Journal of Air Transport Management 16 (2010) 86–93

Contents lists available at ScienceDirect

Journal of Air Transport Management

journal homepage: www.elsevier.com/locate/jairtraman

Assessing airport terminal performance using a system dynamics model

Ioanna E. Manataki*, Konstantinos G. Zografos
Transportation Systems and Logistics Laboratory, Department of Management Science and Technology, Athens University of Economics and Business, Evelpidon 47A & Lefkados 33,
Athens 113 62, Greece

a b s t r a c t

Keywords: Current models for airport terminal performance assessment require substantial modeling effort to be
Airport terminal customized to the configuration of a particular airport terminal and to reflect adequately the alternative
Generic model airport operational policies adopted in a user-friendly way. Therefore, there is a need to develop
Performance assessment
a generic, yet flexible decision-support tool that will facilitate high-level decision-making related to
System dynamics
fundamental changes in the structure and operation of the airport terminal system. This paper presents
a generic and easily customizable system dynamics based tool for assessing the performance of the
Athens International Airport passenger terminal under different demand and resource deployment
scenaria.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction 2. Problem context

Strategic decision-making related to airport terminal is The airport terminal constitutes a major element of the airport
a process of high complexity, as it demands the effective planning landside, as it is the boundary of the airport towards the airside. It
and coordination of multiple complex dynamic processes. The is associated with the processes and the facilities that airport
airport terminal is a highly complex large-scale system, as it customer groups visit while at the airport. Airport terminal strategic
involves a large number of entities, a large variety of types of planning and operations exhibits some distinct characteristics:
services, complex interrelations between processes. In addition,
airport terminal performance assessment involves multiple and  It involves multiple stakeholders, representing multiple and
sometimes conflicting objectives of the various stakeholder groups sometimes conflicting operational objectives.
involved in and affected by airport terminal operations. In this  Most airport decision makers are interested in addressing
context, to support effective airport terminal decision-making, a standard set of issues pertaining to airport terminal opera-
several models, both analytical and simulation, have been devel- tions, such as capacity, delays, level of service, etc.
oped. Most of the existing models require substantial modeling  The airport terminal is a large-scale system involving a large
effort to be customized to the configuration and operational number of entities and a large variety of types of services (e.g.,
environment of any airport terminal in a user-friendly way. ticketing, check-in, boarding pass control, passport control,
Therefore, there is an urgent need for developing a generic, yet security screening, customs control, baggage claim, ancillary
flexible decision-support tool that will facilitate airport terminal services, etc.).
strategic analysis.  Different customer groups (i.e., departing/arriving/transit
Recently, a generic system dynamics based airport performance passengers, well-wishers, greeters, visitors) present different
tool has been developed (Manataki and Zografos, 2009a, 2009b) service requirements and are involved in different facilities and
that can assist decision makers to examine, in an efficient way, processes.
airport terminal performance by generating alternative demand  Complex interrelations exist between the various processes.
and resource deployment scenaria. The model is customized to  Passenger flows are guided through the airport terminal
reflect the operational environment of the Athens International passenger path network, according to the terminal signage,
Airport (AIA) and is applied to assess the performance of the and interact (or are accumulated) at various terminal areas,
airport’s passenger terminal. depending on the airport terminal layout.
 There exist a large variety of airport terminal physical config-
* Corresponding author. Tel.: þ30 210 82 03 673; fax: þ30 210 82 03 684. urations and operational policies, depending on the local
E-mail address: [email protected] (I.E. Manataki). airport conditions.

0969-6997/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jairtraman.2009.10.007
 I.E. Manataki, K.G. Zografos / Journal of Air Transport Management 16 (2010) 86–93 87

 Depending upon the airport terminal configuration, some that require substantial modeling effort and knowledge to repre-
service processes may be conducted at various locations in sent a given airport terminal (Manataki and Zografos, 2009a). To fill
a terminal (e.g., centralized vs. decentralized processing this gap, a generic, yet flexible simulation model that supports
concept). airport terminal strategic analysis has been recently developed.
 Variability and stochastic events take place, such as passenger
stochastic behaviour, flight cancellations, delays, etc. 3. Modeling approach and design
 Demand for airport terminal services is primarily determined
by the airport flight schedule. The selection of the modeling approach for the development of
 Demand for airport terminal services exhibit seasonal and the airport terminal model is based on the need to address the
peaking patterns in alignment with air transport demand requirements for a generic tool, so as to deal with the basic needs of
patterns. any airport, for a tool capable to support strategic level perfor-
 Arrival patterns at airport terminal vary between airports, mance assessment, and for flexibility, so as to capture easily any
depending on the airport accessibility. Another important airport terminal configuration and operational environment.
factor affecting arrival patterns is the variability of the These requirements influence the adoption of a holistic
passenger traveling time to the airport (DeNeufville and perspective for the design and development of a macroscopic and
Odoni, 2003). aggregate simulation model, using a modular and hierarchical
 Intermodal operations at major airports affect passenger modeling approach. Specifically, system dynamics have been used
arriving/departing patterns. Depending on the modal split, as the underlying theoretical basis, since they provide a framework
different percentages of passengers may arrive at or depart to understand the operations of complex dynamic systems and
from the airport using various modes of transport (e.g., metro, view the impacts of any decision on the entire system (Forrester,
regional rail, bus, taxi, private car, rented car, etc.). Depending 1994; Sterman, 2000). A primary advantage in adopting system
on the mode used to access the airport terminal, different dynamics is the holistic view of airport terminal components and
passenger arrival patterns result for the terminal and different the analysis of their interdependencies, which leads to a clearly
travel times are observed from the platforms of the various defined system’s behaviour.
transport modes to the terminal entrance. The tool is structured into two hierarchical levels: the first level
of the hierarchy reflects the airport terminal system decomposition
Taking into account all these peculiarities of an airport terminal, into a set of airport functional areas; at the second level, the
to perform effective airport terminal strategic performance functional areas are decomposed into service facilities (modules) of
assessment, it is highly desirable to utilize decision-support the airport terminal, e.g., the passport control facility. In this
models/tools that will be able to cope with these requirements. context, every airport terminal service facility is represented by
Towards this end, researchers in the air transportation field have a module. Modules are further interconnected to compose airport
developed and applied models and tools, both analytical and functional areas. At a higher level, airport functional areas are
simulation, to model airport terminal operations and passenger composed to form the airport terminal. The composition procedure
flows (Andreatta et al., 1999; Mumayiz, 1990; Zografos and Madas, of facilities and functional areas is driven by the airport terminal
2006). For an overview of the state-of-the-art and state-of-practice physical layout, the operational policies concerning the successive
and a thorough analysis and discussion on existing airport terminal processing of terminal users through the terminal facilities, and
models/tools the reader is referred to (Manataki and Zografos, the flow characteristics of the associated passenger streams, i.e.,
2009b). Previous work reveals that existing analytical models density, flow, and speed.
generally present limitations, since they use specific distributions An illustrative example of the model structure is presented in
(or combinations of distributions) for their variables, or simplifi- Fig. 1, where a hypothetical arriving passenger flow is also depicted.
cations of many assumptions, failing thus to capture the As it can be observed, the airport terminal model for departing
complexity, variability and stochasticity of airport terminal opera- passengers is composed by four airport functional areas: Unre-
tions and flows. On the other hand, existing simulation models are stricted airport functional area, Controlled airport functional area,
either models of specific airports, or general simulation platforms Gates airport functional area, and Arrivals’ Controlled airport

Fig. 1. Model structure.

88 I.E. Manataki, K.G. Zografos / Journal of Air Transport Management 16 (2010) 86–93

functional area. Concerning the representation of airport functional passengers, after performing the necessary controls, leave the
areas, Fig. 1 presents an example of the facilities contained in every facility (‘‘arriving pax after Passport Control’’). Following the
airport functional area, e.g., the Unrestricted Airport Functional previously described modeling approach, relevant modules have
Area contains the following facilities: Ticketing, Check-in, Ancillary been developed for all processes/facilities/functional areas of the
Service facilities (i.e., Retail and Food & Beverage facilities), Waiting airport terminal, e.g., Ticketing, Check-in, Boarding Pass/Ticket
Lounges, Arrival halls and Boarding Pass/Ticket Control. Control, Passport Control, Security Screening, Ancillary Service
To provide insight to the modeling process followed for the facilities (i.e., Retail and Food & Beverage facilities), Gate Lounge,
development of the modules, the passport control facility for arrivals Baggage Claim, Customs, Arrival Hall, etc.
airport functional area is illustrated in Fig. 2. Based on system Finally, the airport terminal model is developed by the
dynamics representation notation (Forrester, 1994; Sterman, 2000), composition of airport functional areas using specific facilities as
the service facility is depicted as a stock-and-flow diagram, using interfaces. For example, the interface between Unrestricted
three basic building blocks: airport functional area and Controlled airport functional area is
the Boarding Pass/Ticket Control facility, which constitutes the
 stocks (rectangles), which represent the state of a process, e.g., physical boundary between them, as clearly illustrated in Fig. 1.
passenger accumulation in queue for passport control, At this point, it is important to mention that what determines
 flows (pipes), which represent the rate of change of this state, and the interfaces between airport functional areas is the airport
 converters (circles) which hold information about the system terminal layout and the operational policies followed, and
and affect the rate of the flows. therefore, these may differ from one airport to another. To
demonstrate this composition process, consider an abstract view
For the case of the passengers arriving with international non- of the model in Fig. 3, representing the composition of Unre-
EU, non-Schengen flights, the passport control process is depicted stricted and Controlled airport functional area of an airport
in Fig. 2. These passengers ‘‘leave’’ the Arrivals Airport Functional through the Boarding Pass/Ticket Control and the Passport
Area (‘‘pax leaving Arrivals Area for Passport Control’’) and proceed Control facilities.
to the facility (‘‘arriving pax proceeding to Passport Control’’). They As far as demand for airport terminal facilities is concerned,
wait in queue (‘‘arriving pax in queue for Passport Control’’), and Fig. 4 provides an illustrative representation of the module devel-
receive passport control service (‘‘arriving pax receiving Passport oped for determining demand patterns for the various facilities in
Control service’’). The service process is determined by the number a typical Unrestricted Airport Functional Area. Specifically, the
of passport control channels (‘‘No. of open Passport Control Points demand placed upon a service facility is expressed as the number of
in Arrivals Area’’) and the associated service rate (‘‘Passport Control customers ‘‘leaving’’ the associated airport functional area to visit
rate’’). Passengers succeeding the passport control can proceed to the service facility. It is determined taking into account: passenger
the next process. However, a percentage of passengers fail passport availability in the associated airport functional area (excluding
control and have to proceed to secondary interview. If a queue has passengers who are currently processed in other facilities), and the
been formed they wait (‘‘arriving pax in queue for secondary demand placed upon other facilities of the associated functional
interview’’), and then get serviced. Again, the service process at the area. For example, the demand placed upon the Passport Control
secondary interview depends on the number of the available facility by departing passengers is determined based on the
servers and the associated service rate. The operation ends when all following parameters (see Fig. 4):

Fig. 2. Passport control module for arriving passengers.

 I.E. Manataki, K.G. Zografos / Journal of Air Transport Management 16 (2010) 86–93 89

Fig. 3. Generic view of the model composition process for the AIA passenger terminal.

 Passengers’ availability in various locations of the Unrestricted

Airport Functional Area (‘‘pax in UnrAreas’’);
 Percentage of Extra-Schengen passengers, since only Extra-
Schengen passengers pass through Passport Control;
 Passenger allocation per Passport Control facility, which
represents the percentage of passengers visiting each facility,
in case that more than one facility exist in the airport terminal,
providing thus the ability of distinguishing between central-
ized versus decentralized concepts;
 Percentage of passengers visiting ancillary facilities (such as
retail and food/beverage facilities) in Unrestricted Airport
Functional Area;
 Percentage of passengers without ticket; and
 Percentage of passengers performing e-check-in.

Therefore, the main passenger flows that arrive at the departing

passengers Passport Control facility, and therefore describe the
demand pattern for passport control services, embrace the Extra-
Schengen passengers who do not visit ancillary service facilities
after visiting Check-in facility, the Extra-Schengen passengers
who do not visit ancillary service facilities, have a ticket and
perform e-Check-in, and the Extra-Schengen passengers, after
Fig. 4. Demand module for facilities pertaining to unrestricted airport functional area. visiting ancillary service facilities.
90 I.E. Manataki, K.G. Zografos / Journal of Air Transport Management 16 (2010) 86–93

Table 1
Model assumptions for the base case scenario at the Athens International Airport.

Operational characteristics

Service facility Number of Service rate

service (i.e., pax per min)
channels
Check-in
Business 2 0.67
Economy 12 0.67
Self-Service 3 0.5
Boarding pass/ticket control 8 6.0
Passport control 4 3.0

Security screening
Security screening 6 3.0
Item deduction 4 1.5
Secondary screening 4 2.0
Fig. 6. LOS for the check-in facility.
Passengers’ characteristics
Percentage of Schengen passengers 68.5%
Percentage of extra-Schengen passengers 31.5%
Percentage of business class passengers 12% concerning departing passenger processing and associated flows
Percentage of passengers performing 5% through the various airport functional areas and facilities are
self check-in analytically presented to demonstrate the model capabilities. In
Percentage of passengers performing 5%
particular, the objective of the application of the model is to
e-check-in
demonstrate the interactions existing among the various processes
and the utility of the model to rationally deploy resources by taking
into account the interaction effects.
4. Model application: the Athens International Airport A Base Case scenario has been developed representing a typical
peak day in terms of passenger traffic at the Athens International
The proposed model has been customized and demonstrated Airport. The basic model assumptions with regards to passenger
using real data of the Athens International Airport. Specifically, the traffic mix, passengers’ characteristics and the operational/service
model has been configured to account for both the topological and policies followed by the airport, defining the service processing at
the operational dimensions of the Athens International Airport, in the various airport terminal facilities, are presented in Table 1. In
terms of the physical airport layout and the operational policies
followed. In this paper, the outcomes of the model application

Fig. 5. a,b Check-in facility results. Fig. 7. a,b Security screening facility results.
 I.E. Manataki, K.G. Zografos / Journal of Air Transport Management 16 (2010) 86–93 91

Fig. 9. a,b Passport control facility results.

Fig. 8. a,b. Boarding pass/ticket control facility results.
of service (LOS) in terms of IATA space standards is graphically
illustrated in Fig. 6 (where LOS A ¼ 1, B ¼ 2, C ¼ 3, D ¼ 4, E ¼ 5,
this respect, the results of the simulation run for the major facilities F ¼ 6), depicting the LOS deterioration during the peak periods.
located at the Unrestricted and Controlled airport functional areas Concerning Security Screening, Athens International Airport has
(see Figs. 1, 3a, and 3b) are presented in the herewith analysis, adopted a decentralized concept, resulting in 9 Security Screening
including: i) Check-in, ii) Boarding Pass/Ticket Control, iii) Passport facilities, each one located before common gate lounges. In Fig. 7a
Control, and iv) Security Screening, under the Base Case scenario. and Fig. 7b, the simulation results for Security Screening facility
As far as the Check-in process is concerned, this is primarily B3-9, which serves Intra-Schengen passengers, are shown. Based on
governed by airlines’ policy. In this sense, airlines may apply the Base Case Scenario analysis performed, under which the
different check-in rules depending on whether the flight is Intra or number of available Security Screening channels is 6, and the
Extra-Schengen. Additionally, passengers are distinguished into number of item deduction staff and secondary screening staff is 4,
those who perform ordinary check-in (employee assisted check-in) the average waiting time amounts to 1.6 min, whereas the
and those who perform self check-in. A further distinction exists, maximum waiting time reaches 19 min. Further, Fig. 7a and Fig. 7b
between business or first-class passengers and economy-class graphically illustrate the pattern of passengers’ accumulation and
passengers, in case that dedicated check-in counters are allocated waiting time at the security screening facility, clearly demonstrating
to business or first-class passengers. Under the Base Case Scenario, that several peaks occur during the day, concentrating mainly
the number of open check-in counters are: 2 for business check-in, around 11:50–12:25 and 17:25–18:10. Since each peak lasts for
12 for economy check-in, and 3 self-service check-in counters. about 30 min with a maximum passenger waiting time of 19 min, no
The average waiting time amounts to 0.2 min, 0.8 min, and 0 min additional security screening channel is considered necessary.
for business, economy, and self-service check-in respectively, Figs. 8a, 8b, and 9a, 9b present the simulation results for the
whereas the maximum waiting time is 5 min, 13 min, and 1 min Boarding Pass/Ticket Control and the Passport Control facilities
respectively. To provide further analysis, Fig. 5a and Fig. 5b present based on the Base Case Scenario. To summarize, based on the
the simulation results during the day (x-axis of the graph repre- service control parameters, i.e., the number of available servers, and
sents time in minutes corresponding to the simulation day, i.e., the service rate of each process (Table 1), a number of performance
from 0 to 1440 min) for the check-in facility based on the Base Case measures have been evaluated by the system dynamics model, e.g.,
Scenario. In particular, Fig. 5a and Fig. 5b display the patterns of mean and maximum waiting times, patterns of passengers’ waiting
passengers’ accumulation throughout the day (i.e., the volume of time, patterns of passengers’ accumulation (i.e., volume of passen-
passengers waiting in queue), and the passengers’ waiting time in gers waiting in queue), resource utilization, and level of service.
minutes respectively (both represented in y-axis of the graph) for In addition, a number of ‘‘what-if’’ scenarios can be evaluated by
the economy check-in facility throughout the day, where it can be changing the model assumptions, i.e., by developing scenarios with
observed that two peaks occur early in the morning, i.e., around different passenger traffic volumes, or by controlling the service
02:15–02:50 a.m. and 05:20–06:10 a.m. The corresponding level parameters of the process, e.g., by increasing or decreasing the
92 I.E. Manataki, K.G. Zografos / Journal of Air Transport Management 16 (2010) 86–93

Table 2
Results for the intense demand scenario

Service facility Base case scenario Intense demand scenario

Number of servers Waiting time (minutes) Number of servers Waiting time (minutes)

mean max mean max

Check-in
Business 2 0.2 5.0 2 1.3 15.0
Economy 12 0.8 13.0 12 5.1 35.0
Self-service 3 0.0 1.0 3 0.0 1.0
Boarding pass/ticket control 8 0.2 2.0 8 0.3 4.0
Passport control 4 0.5 3.0 4 0.6 3.0

Security screening
Security screening 6 1.6 19.0 6 2.3 21.0
Item deduction 4 0.1 2.0 4 0.1 2.0
Secondary screening 4 0.0 1.0 4 0.0 1.0

Intense demand scenario 1st Run 2nd Run

i) Check-in i) Check-in
ii) Security screening

Service facility Number of servers Waiting time (minutes) Number of servers Waiting time (minutes)

mean max mean max

Check-in
Business 3 0.1 2.0 3 0.1 2.0
Economy 15 1.0 13.0 15 1.0 13.0
Self-service 3 0.0 1.0 3 0.0 1.0
Boarding pass/ticket control 8 0.4 5.0 8 0.4 5.0
Passport control 4 0.6 3.0 4 0.6 3.0

Security screening
Security screening 6 2.2 21.0 8 0.5 10.0
Item deduction 4 0.1 2.0 5 0.1 1.0
Secondary screening 4 0.0 1.0 4 0.0 1.0

number of available service points, and simulating the generated results, the increase in the number of check-in counters affected
scenarios with the help of the system dynamics model developed. positively the performance of the check-in facility, by reducing
To this end, the decision maker can obtain a clear insight of the passenger waiting times, but affects negatively mainly the security
process studied, concerning the level of service, the passenger screening process (and at a lower degree the boarding pass/ticket
waiting times, and the recourses required, along with their impli- control process), due to higher passenger accumulation. In other
cations to overall system performance. words, accelerating the preceding check-in process causes the
To elaborate on the development of alternative scenarios and bottleneck to be transferred to subsequent processes. To alleviate
the assessment of the respective airport terminal performance, an this effect, 2 service channels are added to the security screening
Intense Demand Scenario has been developed, based on the facility (2nd Run), along with increasing the number of item
assumption of 25% increase in passenger traffic. In this respect, the deduction staff by one, eliminating any delays. Based on the above
model has been applied at the Athens International Airport, analysis, it is apparent that the proposed system dynamics model
increasing passenger traffic volumes, while maintaining all other facilitates the evaluation of performance of the airport terminal
model assumptions, as presented in Table 1. The impact of the under alternative scenarios and assists airport decision-makes in
intense demand on airport terminal facilities performance is airport terminal planning and operations.
presented in Table 2, taking also into account the interactions
among them. In relation to the Base Case scenario, already analyzed 5. Concluding remarks
through Figs. 5–9, passenger delays occur in most of the critical
processes presented, with significant effect on the security A system dynamics simulation model for airport terminal anal-
screening facility. Delays are also observed at the economy and ysis has been presented. The proposed tool supports effective stra-
business check-in, and at the passport control facility, along with tegic decision-making, providing the capability of conducting
LOS deterioration. To control the impact of the intense traffic on impact analyses with respect to a variety of airport terminal
airport terminal performance, a number of what-if scenarios have performance measures, i.e., capacity, delays/waiting times, level of
been further developed, by increasing the number of available service, resource utilization, passenger accumulations at various
servers, and running the simulation. To this end, Table 2 presents facilities and time frames, etc. The major strengths of the proposed
also the model results after two alternative simulation runs: model is that it is generic, yet flexible, and can be easily adapted to
represent any airport terminal local conditions, as it enables quick
Run 1: Adding 1 check-in counter for business check-in and and easy model building through the selection of objects from
3 check-in counters for economy check-in. a library of modules. To this end, it does not require prior use
Run 2: Adding 1 check-in counter for business check-in and experience and familiarity of the user/decision maker, to customize
3 check-in counters for economy check-in, and and adapt it to the needs and characteristics of the particular airport
Adding 2 security screening points, along with increasing terminal cases under consideration. Furthermore, it allows for
the number of item deduction staff by 1. a holistic evaluation of terminal performance, capturing important
interactions and trade-offs among the various terminal facilities and
Based on the model outcomes, the interactions among the services, as opposed to analyzing in isolation the performance of
various facilities can be easily perceived. Based on the 1st Run individual airport terminal facilities. The deployment of the
 I.E. Manataki, K.G. Zografos / Journal of Air Transport Management 16 (2010) 86–93 93

proposed tool is expected to improve the quality of airport terminal Forrester, J.W., 1994. System dynamics, systems thinking, and soft OR. System
Dynamics Revue 10, 245–256.
decision-making and management, by providing insight to airport
Manataki, I.E., Zografos, K.G., 2009a. A generic system dynamics based tool for
terminal dynamics, evaluating the performance of the terminal as airport terminal performance analysis. Transportation Research C 17, 428–443.
a whole, and also identifying specific problematic areas. Manataki, I.E., Zografos, K.G., 2009b. Development and demonstration of
a modeling framework for airport terminal planning and performance evalu-
ation. Transportation Research Record 2106, 66–75.
References Mumayiz, S.A., 1990. Overview of airport terminal simulation models. Trans-
portation Research Record 1273, 11–20.
Andreatta, G., Brunetta, L., Odoni, A.R., Righi, L., Stamatopoulos, M.A., Zografos, K.G., Sterman, J.D., 2000. Business Dynamics: Systems Thinking and Modeling for
1999. A set of approximate and compatible models for airport strategic plan- a Complex World. McGraw-Hill, Boston.
ning on airside and on landside. Air Traffic Control Quarterly 7, 291–317. Zografos, K.G., Madas, M.A., 2006. Development and demonstration of an integrated
DeNeufville, R., Odoni, A., 2003. Airport Systems: Planning, Design and Manage- decision support system for airport performance analysis. Transportation
ment. McGraw-Hill, Boston. Research C 14, 1–17.