguesa, accessed 28-03-2017, http://www.emfa.pt/
www/unidade-21-aerodromo-de-transito-n-1.
[10] “Esquadra 504 - ”linces”,” Forca Aerea Portuguesa, accessed 28-03-2017, http://www.emfa.pt/
www/esquadra-41.
[11] Lisboa ACC/TMA Interface Study - Phase III and Phase IV Report, 1st ed., EUROCONTROL, Dec.
2016.
[12] J. A. Rodrigues, “Lisbon Airport Capacity Enhancement, Airspace capacity estimation and en-
hancement,” Master’s thesis, Instituto Superio Tecnico, Nov. 2014.
[13] DOC 9883 Manual on Global Performance of the Air Navigation System, 1st ed., ICAO, 2009.
[14] S. O’Flynn, Airport Capacity Assessment Methodology, 1st ed., EUROCONTROL, Oct. 2016.
79
[15] Airtopsoft, “AirTOp,” accessed 04-05-2017, http://airtopsoft.com.
[16] A. R. C. GmbH, “CAST,” accessed 04-05-2017, http://www.airport-consultants.com/cast-simulation.
[17] EUROCONTROL, “Performance Indicator and Analysis Tool for Airports (PIATA),” accessed 03-07-
2017, http://www.eurocontrol.int/services/performance-indicator-and-analysis-tool-airports-piata.
[18] I. Software, “RAMS Plus,” accessed 04-05-2017, http://ramsplus.com.
[19] ATAC, “Simmod PLUS!” accessed 04-05-2017, http://www.atac.com/ITL/simmod-plus.html.
[20] Simmod PLUS! Reference Manual, ATAC, 2770 De La Cruz Blvd, Santa Clara, CA 95050-2624,
USA, Feb. 2015.
[21] N. L. Johnson, S. Kotz, and N. Balakrishnan, Continuous Univariate Distributions, 2nd ed., 1995,
vol. 2, ch. 21:Beta Distributions.
[22] “Flighradar24,” accessed 05-05-2017, https://www.flightradar24.com/.
[23] DOC 4444 Procedures for Air Navigation Services, Air Traffic Management, 16th ed., ICAO, 2016.
[24] “Lisbon capacity enhancement exercise 2010-2011,” NAV Portugal E.P.E., Tech. Rep. 1, July 2011.
[25] C. J. Cashman, “Crosswind guidelines,” Boeing, Tech. Rep.
80
ALPPT Charts
Contents
A.1 LPPT Aerodrome Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
A.2 LPPT Ground Movement Chart RWY03 . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
A.3 LPPT Ground Movement Chart RWY21 . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
81
82
AIP
PO
RTU
GA
LLP
PT A
D 2
.24.
01 -
1
© N
AV P
ortu
gal,
E.P.
E.
30-M
AR
-201
7
AIR
AC
002
-17
30-M
AR
-201
7
New layout Taxiways M and N
A.1 LPPT Aerodrome Chart
83
AIP
PO
RTU
GA
LLP
PT A
D 2
.24.
03 -
1
© N
AV P
ortu
gal,
E.P.
E.
30-M
AR
-201
7
AIR
AC
002
-17
30-M
AR
-201
7
New layout Taxiways M and N
A.2 LPPT Ground Movement Chart RWY03
84
AIP
PO
RTU
GA
LLP
PT A
D 2
.24.
03 -
3
© N
AV P
ortu
gal,
E.P.
E.
30-M
AR
-201
7
AIR
AC
002
-17
30-M
AR
-201
7
New layout Taxiways M and N
A.3 LPPT Ground Movement Chart RWY21
85
86
BFlights Timetable
Contents
B.1 Current Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
B.2 Future Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
87
88
B.1 Current Demand
Flight Aircraft Airline Time A or DLX2089 A320 SWISS 02:45 DKL1692 B738 KLM 05:00 DTP1528 A320 TAP 05:10 ATP224 A332 TAP 05:25 ATP82 A343 TAP 05:30 ATP74 A332 TAP 05:35 A
AF1125 A321 Air France 05:40 DTP218 A332 TAP 05:50 A
LH1497 A320 Lufthansa 05:55 DTP202 A332 TAP 05:55 ATP172 A332 TAP 05:55 A
LH1793 A321 Lufthansa 06:05 DTP46 A332 TAP 06:05 A
TP104 A332 TAP 06:05 AU27651 A319 easyJet 06:15 DFR1083 B738 Ryanair 06:25 DFR2941 B738 Ryanair 06:30 DFR1885 B738 Ryanair 06:30 DS4121 A320 SATA 06:30 D
TP1943 AT76 TAP 06:30 ATP436 A320 TAP 06:35 DS66646 B762 StarAir 06:35 AU28716 A320 easyJet 06:40 DTP578 A319 TAP 06:45 D
TP1044 A320 TAP 06:45 DTP1900 A320 TAP 06:50 ATP834 A320 TAP 06:50 D
TP1480 A321 TAP 06:50 ADT650 B77W TAAG 06:50 AIB3107 A319 Iberia 06:55 DTP1532 A319 TAP 06:55 ATP1012 E190 TAP 06:55 DU27611 A319 easyJet 07:00 DTP1926 AT76 TAP 07:00 DQY8165 B752 DHL 07:00 ATP336 A319 TAP 07:05 DTP754 A321 TAP 07:05 DTP802 A321 TAP 07:05 D
UX1159 E195 Air Europa 07:10 AU27603 A319 easyJet 07:10 DLH1173 A321 Lufthansa 07:15 DBA499 A320 British Airways 07:25 DTP473 E190 TAP 07:25 A
U21442 A319 easyJet 07:30 D
89
TP616 A319 TAP 07:30 DTP864 A320 TAP 07:30 D
TP1925 AT76 TAP 07:30 AFR6066 B738 Ryanair 07:35 DTP1049 A319 TAP 07:35 ATP1670 A320 TAP 07:35 AFR2093 B738 Ryanair 07:40 ATP1827 A319 TAP 07:40 DTP374 A332 TAP 07:40 D
TP1091 AT76 TAP 07:40 ATP481 E190 TAP 07:40 A
WT8023 AT43 Swiftair 07:43 ATP1023 A319 TAP 07:45 ATP1679 A319 TAP 07:45 D
TAY646C B734 ASL AirlinesBelgium
07:45 A
UX1152 E195 Air Europa 07:50 DU21443 A320 easyJet 07:55 ATP447 A319 TAP 07:55 A
TP1272 A320 TAP 08:00 DTP434 A321 TAP 08:00 D
TP1082 AT76 TAP 08:00 DTP1134 AT76 TAP 08:00 DTP1924 AT76 TAP 08:00 DFR2094 B738 Ryanair 08:05 DTP929 A320 TAP 08:05 ATP611 A320 TAP 08:05 A
TP1065 AT76 TAP 08:05 ATP404 E190 TAP 08:10 D
FR2087 B738 Ryanair 08:15 AVY8460 A321 Vueling 08:15 ATP669 A320 TAP 08:15 ATP837 A319 TAP 08:20 ATP811 A320 TAP 08:20 A
TP1096 AT76 TAP 08:20 DU21444 A320 easyJet 08:25 DTP567 A319 TAP 08:25 ATP574 A319 TAP 08:25 DTP571 A320 TAP 08:25 ATP488 E190 TAP 08:25 D
TP1933 E190 TAP 08:25 ATP952 A319 TAP 08:30 DTP361 A320 TAP 08:35 ATP694 E190 TAP 08:35 DTP531 A320 TAP 08:40 A
TP1454 A320 TAP 08:40 DTP542 E190 TAP 08:40 D
90
FR2086 B738 Ryanair 08:45 DTP1672 A319 TAP 08:45 ATP753 A319 TAP 08:45 ATP784 A320 TAP 08:45 D
TP1439 AT76 TAP 08:45 ATP558 A319 TAP 08:50 D
TO3412 B738 Transavia France 08:55 AVY8461 A321 Vueling 09:00 DTP664 A319 TAP 09:00 DTP832 A319 TAP 09:00 D
TP1930 AT76 TAP 09:00 DTP1034 E190 TAP 09:00 DPV6324 C25B Private 09:00 ATP806 A320 TAP 09:05:00 D
HV9145 B737 Transavia 09:10 AIB3108 A320 Iberia 09:10 ATP538 A319 TAP 09:10 D
FR1786 B738 Ryanair 09:15 ATP443 A319 TAP 09:15:00 AAT982 B738 Royal Air Maroc 09:20 ATP932 A320 TAP 09:20 DZI801 A320 Aigle Azur 09:25 A
TP1020 A319 TAP 09:25 DTO3413 B738 Transavia France 09:30 DU22365 A320 easyJet 09:30 ATP1905 A319 TAP 09:30 DTP1262 A320 TAP 09:30 DTP1671 A320 TAP 09:30 DTP1533 A320 TAP 09:30 DTP1483 A320 TAP 09:30 DTP1941 AT76 TAP 09:30 APV2785 LJ35 Private 09:30 AFR7328 B738 Ryanair 09:35 A
TP59 A332 TAP 09:35 DFR1787 B738 Ryanair 09:40 DHV5951 B738 Transavia 09:40 ATP103 A332 TAP 09:40 DTP117 A332 TAP 09:45 D
HV9146 B737 Transavia 09:50 DIB3111 A320 Iberia 09:50 DFR2098 B738 Ryanair 09:55 ATP440 A319 TAP 09:55 D
FR7319 B738 Ryanair 10:00 DU22366 A320 easyJet 10:00 DTP1944 AT76 TAP 10:00 DTP336 A319 TAP 10:05 D
TP1011 E190 TAP 10:10 A
91
BA500 A320 British Airways 10:15 AU27612 A319 easyJet 10:15 AZI802 A320 Aigle Azur 10:15 DTP223 A332 TAP 10:15 DTP36 A332 TAP 10:15 AAT983 B738 Royal Air Maroc 10:20 D
FR2097 B738 Ryanair 10:20 DHV6204 B738 Transavia 10:20 DFR2622 B738 Ryanair 10:30 ATK1755 B739 Turkish Airlines 10:30 ATP1927 AT76 TAP 10:30 AWI5773 A320 White 10:30 DEI482 A320 Aer Lingus 10:35 A
FR2925 B738 Ryanair 10:35 AU27641 A319 easyJet 10:45 DUX1153 E195 Air Europa 10:45 ATP371 A320 TAP 10:45 ATP217 A332 TAP 10:45 D
FR2623 B738 Ryanair 10:55 DTP1867 A319 TAP 10:55 DTP1038 E190 TAP 10:55 DDT651 B77W TAAG 11:00 D
FR2926 B738 Ryanair 11:00 DTP1934 AT76 TAP 11:00 DBA501 A320 British Airways 11:10 DLH1166 A321 Lufthansa 11:10 AU27604 A319 easyJet 11:10 AS4320 A310 SATA 11:10 A
TP1051 A320 TAP 11:10 AEI483 A320 Aer Lingus 11:15 DTP89 A343 TAP 11:15 D
PV6323 C25B Private 11:15 DAF1024 A319 Air France 11:20 ATP1680 A319 TAP 11:20 ATP208 A332 TAP 11:20 ATP12 A343 TAP 11:20 A
UX1150 E195 Air Europa 11:25 DTP26 A332 TAP 11:25 A
TK1756 B739 Turkish Airlines 11:30 DTP88 A343 TAP 11:30 A
TP1931 AT76 TAP 11:30 AU27637 A319 easyJet 11:40 DTP1085 AT76 TAP 11:40 ATP1135 AT76 TAP 11:40 ATP1902 A319 TAP 11:45 ASN3815 A320 Brussels Airlines 11:50 ATP1674 A319 TAP 11:50 A
92
IB3110 A320 Iberia 11:55 ATP1936 AT76 TAP 12:00 D4U602 A319 Germanwings 12:05 A
TP1025 A319 Air France 12:10 DFR1084 B738 Ryanair 12:10 ATP1167 A321 Lufthansa 12:10 DTP1681 A319 TAP 12:10 DTP437 A320 TAP 12:15 A
PV2785 LJ35 Private 12:15 DEW2604 A320 Eurowings 12:20 ATP1304 A320 TAP 12:20 DTP1028 A320 TAP 12:20 DPV8148 GLF5 Private 12:20 AA3668 A321 Aegean Airlines 12:25 A
TP1102 AT76 TAP 12:25 DU27083 A319 easyJet 12:30 AU23759 A320 easyJet 12:30 ATP836 A319 TAP 12:30 DTP442 A319 TAP 12:30 D
TP1935 AT76 TAP 12:30 AEK191 B77W Emirates 12:35 AFR1671 B738 Ryanair 12:35 DFR1884 B738 Ryanair 12:35 AIB3109 A320 Iberia 12:35 DSN3816 A320 Brussels Airlines 12:40 DU28717 A320 easyJet 12:40 ATP1013 A319 TAP 12:40 A4U603 A319 Germanwings 12:45 D
LG3751 B737 Luxair 12:45 ATP1453 A320 TAP 12:45 ATP201 A332 TAP 12:45 D
TP1140 AT76 TAP 12:45 DS4321 A310 SATA 12:50 DTP342 A320 TAP 12:55 D
EW2605 A320 Eurowings 13:00 DU27084 A319 easyJet 13:00 DU23760 A320 easyJet 13:00 DVY8412 A320 Vueling 13:00 ATP1938 AT76 TAP 13:00 DU21445 A319 easyJet 13:05 AFR6067 B738 Ryanair 13:10 AU22715 A320 easyJet 13:10 ATP764 A320 TAP 13:10 D
TP1099 AT76 TAP 13:10 AA3669 A321 Aegean Airlines 13:15 D
TP1828 A319 TAP 13:15 ATP339 A319 TAP 13:15 A
93
TP1235 A320 TAP 13:15 ATP803 A321 TAP 13:15 A
TP1042 A319 TAP 13:20 DTP1035 E190 TAP 13:25 ALX3752 B737 Luxair 13:30 DTP874 A320 TAP 13:30 D
TP1937 AT76 TAP 13:30 ATP403 E190 TAP 13:30 A
LH1790 A321 Lufthansa 13:35 ATO3414 B738 Transavia France 13:35 AVY8413 A320 Vueling 13:35 DTP1682 A320 TAP 13:35 ATP831 A320 TAP 13:35 A
U22716 A320 easyJet 13:40 DTP617 A319 TAP 13:40 ATP433 A321 TAP 13:40 A
FR2942 B738 Ryanair 13:45 ALX2084 A321 SWISS 13:50 ATP581 A319 TAP 13:50 A
TP1434 AT76 TAP 13:50 DTP363 A332 TAP 13:55 AYU606 B763 euro Atlantic
Airways14:00 A
VR606 B763 TACV 14:00 ATP1946 AT76 TAP 14:00 DTP483 E190 TAP 14:00 AWI5772 A320 White 14:00 ATP662 A319 TAP 14:05 D
FR2077 B738 Ryanair 14:10 DTO3415 B738 Transavia France 14:10 DTP961 A319 TAP 14:10 AEK192 B77W Emirates 14:15 DU27652 A319 easyJet 14:20 ATP867 A320 TAP 14:20 ATP926 A332 TAP 14:20 DTP324 E190 TAP 14:20 D
LH1791 A321 Lufthansa 14:25 DTP472 E190 TAP 14:25 DLX2085 A321 SWISS 14:30 DTP576 A320 TAP 14:30 DTP614 A321 TAP 14:30 D
TP1939 AT76 TAP 14:30 AKL1693 B738 KLM 14:35 AS4201 A310 SATA 14:35 DTP432 A319 TAP 14:35 DTP804 A319 TAP 14:40 D
FR1886 B738 Ryanair 14:45 A
94
TP693 E190 TAP 14:45 ALX2092 A320 SWISS 14:50 ATK1759 A321 Turkish Airlines 14:50 AAF1624 A321 Air France 14:55 AFR1672 B738 Ryanair 14:55 ATP554 A320 TAP 14:55 D
U27615 A319 easyJet 15:00 DVR613 B738 TACV 15:00 DS4143 A320 SATA 15:00 DTP948 A321 TAP 15:00 D
TP1952 AT76 TAP 15:00 DTP402 E190 TAP 15:00 D
UX1155 E195 Air Europa 15:05 AZI307 A319 Aigle Azur 15:05 ATP356 A319 TAP 15:05 D
FR1887 B738 Ryanair 15:10 DLH1168 A321 Lufthansa 15:10 ATP757 A321 TAP 15:10 A
U22367 A319 easyJet 15:15 ATP801 A320 TAP 15:15 A
U27642 A319 easyJet 15:20 ATP1687 A319 TAP 15:20 DTP1105 AT76 TAP 15:20 ATP1039 E190 TAP 15:25 ATP862 A320 TAP 15:30 D
TP1947 AT76 TAP 15:30 ATP541 E190 TAP 15:30 AKL1694 B738 KLM 15:35 DTP573 A319 TAP 15:35 ATP439 A319 TAP 15:35 ATP663 A319 TAP 15:40 ATP839 A319 TAP 15:40 ATP842 A320 TAP 15:40 D
TP1027 A320 TAP 15:40 ATP1141 AT76 TAP 15:40 ATK1760 A321 Turkish Airlines 15:45 DU22368 A319 easyJet 15:45 DUX1156 E195 Air Europa 15:45 DTP557 A319 TAP 15:45 ATP931 A320 TAP 15:45 A
AF1625 A321 Air France 15:50 DZI308 A319 Aigle Azur 15:50 D
U27638 A319 easyJet 15:55 ATP1271 A320 TAP 15:55 ATP1024 E190 TAP 15:55 DIB3102 A320 Iberia 16:00 ALX2093 A320 SWISS 16:00 D
95
U28722 A319 easyJet 16:00 DTP1062 AT76 TAP 16:00 DTP1960 AT76 TAP 16:00 DAT980 B738 Royal Air Maroc 16:10 A
LH1169 A321 Lufthansa 16:10 DVY7982 A320 Vueling 16:10 ATP364 A321 TAP 16:10 D
TP1036 E190 TAP 16:10 DTP496 E190 TAP 16:10 D
TP1676 A319 TAP 16:15 ATP1868 A319 TAP 16:20 ATP359 A319 TAP 16:20 ATP438 A319 TAP 16:25 DTP288 A343 TAP 16:25 A
TP1112 AT76 TAP 16:30 DTP1949 AT76 TAP 16:30 APV8149 GLF5 Private 16:30 DU26253 A319 easyJet 16:35 ATP1903 A319 TAP 16:35 DIB3103 A320 Iberia 16:40 DU27655 A319 easyJet 16:40 DU27687 A319 easyJet 16:40 DFR1798 B738 Ryanair 16:45 D
TP11 A332 TAP 16:50 DFR1142 B738 Ryanair 17:00 AVY7983 A320 Vueling 17:00 DTP209 A332 TAP 17:00 D
TP1962 AT76 TAP 17:00 DU26254 A319 easyJet 17:05 DTP533 A319 TAP 17:05 AAT981 B738 Royal Air Maroc 17:10 D
U21448 A320 easyJet 17:10 DTP1016 A319 TAP 17:15 DFR1143 B738 Ryanair 17:25 DFR2932 B738 Ryanair 17:30 DTP1689 A319 TAP 17:30 DTP1951 AT76 TAP 17:30 AAF1194 A321 Air France 17:40 AFR2253 B738 Ryanair 17:40 DU27664 A319 easyJet 17:45 ATP1037 A319 TAP 17:50 ATP1511 A319 TAP 17:55 DEW9602 A320 Eurowings 18:00 AFR3091 B738 Ryanair 18:00 ATP566 A320 TAP 18:00 D
TP1964 AT76 TAP 18:00 DBA502 A320 British Airways 18:05 A
96
TP441 A319 TAP 18:05 ATO3928 B738 Transavia France 18:10 AVY8466 A320 Vueling 18:10 AS4124 A320 SATA 18:10 ATP548 A319 TAP 18:10 DTP572 A320 TAP 18:10 D
TP1443 AT76 TAP 18:10 ATP532 A319 TAP 18:15 DTP783 A320 TAP 18:15 A
TP1482 A320 TAP 18:15 AU27663 A319 easyJet 18:20 DU27616 A319 easyJet 18:20 AVY1981 A321 Vueling 18:20 AFR3092 B738 Ryanair 18:25 DTO3416 B738 Transavia France 18:30 ATP1265 A320 TAP 18:30 ATP1558 A320 TAP 18:30 ATP1953 AT76 TAP 18:30 AAF1195 A321 Air France 18:35 DTU318 A319 Tunisair 18:35 A
EW9603 A320 Eurowings 18:40 DHV6203 B738 Transavia 18:40 ATO3417 B738 Transavia France 18:45 DTP1046 A320 TAP 18:45 DBA503 A320 British Airways 18:50 D
SN3819 A319 Brussels Airlines 18:50 AU27607 A319 easyJet 18:50 DVY8467 A320 Vueling 18:50 DTP1904 A319 TAP 18:50 AUX1157 E195 Air Europa 18:55 AW62393 A320 Wizz Air 18:55 AWT7023 AT43 Swiftair 18:55 DFR2692 B738 Ryanair 19:00 AU24435 A319 easyJet 19:00 ATP928 A320 TAP 19:00 DTP283 A343 TAP 19:00 D
TP1966 AT76 TAP 19:00 DTO3929 B738 Transavia France 19:05 DVY1982 A321 Vueling 19:05 DTP341 A320 TAP 19:05 ATP358 A320 TAP 19:05 DS4137 A320 SATA 19:10 DTP833 A319 TAP 19:10 A
TP1025 E190 TAP 19:10 ATP1677 A320 TAP 19:15 DFR2078 B738 Ryanair 19:20 ATP552 A319 TAP 19:20 D
97
FR2693 B738 Ryanair 19:25 DHV5952 B738 Transavia 19:25 DTU319 A319 Tunisair 19:25 DTP1686 A319 TAP 19:25 AU24434 A319 easyJet 19:30 DW62394 A320 Wizz Air 19:30 DTP1955 AT76 TAP 19:30 AS66645 B762 StarAir 19:30 DUX1160 E195 Air Europa 19:40 DTP446 A320 TAP 19:40 DTP838 A320 TAP 19:40 D
SN3820 A319 Brussels Airlines 19:45 DTP1018 E190 TAP 19:50 DTP477 E190 TAP 19:50 A
FR2096 B738 Ryanair 19:55 DIB3106 A319 Iberia 19:55 ATP668 A319 TAP 19:55 DTP612 A320 TAP 19:55 D
TP1305 A320 TAP 19:55 ATP873 A320 TAP 20:00 A
TP1970 AT76 TAP 20:00 DQY8166 B752 DHL 20:00 DTP431 A319 TAP 20:15 ATP362 A319 TAP 20:15 DTP954 A319 TAP 20:15 D
VY8462 A321 Vueling 20:20 ATP401 E190 TAP 20:20 A
TAY247B B734 ASL AirlinesBelgium
20:20 D
FR1882 B738 Ryanair 20:30 ATP1029 A319 TAP 20:30 ATP1064 AT76 TAP 20:30 DTP1957 AT76 TAP 20:30 AIB3105 A319 Iberia 20:35 DTP1113 AT76 TAP 20:35 ATP321 E190 TAP 20:35 A
TP1090 AT75 TAP 20:40 DTP1067 AT76 TAP 20:40 ATP1045 E190 TAP 20:40 AW61165 A320 Wizz Air 20:45 ATP808 A319 TAP 20:45 DTP953 A321 TAP 20:45 ATP619 A321 TAP 20:45 ATP493 E190 TAP 20:45 ATP809 A319 TAP 20:50 ATP661 A319 TAP 20:50 ATP927 A332 TAP 20:50 A
98
TP494 E190 TAP 20:50 DFR1883 B738 Ryanair 20:55 DU22369 A320 easyJet 20:55 ALY5161 B738 ElAl 21:00 AFR1799 B738 Ryanair 21:00 ATP1048 A319 TAP 21:00 DTP1968 A320 TAP 21:00 DVY8463 A321 Vueling 21:05 DS4142 A320 SATA 21:05 AS4126 A320 SATA 21:05 A
W61166 A320 Wizz Air 21:20 DTP369 A319 TAP 21:20 A
TP1136 AT76 TAP 21:20 DU22370 A320 easyJet 21:25 DTP1691 A319 TAP 21:25 DLH1792 A321 Lufthansa 21:30 ATP1959 AT76 TAP 21:30 ATP1022 E190 TAP 21:35 DTP478 E190 TAP 21:35 DEI486 A320 Aer Lingus 21:40 ATP579 A320 TAP 21:40 A
TP1104 AT76 TAP 21:45 DTP551 A320 TAP 21:50 A
FR2624 B738 Ryanair 21:55 AS4129 A320 SATA 21:55 D
TP1481 A321 TAP 21:55 DLH1496 A320 Lufthansa 22:00 AU28721 A319 easyJet 22:00 ATP763 A320 TAP 22:00 A
TP1958 E190 TAP 22:00 DTP449 A319 TAP 22:05 ALY5162 B738 ElAl 22:10 DTP1432 AT76 TAP 22:10 DFR2095 B738 Ryanair 22:15 ATP863 A320 TAP 22:15 AYU609 B763 euro Atlantic
Airways22:20 D
EI487 A320 Aer Lingus 22:20 DFR2625 B738 Ryanair 22:20 DAF1124 A320 Air France 22:20 ATP843 A320 TAP 22:25 ATP367 A321 TAP 22:30 A
TP1971 AT76 TAP 22:30 AU21449 A320 easyJet 22:45 AKL1697 B738 KLM 22:50 AU27688 A319 easyJet 22:50 AU27608 A319 easyJet 22:50 A
99
U27656 A319 easyJet 22:50 ABA504 A320 British Airways 22:55 ALH1172 A321 Lufthansa 23:00 ATP1909 A319 TAP 23:05 DTP1693 A320 TAP 23:05 DTP1021 E190 TAP 23:05 ATP1928 A319 TAP 23:15 DTP1047 A320 TAP 23:15 ATP1688 A320 TAP 23:20 A
TP75 A332 TAP 23:20 DTP289 A343 TAP 23:20 DIB3118 A321 Iberia 23:25 ATP792 A320 TAP 23:25 DTP87 A343 TAP 23:30 D
FR2931 B738 Ryanair 23:40 AFR2252 B738 Ryanair 23:45 AS4136 A320 SATA 24:40 A
B.2 Future Demand
Flight Aircraft Airline Time A or DTP2222 A321 TAP 07:40 AVY1983 A321 Vueling 08:05 ATP1110 A320 TAP 08:10 AUA64 B752 United Airlines 08:20 A
TP2223 A321 TAP 08:30 DTS580 A332 Air Transat 08:35 A
FR1316 B738 Ryanair 08:40 AHV5953 B738 Transavia 08:40 AVY1984 A321 Vueling 08:50 DZI301 A320 Aigle Azur 08:55 A
TP1111 A320 TAP 09:00 DFR1317 B738 Ryanair 09:05 DU24433 A319 easyJet 09:10 AAA738 B752 American Airlines 09:15 A
HV5954 B738 Transavia 09:20 DZB1720 A321 Monarch 09:25 ATP3333 A319 TAP 09:30 AU24432 A319 easyJet 09:35 DD83616 B738 Norwegian Air
International09:35 A
ZI302 A320 Aigle Azur 09:45 DZB480 A321 Monarch 09:50 A0B157 B738 Blue Air 09:55 ATS581 A332 Air Transat 10:00 D
100
OU700 A319 Croatia Airlines 10:10 ADY1786 B738 Norwegian Air
Shuttle10:10 A
ZB1721 A321 Monarch 10:20 DD83617 B738 Norwegian Air
International10:20 D
UA65 B752 United Airlines 10:25 DFR2088 B738 Ryanair 10:35 AUA168 B752 United Airlines 10:35 AZB481 A321 Monarch 10:40 D
TP3334 A319 TAP 10:40 DU26981 A320 easyJet 10:45 AOU701 A319 Croatia Airlines 10:55 DDY1787 B738 Norwegian Air
Shuttle10:55 D
0B158 B738 Blue Air 11:00 DKL1695 B738 KLM 11:00 AFR2089 B738 Ryanair 11:05 DZB1324 A320 Monarch 11:15 AU26982 A320 easyJet 11:15 DKL1696 B738 KLM 11:45 DZB1325 A320 Monarch 12:05 DUA167 B752 United Airlines 12:15 DAA739 B752 American Airlines 12:15 DU27626 A319 easyJet 12:45 AU27681 A319 easyJet 15:00 DFR8038 B738 Ryanair 18:05 AFR8039 B738 Ryanair 18:30 DTO3940 B738 Transavia France 19:35 ATO3941 B738 Transavia France 20:10 D
101
