strategic and operational issues in the ... sulaiman...extendsim v9.2 simulator software from...

STRATEGIC AND OPERATIONAL ISSUES

IN THE INTEGRATED MANAGEMENT OF AN

AIRPORT – AN OPERATIONS

MANAGEMENT APPROACH

Sultan Sulaiman Alodhaibi

BSc, MSc (Mathematical Science)

Submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy (Research)

School of Chemistry, Physics and Mechanical Engineering

Science and Engineering Faculty

Queensland University of Technology

2019

Strategic and Operational issues in the integrated management of an airport – An operations management

approach i

Keywords

Airport modelling; passenger flow analysis; airport operational planning;

performance evaluation; arrival patterns; simulation; Optimisation; capacity

expansion; Simulated Annealing; Discrete-Event Simulation.

ii Strategic and Operational issues in the integrated management of an airport – An operations management

approach

Abstract

The global air transport industry is expanding rapidly. New approaches to airport

management are required to ensure that ever-increasing consumer demand is met with

adequate developments of ground operational and processing facilities; particularly

those related to effective and safe processing of passenger flows. The solution to this

problem requires the development of a new generation of fast, reliable decision-

making tools to quickly mobilise the human and technical resources available at

modern airports. Operational research aimed at developing novel airport optimisation

simulations to empower efficient management decisions is therefore a rapidly

advancing field. The research conducted highlights the improvements made in our

understanding of passenger flow modelling to date. These models can be classified as

either ‘analytical’, ‘simulation’, or ‘hybrid’ models, giving decision support

capabilities at all levels of detail: from macroscopic, through mesoscopic, to

microscopic. However, despite the current developments in understanding passenger

flow, the literature suggests that an aggregate model, integrating both outbound and

inbound processes, is still needed.

The main aim of this research is to develop a generic, holistic simulation model

that can optimise passenger flow in an international airport. To achieve such a goal,

the holistic model has been split into three phases. The first phase uses Discrete-Event

Simulation (DES) to develop a generic model of passenger flow through the outbound

processes of an international airport. The DES is built using ExtendSim V9.2 simulator

software from Imagine That. Starting from this simulation of outbound processes, the

DES was also used to investigate how the arrival pattern of outbound passengers at the

front door affects international terminal operations. Included in this are major

outbound processes such as check-in, security screening, and immigration.

Experiments demonstrated that different arrival patterns have a significant impact on

the performance of operational processes, so the best policy for passenger arrival time

can be determined. The second phase is developing a simulation model for inbound

processes, while the third phase is integrating the Advanced Resource Management

(ARM) approach inside the simulation model. In this way, an overall view of airport

operations can be realised, helping operations managers identify potential bottlenecks,


approach iii

the optimal utilisation of available resources, and both typical and maximal capabilities

with the available staff resources.

The model supports what-if and trade-off analyses by inclusion of a problem-

oriented approach. Hence, best practice can be identified for staff allocation or the

opening and closing of processing points counters. The simulation results demonstrate

that flight schedules have a large impact on passenger flows. The proposed simulation

framework and model can be used to predict ahead of time the effect of different flight

schedules and may be used as a feedback mechanism to improve the simulation model

before implementation. The results also demonstrated that airport operations

performance was significantly affected by different arrival distribution patterns, i.e.

the rate of arrivals at the airport based on the policy of opening check-in counters prior

to scheduled departure time. The developed ARM model balanced the average waiting

time and operation hours since the staff is only allocated if needed.

In summary, studying passenger flow within an international terminal using DES

allowed the integration of outbound and inbound flow processes to enhance

operational efficacy. By integrating the airport simulation model within an analytical

optimisation framework, the model can determine where additional resources should

be best allocated to reduce the overall cost of waiting. The ARM can be a decision

support tool and efficiently used to support and model real-world airport staff

allocation planning problems.

iv Strategic and Operational issues in the integrated management of an airport – An operations management

approach

List of Publications

1. Alodhaibi, Sultan, Robert L. Burdett, and Prasad K. D. V. Yarlagadda. 2017.

"Framework for Airport Outbound Passenger Flow Modelling." Procedia

Engineering 174:1100-9. doi: https://doi.org/10.1016/j.proeng.2017.01.263.

2. Alodhaibi, Sultan, Robert L. Burdett, and Prasad K. D. V. Yarlagadda. “Impact of

Passenger-Arrival Patterns in Outbound Processes of Airports." Procedia

manufacturing Engineering 30:323-30. doi:

https://doi.org/10.1016/j.promfg.2019.02.046.

3. Alodhaibi, Sultan, Robert L. Burdett, and Prasad K. D. V. Yarlagadda. “A model

to simulate passenger flow congestion in airport environment.” International

Journal of Engineering & Technology” (In press).

4. Alodhaibi, Sultan, Robert L. Burdett, and Prasad TK. D. V. Yarlagadda. “A

review of the challenges in airport terminal planning and future directions”.

(Submitted).

5. Alodhaibi, Sultan, Robert L. Burdett, and Prasad K. D. V. Yarlagadda. “A

framework for sharing staff between outbound and inbound airport processes”.

(To be submitted).

6. Alodhaibi, Sultan, Robert L. Burdett, and Prasad K. D. V. Yarlagadda. “An

analytical optimization framework for airport terminal capacity planning”. (To be

submitted).

https://doi.org/10.1016/j.promfg.2019.02.046


approach v

Table of Contents

Keywords .................................................................................................................................. i

Abstract .................................................................................................................................... ii

List of Publications ................................................................................................................. iv

Table of Contents ......................................................................................................................v

List of Figures ......................................................................................................................... ix

List of Tables ..........................................................................................................................xv

List of Abbreviations ........................................................................................................... xvii

Statement of Original Authorship ....................................................................................... xviii

Acknowledgements ............................................................................................................... xix

Chapter 1: Introduction ...................................................................................... 1

1.1 Background and motivation ............................................................................................2

1.2 Research problem ...........................................................................................................4

1.3 Research objectives ........................................................................................................5

1.4 Research significance and innovation.............................................................................5

1.5 Thesis outline ..................................................................................................................7

Chapter 2: Literature Review ............................................................................. 9

2.1 Overview ........................................................................................................................9

2.2 Current issues in airport planning and management .....................................................10 2.2.1 Passenger flow issues .........................................................................................10 2.2.2 Security issues ....................................................................................................13 2.2.3 Staff allocation issues .........................................................................................14

2.3 Studying complex systems ...........................................................................................17

2.4 Analytical methods .......................................................................................................18

2.5 Simulations methods .....................................................................................................21 2.5.1 Simulation models for passenger flow ...............................................................24 2.5.2 Simulation models of security processes ............................................................30

2.6 Optimisation methods ...................................................................................................32

2.7 Summary of the reviewed literature..............................................................................41

2.8 Knowledge gap identified .............................................................................................42

2.9 Formulation of research scope and research contributions ...........................................42

Chapter 3: Simulation Model Framework for the Outbound Passenger

Processes at an International Airport .................................................................... 43

3.1 Overview ......................................................................................................................43

3.2 The conceptual framework ...........................................................................................44

vi Strategic and Operational issues in the integrated management of an airport – An operations management

approach

3.3 Passenger flow characteristics...................................................................................... 46

3.4 Outbound processes modelling .................................................................................... 47 3.4.1 Arrival at the terminal ........................................................................................ 49 3.4.2 Check-in module ................................................................................................ 51 3.4.3 Security screening module ................................................................................. 54 3.4.4 Immigration module .......................................................................................... 55 3.4.5 Boarding procedure module ............................................................................... 57

3.5 ExtendSim models for outbound processes .................................................................. 57 3.5.1 Hierarchical blocks ............................................................................................ 59 3.5.2 ExtendSim modules description ......................................................................... 59

3.6 Numerical testing ......................................................................................................... 79 3.6.1 Impact on arrival process ................................................................................... 80 3.6.2 The impact on terminal facilities ....................................................................... 81

3.7 Chapter summary ......................................................................................................... 83

Chapter 4: The Impacts of Arrival Patterns on Airport Mandatory

Processes

……………………………………………………………………………………85

4.1 Introduction .................................................................................................................. 85

4.2 Development of passenger arrival process model ........................................................ 86

4.3 Case study 1: Impacts of different time before departure values ................................. 88 4.3.1 Behaviour of CDF of time before flight. ........................................................... 89 4.3.2 Behaviours of arrival pattern ............................................................................. 90 4.3.3 Results of simulation and discussion ................................................................. 90

4.4 Case study 2: Impacts of different mean values ........................................................... 94 4.4.1 Behaviour of CDF .............................................................................................. 95 4.4.2 Behaviours of arrival pattern ............................................................................. 96 4.4.3 Results of the simulation and discussion ........................................................... 97

4.5 Selection of best time to arrive at the airport based on the normal distribution ......... 100 4.5.1 Selection of the best scenario at each process ................................................. 102 4.5.2 Aggregation of all processes ............................................................................ 103

4.6 Chapter summary ....................................................................................................... 107

Chapter 5: A Framework for Sharing Staff between Outbound and Inbound

Airport Processes. ................................................................................................... 109

5.1 Introduction ................................................................................................................ 109

5.2 Inbound passenger flows modelling ........................................................................... 110 5.2.1 Outline of inbound flow processes .................................................................. 110 5.2.2 Inbound process simulation modelling ............................................................ 111

5.3 Inbound ExtendSim module description ..................................................................... 113 5.3.1 Block 1: Hierarchy blocks for inbound processes ........................................... 115 5.3.2 Block 2: Creating inbound passengers’ entities of an arrived flight ................ 115 5.3.3 Block 3: Inbound security module ................................................................... 122 5.3.4 Block 4: Duty free............................................................................................ 126 5.3.5 Block 5: Inbound immigration and customs module ....................................... 127 5.3.6 Block 6: Baggage claim module ...................................................................... 132 5.3.7 Block 7: Inbound quarantine module ................................................................ 133

5.4 Integrated inbound and outbound processes .............................................................. 140 5.4.1 Advanced resource management (ARM) ......................................................... 140


approach vii

5.4.2 Mechanism of development of algorithms .......................................................142 5.4.3 The logic of development algorithms ...............................................................144 5.4.4 Input data of integrated model ..........................................................................145 5.4.5 Categories of algorithms...................................................................................146

5.5 Chapter summary ........................................................................................................157

Chapter 6: Case Study - Validation of the Simulation Model ..................... 158

6.1 Introduction ................................................................................................................158

6.2 KKAI operational data ................................................................................................161

6.3 Model application and simulation process ..................................................................165

6.4 Simulation results and analysis .....................................................................................168 6.3.1 Description of Terminal 1 results .....................................................................169 6.3.2 Description of Terminal 2 results .....................................................................173 6.3.3 Results analysis and discussion ........................................................................177

6.5 Chapter summary ........................................................................................................180

Chapter 7: Application of Advanced Resource Management (ARM) ........ 182

7.1 Introduction ................................................................................................................182

7.2 Overview of airport resource management .................................................................183 7.2.1 Model demonstration ........................................................................................184 7.2.2 General input data ............................................................................................184

7.3 Simulation results and analyses ..................................................................................185 7.3.1 Static Allocation Base Case method .................................................................186 7.3.2 Dynamic resource allocation method ...............................................................191 7.3.3 Comparison of overall impact of static and dynamic allocation models ..........194 7.3.4 Identify the variation of static and dynamic allocation methods ......................196

7.4 Dynamic approach demonstration ..............................................................................197 7.4.1 Adding and removing staff polices for non-integrated processes ....................197 7.4.2 Sharing staff policy for integrated processes ....................................................200

7.5 Chapter summary ........................................................................................................203

Chapter 8: An Analytical Optimization Framework ................................... 204

8.1 Problem description and formulation .........................................................................205 8.1.1 Model notation .................................................................................................205

8.2 Simulated annealing....................................................................................................206 8.2.1 Simulated annealing algorithm description ......................................................209

8.3 Numerical testing and analysis ...................................................................................211

8.4 Chapter summary ........................................................................................................217

Chapter 9: Conclusion ..................................................................................... 218

9.1 Introduction ................................................................................................................218

9.2 Summary and discussion ............................................................................................218

9.3 Research contributions................................................................................................221 9.3.1 Framework for airport outbound passenger flow modelling ............................222 9.3.2 Investigating the effect of arrival patterns of departing passengers on the

departure terminal operations ...........................................................................222 9.3.3 Advanced resource management strategies ......................................................223

viii Strategic and Operational issues in the integrated management of an airport – An operations management

approach

9.3.4 Development of a novel holistic model for facilitating outbound and

inbound processes ............................................................................................ 224 9.3.5 Strategic and operational planning techniques ................................................ 224 9.3.6 Practice contribution ........................................................................................ 225

9.4 Limitations and future research directions ................................................................. 225

Bibliography ........................................................................................................... 229

Appendices .............................................................................................................. 239


approach ix

List of Figures

Figure 2 - 1: The relationships between consecutive and non-consecutive

processes (Hsu et al., 2014). ........................................................................ 20

Figure 2 - 2: Mathematical model of the SSCP using a Jackson open queuing

network (Dorton & Liu, 2015) ..................................................................... 20

Figure 2 - 3: An illustration of the ExtendSim model used for simulation and

optimisation of an airport baggage handling system (Savrasovs, et al.

2009). ........................................................................................................... 25

Figure 2 - 4: Scheme for processing the inbound international passengers using

the Anylogic software package (Curcio, et al. 2007). .................................. 26

Figure 2 - 5: Grid element scheme for the probabilistic discrete determination

of human motion from a given position to the nearby positions on the

grid (Schultz & Fricke, 2011). ..................................................................... 27

Figure 2 - 6: Diagram of stock and flow (Manataki & Zografos, 2009b) .................. 29

Figure 3 - 1: Overview of an airport’s terminal processes, including outbound

and inbound processes ................................................................................. 44

Figure 3 - 2: Airport system model. ........................................................................... 45

Figure 3 - 3: Airport outbound processes (Shuchi, 2016). ......................................... 46

Figure 3 - 4: The input modelling of an outbound simulation model. ....................... 48

Figure 3 - 5: Flowchart for generating outbound passenger attributes. ..................... 49

Figure 3 - 6: The relationship between departing passengers’ arrival times and

the type of flight (Ashford et al. 2011). ....................................................... 50

Figure 3 - 7: Flowchart of check-in processing at international airport. .................... 52

Figure 3 - 8: Module hierarchy of check-in system. .................................................. 53

Figure 3 - 9: Flowchart of screening checkpoints for processing at the

international airport. ..................................................................................... 55

Figure 3 - 10: Flowchart of immigration system at international airport. .................. 56

Figure 3 - 11: Proposed logic design for outbound system consisting of eight

blocks. .......................................................................................................... 60

Figure 3 - 12: Input data represented by passenger attributes. ................................... 60

Figure 3 - 13: Block 1: ExtendSim simulation for outbound system. ........................ 62

Figure 3 - 14: Block 2: Prioritise arrivals. ................................................................. 63

Figure 3 - 15: Block 2: Algorithm for assigning passenger high priority. ................. 64

Figure 3 - 16: Block 3: Assigning check-in type using the decision module. ........... 65

Figure 3 - 17: Block 3: Self-service module. ............................................................. 66

Figure 3 - 18: Block 3: Hierarchical block for check-in group module. .................... 66

Figure 3 - 19: Block 4: Decision module for selecting class of travellers. ................ 67

x Strategic and Operational issues in the integrated management of an airport – An operations management

approach

Figure 3 - 20: Block 4: Decision modules for number of bags and check-in

type. .............................................................................................................. 68

Figure 3 - 21: Block 4: Delay time module. ............................................................... 69

Figure 3 - 22: Block 4: Workstation control module. ................................................ 69

Figure 3 - 23: Block 5: Diplomatic decision module. ................................................ 70

Figure 3 - 24: Block 5: Queue system module. .......................................................... 71

Figure 3 - 25: Block 5: Processing time distribution module. ................................... 71

Figure 3 - 26: Block 5: Security first failure module. ................................................ 72

Figure 3 - 27: Block 5: Workstation control module. ................................................ 72

Figure 3 - 28: Block 6: Probability of random explosive check module. .................. 73

Figure 3 - 29: Block 6: Random check decision module ........................................... 73

Figure 3 - 30: Block 6: Processing time distribution at random explosive check. ..... 74

Figure 3 - 31: Block 7: SmartGate user decision module check. ............................... 75

Figure 3 - 32: Block 7: SmartGate processing time distribution................................ 75

Figure 3 - 33: Block 7: Immigration queue system. ................................................... 76

Figure 3 - 34: Block 7: Immigration workstation control module. ............................ 76

Figure 3 - 35: Block 8: Walking time distribution for boarding gate module. .......... 77

Figure 3 - 36: Block 8: Walking time distribution to boarding gate module. ............ 78

Figure 3 - 37: Block 8: Flowchart for boarding procedure algorithms. ..................... 79

Figure 3 - 38: Arrivals patterns for 100% flights full. ............................................... 80

Figure 3 - 39: Arrivals patterns for 50% flights full. ................................................. 80

Figure 3 - 40: Security queue length 100% flights full. ............................................. 82

Figure 3 - 41: Security queue length 50% flights full. ............................................... 82

Figure 3 - 42: Immigration queue length 100% flights full. ...................................... 82

Figure 3 - 43: Immigration queue length 50% flights full. ........................................ 82

Figure 4 - 1: Flowchart for modelling passenger arrivals at the international

terminal ........................................................................................................ 87

Figure 4 - 2: CDF of passengers arriving before flights for a given mean (µ):

(a) µ = 60 min; (b) µ = 90 min; (c) µ = 120 min; (d) µ = 150 min; (e)

µ = 180 min (f) µ = 210 min. ....................................................................... 89

Figure 4 - 3: Departing passenger arrival profiles at airport terminal for

different (Ω) under given (µ): (a) µ = 60 min; (b) µ = 90 min; (c) µ =

120 min; (d) µ = 150 min; (e) µ = 180 min (f) µ = 210 min. ....................... 91

Figure 4 - 4: Queue lengths of different time before flight given µ = 60 .................. 93

Figure 4 - 5: Queue lengths of different time before flight given µ = 90 .................. 93

Figure 4 - 6: Queue lengths of different time before flight given µ = 120 ................ 94


approach xi

Figure 4 - 7: CDF of passengers arriving at airport for flight (i) for a given time

before the flight under different (µ): time of passenger arriving (a) Ω =

120 min; (b) Ω = 150 min; (c) Ω = 180 min; (d) Ω = 210 min; (e) Ω =

240 min ........................................................................................................ 95

Figure 4 - 8: Departing passenger arrival profiles at airport terminal for

different (µ) given (Ω): (a) Ω = 120 min; (b) Ω = 150 min; (c) Ω =

180 min; (d) Ω = 210 min; (e) Ω = 240 min ................................................ 96

Figure 4 - 9: Queue lengths of different mean value at a given time before

flight ............................................................................................................. 99

Figure 4 - 10: Check-in queue length and waiting time for all scenarios ................ 102

Figure 4 - 11: Security screening queue length and waiting time for all

scenarios ..................................................................................................... 102

Figure 4 - 12: Immigration queue length and waiting time for all scenarios ........... 103

Figure 4 - 13: (a-g) the impacts of different arrival patterns based on the

priority for each processes ......................................................................... 106

Figure 5 - 1: An illustration of inbound passenger facilitation processes (Wu et

al., 2014). ................................................................................................... 110

Figure 5 - 2: Flowchart of the upper level of the inbound process flow model ....... 112

Figure 5 - 3: The input modelling of the inbound simulation model ....................... 113

Figure 5 - 4: Block 1: The high level of inbound flow modelling ........................... 115

Figure 5 - 5: Block 2: Structure of the hierarchical block ‘passengers

disembarking’ ............................................................................................ 116

Figure 5 - 6: Inbound flight attributes ...................................................................... 116

Figure 5 - 7: Algorithm for creating inbound passenger attributes .......................... 117

Figure 5 - 8: Inbound passenger attributes ............................................................... 118

Figure 5 - 9: Block 2: Mechanism of linking inbound passenger attributes with

the ExtendSim model.................................................................................. 119

Figure 5 - 10: Block 2: Walking speed module ....................................................... 120

Figure 5 - 11: Block 2: Arrival calculating gate distance module ........................... 121

Figure 5 - 12: Block 3: Walking time module ......................................................... 121

Figure 5 - 13: Block 3: The hierarchy module of x-ray and routing for random

check .......................................................................................................... 122

Figure 5 - 14: Block 3: Simulated queue of x-ray check and hierarchy block of

workstation ................................................................................................. 123

Figure 5 - 15: Block 3: Characteristics of processing items .................................... 124

Figure 5 - 16: Block 3: Storing the outputs of security ............................................ 124

Figure 5 - 17: Block 3: Random explosive decision module ................................... 125

Figure 5 - 18: Block 3: queuing and processing time characteristics of random

explosive check .......................................................................................... 126

xii Strategic and Operational issues in the integrated management of an airport – An operations management

approach

Figure 5 - 19: Block 4: Assign duty free attributes module ..................................... 127

Figure 5 - 20: Flow chart of inbound immigration checkpoint process ................... 128

Figure 5 - 21: Block 5: Logic design of inbound immigration checkpoint

process ........................................................................................................ 129

Figure 5 - 22: Block 5: Inbound SmartGate user decision module .......................... 129

Figure 5 - 23: Block 5: Inbound immigration queue module ................................... 130

Figure 5 - 24: Block 5: Process characteristics of inbound immigration ................. 131

Figure 5 - 25: Block 5: Logic design of the SmartGate module .............................. 131

Figure 5 - 26: Block 6: Baggage claim decision queue module............................... 132

Figure 5 - 27: Block 6 baggage claim delay time module ....................................... 133

Figure 5 - 28: Flow chart for quarantine process ..................................................... 134

Figure 5 - 29: Block 6: Logic chart of quarantine module ....................................... 135

Figure 5 - 30: Block 7: Inbound declaration decision module ................................. 136

Figure 5 - 31: Block 7: Declaration queue module .................................................. 137

Figure 5 - 32: Block 7: Inbound immigration queue module ................................... 138

Figure 5 - 33: Block 7: Inbound Quarantine queue module ..................................... 138

Figure 5 - 34: Block 7: Nothing to declare queue module ....................................... 139

Figure 5 - 35: Block 7: Quarantine workstation for nothing to declare line ............ 140

Figure 5 - 36: Flowchart framework for ARM model ............................................. 142

Figure 5 - 37: Resource allocation dialog for global array ...................................... 143

Figure 5 - 38: Dynamic link between parameter tables and ExtendSim ................... 144

Figure 5 - 39: Staff attributes for the ARM model ................................................... 145

Figure 5 - 40: Flowchart algorithm for airline staff allocation module ................... 147

Figure 5 - 41: Flowchart for integrated module for boarding procedure ................. 148

Figure 5 - 42: Flowchart algorithm of quarantine staff management module ......... 150

Figure 5 - 43: Flowchart 1-2 of security resource allocation management .............. 152

Figure 5 - 44: Flowchart 2-2 of security resource allocation management .............. 153

Figure 5 - 45: flowchart algorithm of immigration resource allocation ................... 155

Figure 6 - 1: The terminals and runways of the King Khalid international

airport. ...................................................................................................................... 159

Figure 6 - 2: Passenger movement numbers at KKIA from 2005-2016 (Statista,

2019) .......................................................................................................... 159

Figure 6 - 3: Scheme of passenger flow types at KKIA terminals (Kloosterziel

et al., 2009). ................................................................................................ 160

Figure 6 - 4: Processing time distributions for departure processes of Terminal

1 of KKIA .................................................................................................. 163


approach xiii

Figure 6 - 5: Processing time distributions for departure processes of Terminal

2 of KKIA .................................................................................................. 164

Figure 6 - 6: Flowchart of KKIA departure flow processes (researcher’s

illustration) ................................................................................................. 166

Figure 6 - 7: Process of calculating cycle time ........................................................ 168

Figure 6 - 8: Arrival pattern and profile of Terminal 1 passengers. ........................ 170

Figure 6 - 9: a, b Terminal 1 check-in process results; c, d Terminal 1 security

screening process results; e, f Terminal 1 immigration process results ..... 172

Figure 6 - 10: Arrival pattern of Terminal 2 passengers and entering Terminal 2

distribution. ................................................................................................ 173

Figure 6 - 11: a, b Terminal 2 check-in process results; c, d Terminal 2 security

screening results; e, f Terminal 2 immigration process results .................. 175

Figure 7- 1: Example of ExtendSim database for output data .................................. 186

Figure 7- 2: Check-in average waiting time using the SABC method ..................... 187

Figure 7- 3: Influence of the SABC method on boarding procedures ..................... 188

Figure 7- 4: Security screening average waiting time using the SABC method ...... 189

Figure 7- 5: Immigration average waiting time using the SABC method ............... 190

Figure 7- 6: Quarantine average waiting time using the SABC method.................. 190

Figure 7- 7: Check-in average waiting time using the dynamic resource

allocation method ....................................................................................... 191

Figure 7- 8: Influence of dynamic resource allocation method on boarding

procedures .................................................................................................. 192

Figure 7- 9: Security screening average waiting time using the dynamic

resource allocation method ........................................................................ 193

Figure 7- 10: Immigration average waiting time using the dynamic resource


Figure 7- 11: Quarantine average waiting time using the dynamic resource


Figure 7- 12: Static allocation method results.......................................................... 195

Figure 7- 13: Dynamic resource allocation method results ..................................... 195

Figure 7- 14: Variation in the SABC method .......................................................... 196

Figure 7- 15: Variation in the dynamic resource allocation approach ..................... 196

Figure 7- 16: Check-in facility results of adding and removing staffing policies ... 199

Figure 7- 17: Quarantine facility results of adding and removing staffing

policies ....................................................................................................... 200

Figure 7- 18: Outcomes of simulated immigration staff sharing rules ................... 201

Figure 7- 19: Security screening facility results of sharing staffing policies ........... 203

Figure 8- 1: General steps of the simulated annealing ............................................. 207

Figure 8- 2: Selecting the best initial parameters ..................................................... 208

xiv Strategic and Operational issues in the integrated management of an airport – An operations management

approach

Figure 8- 3: snapshot of simulated annealing results ............................................... 211

Figure 8- 4: SA optimisation results using the random method of creating new

solutions ..................................................................................................... 215

Figure 8- 5: AS optimisation results using the method of creating new solution

using local technique .................................................................................. 217


approach xv

List of Tables

Table 2 - 1: Summary of the recent issues associated with airport operations ......... 16

Table 2 - 2: Comparison between the three types of simulation (Ma, 2013;

Owen, 2013) ................................................................................................. 22

Table 2 - 3: Summary of models used to address airport problems ........................... 38

Table 3 - 1: Summaries of major elements and processing facilities of check-in

module.......................................................................................................... 54

Table 4 -1: Selection of Ω values under a fixed µ values .......................................... 88

Table 4-2: Detailed output of ExtendSim simulation model for case study 1 ............ 92

Table 4-3: Selection of Ω values under different µ values ........................................ 95

Table 4-4: Detailed output of ExtendSim simulation model ...................................... 98

Table 4 - 5: Summary of the simulation results…………………………………....101

Table 4- 6: Illustration of different conditions to select the best scenario………....104

Table 4-7: Summary of the results of selection of the best policy of time before

flight………………………………………………………………………………..107

Table 6 - 1: Summary of model default parameters at the KKIA international

airport. ........................................................................................................ 169

Table 6 - 2: Comparisons of waiting time in queue and cycle time at check-in,

security and immigration between the simulation data and real time

data of Terminal 1. ..................................................................................... 178


security and immigration between the simulation data and the real

time data of Terminal 2. ............................................................................. 178


security and immigration between the simulation data and the real

time data at Brisbane International Airport. .............................................. 179

Table 7 - 1: Summary of common operational input data for the experiments ....... 185

Table 7 - 2: Summary of eligible sharing polices .................................................... 202

Table 8- 1: Summary of the input data .................................................................... 212

Table 8- 2: summary of simulated annealing results using random search

technique .................................................................................................... 213

Table 8- 3: summary of simulated annealing results using local search

technique .................................................................................................... 214

Table 8- 4: Check-in additional resource results using the random technique ........ 214

Table 8- 5: Check-in additional resource results using local technique.................. 216

xvi Strategic and Operational issues in the integrated management of an airport – An operations management

approach

Table 9- 1: Comparisons of developed ARM results with Kierzkowski and

Kisiel (2016) .............................................................................................. 224


approach xvii

List of Abbreviations

ABS Agent based simulation

ARM Advanced Resource Management

BHS Baggage Handling Systems

CPM Capacity Planning Model

DES Discrete Event Simulation

FIDS Flight Information Display System

GPSS General Purpose Simulation System

IATA International Air Transport Association

ICAO International Civil Aviation Organization

KPIs Key Performance Indicators

LOS Level of Service

SA Simulated Annealing

SD System Dynamic

SSCP Security Screening Checkpoint’s

PAX Passengers

VBA Visual Basic for Applications

xviii Strategic and Operational issues in the integrated management of an airport – An operations

management approach

Statement of Original Authorship

The work contained in this thesis has not been previously submitted to meet

requirements for an award at this or any other higher education institution. To the best

of my knowledge and belief, the thesis contains no material previously published or

written by another person except where due reference is made.

Signature:

Date: 17/07/2019

QUT Verified Signature


approach xix

Acknowledgements

I would like to express my sincere gratitude to my Principle Supervisor Professor

Prasad K.D.V. Yarlagadda, and my Associate Supervisor Dr. Robert Burdett for their

continuous support and guidance throughout this thesis. This work would not have

been possible without their encouragement, insightful comments and valuable advice.

Also, I would like to thank Professor Clinton Fookes from Airport of the future project

for his kind support by provide me with resource and information that I need to conduct

my research.

I extend my thanks to AbdulAziz Abu Abat a director of event & logistics

support at Saudi Telecom for his kind support for helping me to contact with Abdulaziz

Al-Ruwais a Project manager at Riyadh International airport for sharing the

information that I need.

I would like to thank the staff at the Science and Engineering Faculty and HDR

Support Officers who provide me with an excellent research environment and patiently

answered my questions regarding my enrolment and research progress.

Special thanks to my mother Norah Alshamkh and to my father Sulaiman

Alodhaibi who the reason of being here and for their great love and advices. My sincere

thanks go to my wife Bushra and my lovely daughter Norah for their support during

tough and difficult time to complete this PhD journey.

Finally, my thanks go to my friends for all their love and encouragement. Words

cannot express how grateful I am to all of you.

Chapter 1: Introduction 1

Chapter 1: Introduction

In 2017, the number of airline travellers exceeded 4 billion globally, an increase

in global air transport demand of about 8.1% from 2016 (ICAO, 2018), a number that

will continue to increase and is predicted to exceed 7.1 billion by 2035 (International

Air Transport Association (IATA). Airport management and airlines have discussed

the possibility of changing and updating several policies related to flight schedules,

staff allocation and other operational policies to accommodate future demand growth,

and to provide better quality of services and security. An international airport terminal

is a large and complex system, since it involves inbound and outbound passenger flow

processes, each with unique operations. Some airports have a slightly different process

and new airports designed in the future may require further changes to the standard

process in light of new security concerns being faced in our modern world. Safety

concerns in recent times have caused many changes to security screening procedures

which affect passenger throughput times. After September 11, 2001, when terrorists

brought down the twin towers in New York using passenger planes, airport security

has become more critical. Another problem facing modern airports is the limited

infrastructure and staffing capacity, such as numbers of common check-in counters

and numbers of personnel available, to deal with increasing passenger numbers.

Due to the complexity of the airport terminal, there is a need for new effective

management approaches to ensure that the skyrocketing demands in air travel are met

with adequate developments of ground operational and processing facilities. In this

thesis a holistic simulation framework is developed using Discrete-Event Simulation

(DES) to simulate entire passenger flow processes within international terminals. DES

is used because it can handle stochastic system and temporal variation demands (Chiu,

2002; Guizzi, Murino, & Romano, 2009; Rauch & Kljajić, 2006). The motivation and

background for this thesis is presented in section 1.1. The following section 1.2

describes the research problem and research question. Section 1.3 presents the aims

and objectives of this research, while the contribution statement and the methodology

used to address research gaps are presented in section 1.4. Section 1.5 completes the

chapter with an outline of the thesis structure.

2 Chapter 1: Introduction

1.1 BACKGROUND AND MOTIVATION

In recent years, airports have played a significant role in economic growth,

connecting cities and countries around the world. Numerous passengers choose to

travel by airline in preference to other transportation modes such as trains, buses and

private cars. Based on the International Civil Aviation Organization (ICAO, 2018)

report for 2017, the number of airline travellers globally exceeded 4 billion, a growth

in global air transport demand of about 8.1% compared with the previous year

(Statista, 2018). The direct, indirect and induced contributions of airline travel to the

global GDP in 2017 was US$776 trillion, including almost 2.78 million jobs generated

globally (IATA, 2017b). In Australia, based on the report by the Bureau of

Infrastructure Transport and Regional Economics (2012), the number of air travellers

increased annually at the rate of about 5%, from 27 million in 1977–1978 to 135.1

million in 2010–2011. The same report indicated that the number of passenger

movements in Australia would reach approximately 279.2 million in 2030–2031

(BITRE, 2012).

One of the major causes for this substantial and rapid increase in air transport

demand is the rapid growth of the international trade and globalisation of industries.

Another significant cause lies with rapidly increasing tourism – over 54% of

international tourists now travel by air (IATA, 2017a). This increase in air transport

demand imposes significant strains on air travel operations and facilities that are

expected to keep up with the growing passenger flows. This includes airport capacity

and ability to process the increasing numbers of passengers with high efficiency and

minimum delay. The required expansion of airport capacity may be limited by the

available resources (e.g. limited available land), environmental impacts and lengthy

approval processes (Barnhart, Fearing, Odoni, & Vaze, 2012). In addition, extension

of major airport infrastructure is typically time-consuming and costly, which

highlights the need for the development of smart systems and methods to improve

airport performance within the available infrastructure limitations.

It is conventional to sub-divide or classify airport operations into those relevant

to the arrival procedures of incoming passengers and departure procedures for

outgoing passengers. The arrival processes and facilities include disembarking,

immigration, baggage claim, and quarantine procedures. The departure processes and

facilities include check-in, security screening, immigration and customs, boarding and


take-off procedures. It is these departure flow processes that have the greatest impact

on the entire operation of passenger terminals and other elements of the airport.

According to De Neufville, Odoni, Belobaba, and Reynolds (2013), the departure

process requires significantly more time than the arrival process because it sometimes

involves services provided to transit passengers. Consequently, most research focuses

on achieving greater efficiency in the departure process to alleviate congestion in

airport terminals (Du, Yu, Jiang, & Ji, 2015; Guizzi et al., 2009; Manataki & Zografos,

2009b; Odoni & de Neufville, 1992; Solak, Clarke, & Johnson, 2009; Wu &

Mengersen, 2013; Zidarova & Zografos, 2011).

Airports are very complex, interlinked systems and any operational problems

with any of their elements can jeopardise the performance of other elements, creating

significant bottlenecks, long passenger queues, congestion and overall delays (De

Neufville & Odoni, 2003; Manataki & Zografos, 2009b; Zografos & Madas, 2006).

For example, disruption, congestion and uneven passenger inflow into the terminal

processing points, caused by the operation of the landside element (including the

infrastructure and facilities associated with the arrival of passengers to the airport),

could have a significant impact on the performance of the terminal (such as passenger

boarding and take-off procedures). Similarly, the airside element can influence the

performance of landside elements, for example, through an excessive flow of inbound

or transit passengers, which could take staff from the landside elements.

Different parts of the airport have been analysed or optimised in isolation, for

example, analysis and simulations conducted for outbound flow processes within

airport terminals to achieve optimal and most efficient passenger processing (Du et al.,

2015; Guizzi et al., 2009; Manataki & Zografos, 2009b; Solak et al., 2009; Wu &

Mengersen, 2013; Zidarova & Zografos, 2011). Some simulations and modelling have

examined the departure processes located in the landside including passenger arrival

at the airport and facilities external to the terminal (Correia & Wirasinghe, 2013; Wu

& Mengersen, 2013; Zhou, Huang, Jia, & Jiang, 2014), while others have examined

passenger boarding and plane take-off procedures that are located in the airside area

(Bazargan, 2007; Budesca & Juan, 2014; Jacquillat & Odoni, 2015; Van Landeghem

& Beuselinck, 2002). There has, however, been little holistic research on the

performance and capacity issues for the whole airport system including both outbound

and inbound types of passenger flow processes. The lack of a comprehensive


overarching model can lead to incorrect assessments and the adoption of incorrect

policies. This can result in significant financial losses due to unexpected delays and

inefficient use of the airport resources and related infrastructure. For example, the costs

of airport delays in the US alone exceeded US$32.9 billion in 2007 and were around

$US41 billion in 2008 (Ferguson, Kara, Hoffman, & Sherry, 2013). In addition,

operational efficiency at an airport can also directly impact safety, security and

customer satisfaction (Rauch & Kljajić, 2006), which can also be associated with

significant direct and indirect financial and other losses.

1.2 RESEARCH PROBLEM

As discussed above, the demand for air travel is growing, but the decision support tools

are not fully optimal. The complexity of airport terminal systems creates the need for

models that can provide an integrated view for all terminal operations. As most

existing tools and models are focused only on individual processes and address

fragmented sections of decision making procedures in airports (Wu & Mengersen,

2013; Zografos, Madas, & Salouras, 2013), a holistic simulation model integrated with

advanced resource management and analytical frameworks developed in this thesis.

The proposed modelling includes simulation and analysis of the impacts of

passenger arrival patterns associated with different arrival modes corresponding to

different means of passenger transportation to the airport facilities. In addition, the

proposed research involves modelling and simulating the impacts of the flow of the

inbound passengers on the processing of outbound passengers. A holistic view is

required to not only study the performance of outbound and inbound flow processes,

but to also investigate the possible interactions between the two elements with respect

to resource allocation policies. In this thesis the following research questions are

addressed:

Question 1: How can a holistic simulation model be utilised/developed to

analyse/optimise the passenger flows in an international airport by

integrating inbound and outbound processes?

Question 2: Can the simulation model identify the impacts of different

passenger arrival patterns and help identify the best passenger arrival

policy?


Question 3: Can advanced resource management (i.e. staff allocation,

opening and closing of check-in counters, etc.) policies be integrated within

a holistic simulation model?

Question 4: Can a simulation model be used within an analytical

optimization framework to improve the efficiency and operation of airports?

1.3 RESEARCH OBJECTIVES

The general aim of this research is to develop comprehensive approaches to

model and optimise airport operations, which would involve the integrated

consideration and analysis of the two airport terminal elements of a modern airport –

outbound and inbound. The main objectives of this research can be summarised as

follows:

Develop a simulation framework simulation model using DES for outbound

passengers

Investigate the impacts of different passenger arrival patterns and determine

the best policy for outbound passengers.

Develop advanced resource management strategies and analyse the effects

on airport operations. Use results to provide direction and best policies for

more effective staff allocation and reallocation within the international

airport terminal.

Integrate inbound flow processes with outbound processes to investigate the

effect of inbound passenger flows on outbound flows and vice versa.

Develop an analytical optimisation framework to be used within the

simulation model for capacity planning.

1.4 RESEARCH SIGNIFICANCE AND INNOVATION

This thesis makes several contributions to the understanding of passenger flows

within airport terminals. It contributes to the development of a new decision-making

support tool aimed at operational improvements and performance optimisation of the

airport terminals, considering key performance indicators such as average/maximum

waiting time and average/maximum queue length. Using this tool, airport operators

can (i) effectively identify any potential bottlenecks, (ii) optimally utilise available


resources, and (iii) determine the typical and maximum capabilities and capacity flows

utilising the available personnel and other resources. This will lead to increased

passenger satisfaction, balanced operating hours and acceptable waiting times in front

of processing stations (Andreatta, De Giovanni, & Monaci, 2014; Dowling,

Krishnamoorthy, Mackenzie, & Sier, 1997; Kirk, 2013b; Wu & Mengersen, 2013;

Yamada et al., 2017; Zografos et al., 2013).

The outcomes of the research will not be limited to Australian airports and will

also significantly contribute to the development of the general knowledge of efficient

operational management practices in the air travel industry. The methodology

developed in this research will facilitate further development of airport modelling and

simulations. The innovation of this thesis is based on the following achievements

filling in the current existing knowledge gaps:

The development of a new methodology based on a unique combination of

an optimisation approach within simulation using ExtendSim simulation

software. This methodology can capture the stochastic nature (the effect of

uncertainty) of the airport system and its dynamic changes over time

(Sachidananda, Erkoyuncu, Steenstra, & Michalska, 2016; Yamada et al.,

2017). To achieve this goal, the methodology is divided into four phases:

o Development of framework for outbound processes flow to predict the

effect of different flight schedules. The framework may be used as a

feedback mechanism for improvement before implementation. It is also

used to identify the effect of different arrival patterns on the

performance of check-in, security and immigration processes.

o Integration of outbound and inbound process flows to understand

terminal process capabilities and facilitate the efficient processing of

increasing numbers of air travel passengers with minimum delay. The

structure of the model is based on a hierarchical model using ExtendSim

hierarchy block. This structure gives the model more flexibility to

enable rapid and easy modification according to airport design.

o Development of the advanced resource management approaches to

dynamically allocate and reallocate personnel, and to identify effective

opening and closing counters policies.


o Determination of where additional resources should be allocated using

a capacity expansion model to reduce passenger time spent waiting in

front in front of terminal processes and improve airport operational

efficiency.

The developed model is a flexible decision support tool to deal with

passenger demands and other unexpected phenomena that might occur in

the terminal. This is done by conducting what-if scenarios to evaluate

alternatives and possible changes in policies related to passenger patterns,

dynamic opening and closing of counters, allocating and reallocating of

staff. The proposed approach is significant as it can be used by people who

are non-simulation experts.

This project makes a new contribution to the general area of operational research

through combining the effects and synergistic impacts of the two elements of airport

terminals—outbound and inbound—which to date have largely been considered as

separate entities.

1.5 THESIS OUTLINE

This thesis consists of eight chapters, including this introductory chapter which

presents the research problem, motivation and significance, aims and objectives and

the thesis outline. The rest of the thesis has been organised as follows:

Chapter 2 presents the literature review relating to passenger flows modelling,

security issues modelling and staff allocation issues modelling. The literature is

grouped into three major models including analytical, simulation and hybrid models.

In Chapter 3, a generic framework for outbound passenger flow modelling is

introduced with respect to the input modelling for the ExtendSim model. In Chapter 4

the model is used to investigate the impacts of different arrival patterns on the

performance of departure operational processes. The outputs of the simulation model

are demonstrated to be more intuitive than those of other models. The model can help

airport operators better understand the distribution of arrivals to the airport to provide

better levels of services.

Chapter 5 extends the simulation model including inbound processes flow then

integrated with the advanced resource management approach to manage allocation and


reallocation of staff within airport terminals. This approach can provide insights into

the causes of bottlenecks, passenger logistics, and the relationships between different

processes that share a set of commonalities and functions.

Chapter 6 presents the validation processes of the simulation model to

demonstrate that this model accurately represents an actual airport terminal using King

Khalid International Airport (KKIA) as a case study.

Chapter 7 discusses the application of Advanced Resource Management (ARM)

modelling and how the model can be used to study the variation in the complex

environment of allocating and reallocating staff as well as resource sharing policies.

Chapter 8 presents an analytical optimization framework to perform capacity

planning in order to improve the efficiency of the airport and to meet future demand.

Finally, Chapter 9 presents and discusses the conclusions of the research including the

study limitations and recommendations for further research.

Chapter 2: Literature Review 9

Chapter 2: Literature Review

2.1 OVERVIEW

This chapter reviews past research focused on the current problems associated

with airports’ operational performance and the challenges of planning. It also provides

insight into the capabilities and limitations of the existing models. The models consider

the analysis and performance optimisation of separate airport elements. For example,

analyses and simulations of airport terminals have been conducted to achieve optimal

passenger processing (Du et al., 2015; Guizzi et al., 2009; Manataki & Zografos,

2009b; Solak et al., 2009; Wu & Mengersen, 2013; Zidarova & Zografos, 2011).

Simulations and models were also designed for landside elements, including passenger

arrival at the airport via the landside infrastructure and facilities external to the

terminal (Correia & Wirasinghe, 2013; Eilon & Mathewson, 1971; Tošić, 1992; Wu

& Mengersen, 2013; Zhou et al., 2014), and for airside elements, including passenger

boarding and plane take-off procedures (Bazargan, 2007; Budesca & Juan, 2014;

Jacquillat & Odoni, 2015; Van Landeghem & Beuselinck, 2002).

A wide range set of parameters are used to characterise passenger flow,

including flight schedules, service rates and resources, and the facilitation process and

associated passenger characteristics (e.g. nationality, as it influences which Customs

lane the passenger can use). Section 2.2 explains the current issues associated with

airport planning and management including passenger flow problems and other related

issues such as security and staff allocation problems. Section 2.3 provides an overview

of the types of approaches, such as analytic, simulation and hybrid approaches, used

to solve the proposed issues. Section 2.4 and section 2.5 discuss analytical and

simulation models for the purpose of airport passenger flow and resource allocation

issues. Section 2.6 describes the use of hybrid and optimisation models for a better

understanding of the characteristics of the airport terminal system and its dynamics.

10 Chapter 2: Literature Review

2.2 CURRENT ISSUES IN AIRPORT PLANNING AND MANAGEMENT

2.2.1 Passenger flow issues

Passenger flow is a result of numerous individuals’ movement within airport

terminals. The primary activities of passengers moving in such an environment are

travelling and walking between processes and facilities to be served. This section

analyses existing research conducted to address the issues of passenger flows within

airports. The irregular flows occurring in airport terminal areas represent a significant

management challenge, for instance, determining the number of service counters to

open, and personnel allocation and reallocation issues (Fonseca, Casanovas, & Ferran,

2014; Wu & Mengersen, 2013). Different policies are associated with departure and

arrival patterns, for example, employee rostering and operation schedules, and with

respect to particular airport operations and activities, such as check-in, duty free

shopping, access control, and waiting areas (Fonseca et al., 2014; Manataki &

Zografos, 2009b; Zografos & Madas, 2006). Passenger flow in airport terminals can

be divided into three main categories: departing passengers, arriving passengers, and

transferring passengers (Ma, 2013; Manataki & Zografos, 2009b).

Another significant problem when studying passenger flow is capturing

stochastic elements. This is because, as Guizzi et al. (2009) argued, passengers behave

differently inside airports according to their previous experiences. Thus, in order to

assist decision makers at the airport terminal to address sudden and unforeseen

congestion conditions, extensive research has been conducted on uncertainty. Yamada

et al. (2017) examined links between passenger behaviour and facilities and identified

numerous kinds of probable congestion circumstances. Ma et al. (2011, 2012, 2013)

investigated the uncertainty factors that can impact the route choice decision making

of passengers, as well as the complex behaviours outside the required processes. To

overcome this complexity, researchers integrated the discretionary activities of

passengers with the standard processing units within the terminal. These discretionary

activities were in alignment with major factors such as walking distance and remaining

time. Passenger flow in new terminals was demonstrated using simulation models of

the passenger flow in the new terminal of Heathrow Airport (Beck, 2011). This model

sought to understand the system before the terminal opened. In addition, some

researchers have included group dynamics in passenger flow models (Cheng, 2014;

Cheng, Reddy, Fookes, & Yarlagadda, 2014). These researchers argued that group


dynamics could have a significant impact on the performance and utilisation of

terminal facilities. The effect of group dynamics can be seen in the dwell time at each

processing unit, the level of service (LOS) at the processing units, and in the

discretionary activities (Cheng, 2014).

Estimating departure passenger flow is a prerequisite to improving airport

terminal resources for meeting dynamic travel demands. For this reason, Sun and

Schonfeld (2015) proposed a simulation-based approach to cater to the issue. Firstly,

an estimate of the number of passengers for every booked flight is done. Then, the

passenger behaviour of every passenger is simulated through randomly generating

departure time dependent on probabilistic distribution. This is followed by a count of

the number of passengers for every time interval. The efficiency or suggested approach

was validated by collecting data from Beijing Capital International Airport.

Research looking at limited resources in airports, such as terminal check-in staff

and security equipment, has used prediction methods to determine how best to use

these resources, and how this can change the level of service in the terminal. Mu,

Cheng, Zhang, and Zhang (2014) designed a method based on neural networks and

support vector regression to predict passenger traffic in the departure terminal of

Harbin Taiping International Airport. Similar work was conducted by Chiang and

Taaffe (2014). Their study used a simulation model as a theoretical model to estimate

occupancy of the zone in the concourse. The methodology involved a finger-pier

concourse with twelve gates and four groups of moving walkways to predict passenger

occupancy in every zone. The model proposed in the research can be used for effective

decision-making to manage concourse operations under any Level of Service. The

flexibility in simulation tools developed through research provides airport researchers

and planners with significant information to make informed decisions when

considering passenger conveyance systems and corridor congestion.

Passenger flow analysis and performance evaluation have been widely reported

in the literature, for instance, Rauch and Kljajić (2006) developed a DES to evaluate

the performance of existing departure processing facilities by focusing on ticketing,

check-in, immigration, and boarding in order to improve passenger satisfaction. By

selecting proper thresholds, the length of the queue is controlled in a specific range.

Nikoue, Marzuoli, Clarke, Feron, and Peters (2015) generalised the idea of passenger

walk time to a model which is independent of origin gate, by deploying mixed models.


They focused on simulations and system dynamics to model airport performance. The

study used following information sources: Flight Information Display System (FIDS)

dataset; DIMIA datasets (stamps at immigration); and Dwell time for all the

simulation results compared to actual wait times, it was observed that simulation times

were much higher, nearly 2 pm each day, than actual results.

Past research has also considered the dynamic operation of terminal systems (e.g.

Manataki & Zografos, 2009, 2010; Nikoue et al., (2015). For example, Manataki and

Zografos (2009b, 2010) designed their model to answer questions regarding

operational concepts, including the number of staff needed for a particular process,

and the percentage of passengers who use e-check-in or online check-in. In Manataki

and Zografos (2009b), two parameters (walking speed and path distance distributions)

were used to determine the flow of passengers from one area to another.

Adacher, Flamini, Guaita, and Romano (2017) pointed out that the quick rise in

passenger traffic and reduced expansion of airport capacity have limited the capability

of airports to maintain satisfactory customer services. Hence, the authors propsoed an

optimization model based on a Surrogate method which provided a specific layout and

passenger flow, decided the number of security control checkpoints and check-in desks

to reduce cost function. Tests were performed using the Napoli-Capodichino (IT)

airport as a case study. As it was a preliminary study, the solution approach was applied

to one airline firm. The results demonstrated that queues at check-in desks satisfy the

tolerance threshold Nt. On the other hand, there is an additional discomfort cost for

security control check-in points. The study, in the end, provides an efficient algorithm

to determine various critical resources for the airport terminal departure operations. It

is recommended that it is implemented with a group of firms and compared with other

optimization algorithms. This will help in optimizing the departure area of existing

airports and designing the size of the departure area for a new airport.

The capacity-planning problem has been studied with respect to transient

demand patterns. An example of this is the study carried out by Solak et al. (2009) to

determine the optimal design for airport terminals, and to expand the capacity for

different processing areas in the presence of uncertainty. Sun and Schonfeld (2015)

also investigated the uncertainty within the terminal. This included the non-linearity

of facility performance functions as represented by delay level as a function of

utilisation rates of capacity, and demand fluctuations as indicated by uncertainties in


traffic predictions. It is believed that passenger departure flow is an important process

for any airport facility because of the fixed departure time of flights.

This section has identified current problems associated with passenger flow such

as congestion in queueing areas, delays at workstations, rate of passenger arrival at the

airport, group dynamics. The next section examines security issues in airports.

2.2.2 Security issues

Safety concerns in recent times have caused many changes to security screening

procedures and this impacts passenger throughput times. After the incident on

September 11 2001, when terrorists brought down the World Trade Centre, New York

City, USA using passenger planes, airport security has dramatically increased.

Therefore, security managers at airports require methods for quantifying changes in

the level of security to avoid terrorist attacks (Skorupski & Uchroński, 2018).

There have been a number of publications looking at airport security. These

publications discuss the problem of security control, particularly with regards to

forbidden objects being carried into the restricted areas of airports (van Boekhold,

Faghri, & Li, 2014). The existing research can be categorised based on its related topic

areas, including the significance of the security screening system in airport operations,

the capacity of security screening areas, and dynamic system management (Dorton &

Liu, 2015; Kierzkowski & Kisiel, 2016, 2017; Leone & Liu, 2011; Skorupski &

Uchroński, 2016; van Boekhold et al., 2014).

Security screening systems are influenced by many factors, such as the

distribution of arrivals, limited physical resources (security line), and operational

factors. Dorton and Liu (2015) investigated the main external factors influencing the

security screening checkpoint’s (SSCP) operational efficiency. This study took into

account the system’s dependent measures of throughput and cycle time. Furthermore,

Chitty, Yang, and Gongora (2017) stated that airports face pressures of reducing costs

of waiting time at the security lane zone by decreasing the working hours of the lane

and maintaining passengers’ service level. In this regard, evolutionary methods can

reduce both objectives.

Kierzkowski and Kisiel (2015) investigated the impact of the behavioural

characteristics of the operators and passengers on the reliability of the system.

Additionally, they discovered irregular flow of travellers in the security system, which


caused numerous peaks. To solve this problem, the number of staff and technical

resources needs to be increased to ensure the smooth performance of the process. Li,

Gao, Xu, and Zhou (2018) conducted a recent study on numerous passenger strategies

and built mathematical models to portray them; using structures of network queuing

for airport security check-points and conducting various numerical experiments to

compare the performance of numerous queuing structures.

Many recent papers have focused on the problems of allocating the appropriate

number of resources, including to the security zone, and this issue will be discussed in

more detail in the following section.

2.2.3 Staff allocation issues

Currently, airport passengers and operators are faced with issues of providing

fast access to the facilities of the airport and preventing congestion during peak

periods. Delays are caused by ground operations and the efficiency of terminal

processes are assumed to have huge importance. To resolve these issues, many efforts

have been made to improve passenger travel experience and airport management’s

performance. Scheduling can be defined as the allocation of activities or actions to

resources according to specific performance criteria (Spyropoulos, 2000). Some

research models study staff scheduling and staff rostering problems. For example, Blaž

Rodič (2010) proposed the idea of workforce shift allocation and rapid rescheduling,

as per dynamic flight scheduling. (Abdoul Soukour, Devendeville, Lucet, & Moukrim,

2013) discussed staff scheduling issues in security operations. The memetic algorithm

is used to solve the problem, assuming that it is possible to close or open counters

depending on demand using a dynamic programming approach.

Dowling et al. (1997) discussed the development of a software product utilised

to formulate rosters for nearly 500 staff of a main international airline at one of the

busiest airports in the world. The key issue was to establish a robust algorithm that

offers optimised monthly rosters for airport service employees. The study was

successful in describing an overall system and the algorithm to solve rostering issues

linked to the system. Soukour, Devendeville, Lucet, and Moukrim (2012) proposed an

approach to resolve realistic scheduling issues in the airport security service domain.

The authors divided the problem into three steps: days off schedule, shift scheduling,

and staff assignment. Then, they emphasise the last step by offering two algorithms,

global assignment and greedy algorithms, to offer an initial solution. This solution is


improved through Iterative algorithm Destruction/Construction, IDC shifts. All

algorithms are applied in Java and tested on Intel Xeon Quad Core at 2.4 Ghz. IDC

shifts help include further new limitations therefore, it is concluded as the best

solution. Sabar, Montreuil, and Frayret (2012) aimed to use a multi-agent based

algorithm model for personnel scheduling and rescheduling in a dynamic setting of a

fast-paced multi-product assembly centre.

Effective allocation of Ground Service Equipment (GSE)to aircraft standing on

the apron is determined with the help of a framework (Integrated Airport Apron Safety

Fleet Management – AAS) as emerged in the European sponsored project (Andreatta

et al., 2014). The basic model used in research is the conceptual Integer Programming

formulation of Ground-Service Resource Allocation Problem (GRAP). Andreatta et

al. (2014) proposed a fast heuristic approach to display how the issue can be

decomposed into sub-issues. It was recommended that GRAP can be improved to

consider robustness against unforeseen delays which often happen in airport aprons.

Parlar and Sharafali (2008) conducted their research to determine the optimal

number of open counters over a specified period. Lin, Xin, and Huang (2015)

described the problems with ground crew rostering and shift allocation in an attempt

to better manage the opening and closing of check-in counters. Recent research by

Rodič and Baggia (2017) attempted to solve the problem of a lack of schedules for

airport check-in employees, especially regarding work groups and overlapping skills.

A summary of the issues associated with airport operations reported in the

literature is presented in Table 2-1.


Table 2 - 1: Summary of the recent issues associated with airport operations

Airport Area Issues Inbound/Outbound Papers

Terminal entry/boarding Operational efficiency

problems

Customer satisfaction

Outbound Rauch and

Kljajić (2006)

Queuing areas Quality of service to

PAX

Congestion problems

Outbound

Check-in and security Complex traffic in side

departing system

Delays at processing

station

Outbound Guizzi et al.

(2009)

Processing, holding, and

flow facilities of the

airport terminal

Congestion

System complexity

Airport operation

problem

Outbound Manataki and

Zografos (2009)

Discretionary activities

and processing units Complex movement

flows

Uncertainty factors

Outbound Ma, et al. (2011,

2012, 2013)

Airport waiting room Presence of bottlenecks

Waiting room capacity

Level of service

Outbound Ju, Wang, and

Che (2007)

Discretionary activities

and processing units Passenger group

dynamics

Group behaviour

Evacuation strategy

Outbound Cheng (2014)

Check-in/security Passenger traffic flow

Randomness of

passenger flow

Outbound Mu et al., (2014)

Terminal processing

points and passageways Stochastic future

demand

Capacity expansion

Passenger terminal

design

Outbound Solak, Clarke,

and Johnson

(2009)

Terminal

airfield system

cargo facility

Congestion effects

Air traffic growth

Out/inbound Sun and

Schonfeld

(2015)

Security screening Performance of SSCP

Baggage volume of

PAX

Alarm rate of baggage

screening device

Outbound Dorton and Liu

(2015)

Security screening Efficacy characteristics

of carry process

Intensity of

passenger flow

Outbound Kierzkowski

and Kisiel

(2015, 2016)


Factors affecting

reliability

Security screening Acceptable quality of

passenger service

Lack of management

Outbound Kierzkowski

and Kisiel

(2015, 2016)

Check-in concourse

Manpower planning

New terminal

Outbound Beck (2011)

Security screening Workforce demand

days-off scheduling

Shift scheduling

Staff assignment

Outbound Abdoul Soukour

et al. (2013)

Check-in Allocation of resources

for check-in counters

Outbound Parlar and

Sharafali (2008)

Check-in Allocating staff for

check-in

Rostering of ground

flow

Outbound Lin et al. (2015)

Departure procedures Scheduling staff

Ground crew processes

Outbound Rodič and

Baggia (2017)

Departure procedures and

baggage handling system Delay at processing

stations

Process delay

Delay control

Outbound Hsu and Chao

(2014)

Entire departure

processes Human behaviour

Analysed walking

behaviour

Outbound Schultz and

Fricke (2011)

Check-in and security Human resource

allocation

Outbound Siadat, Arain,

and Ruwanpura

(2012)

Terminal gates Passenger experience

Gate scheduling

Gate assignment

problem

Outbound /Inbound Kim, Feron,

Clarke,

Marzuoli, and

Delahaye

(2013)

2.3 STUDYING COMPLEX SYSTEMS

A system has various definitions; most of the proposed definitions are very

similar. The earliest definition was provided by Schmidt and Taylor (1970), who noted

that the system is about of a group of entities, e.g. people, that interact together for

achievement of some logical end. Backlund (2000) defined the system as a set of

elements with interacting parts forming a complex. A similar definition is used by

Miller (1978, p. 17) “A system is a set of interacting units with relationships among


them”. In practice, Law and Kelton (1991, p. 3) says the system is always based on the

objectives of a specific study. This means that the group of entities forming a system

for one study might be only a subgroup of the overall group of another system.

This section discusses different types of models used to solve the current

problems occurring in airport terminal systems. These models predominantly consider

departure systems to measure the performance of workstations and to understand

significant factors that affect system’s performance. Wu and Mengersen (2013)

pointed out that existing airport models can be categorised into four groups, capacity

planning, operational planning and design, security policy and planning, and airport

performance review. These models can be analytic, simulation, and hybrid approaches

as well. They require different levels of detail (e.g. macroscopic, microscopic, and

mesoscopic) and have deterministic and stochastic characteristics (Wu & Mengersen,

2013; Zografos & Madas, 2006). The models capture different performance metrics

for ‘operational efficiency’, including service time, queue length, and throughput. To

bridge this gap, several models have been developed. These methods can be separated

into four categories: analytical, simulation, optimisation, and integrated models (L.

Cheng, Yarlagadda, Fookes, & Yarlagadda, 2014; Law & Kelton, 1991; Tošić, 1992;

Wu & Mengersen, 2013; Zidarova & Zografos, 2011).

2.4 ANALYTICAL METHODS

Most of the earlier literature looking at airport terminals has focused on

analytical models. These models are exemplified by deterministic queue models that

assess significant performance metrics, such as waiting and service time for

passengers, and queue length at individual processing stations (Barbo, 1967; Tošić,

1992; Wu & Mengersen, 2013). A review by Wu and Mengersen (2013) stated that

deterministic models use a graphic form of the cumulative arrival and departure

profiles from the service facility under inspection. The deterministic model’s

advantages are that the two performance measures (queue length and average wait

time) can be easily determined depending on the one-to-one vertical and horizontal

distance between arrival and departure for a given facility (Wu & Mengersen, 2013).

The disadvantage, however, is that these types of models do not take into account

uncertainties in the arrival and departure profiles, and are unable to find the maximum

wait time for a single traveller (Wu & Mengersen, 2013).


To raise throughput of the system of security checks and maintaining safety

standards, a novel hybrid parallel queuing system dependent on passenger

classification is suggested by Hu and Chen (2017). It is a combination of virtual queue

system, modified model M/M/n, and basic queuing model M/M/1. The assumptions in

the study are made regarding security check progress. To address issues and

bottlenecks in the basic model, a properly optimised and modified model M/M/n is

introduced. The results reveal that when n is increased, wait time of M/M/n reduces

quickly which supports the analysis that throughput of M/M/n is much better than

n*M/M/1.

Hsu et al. (2014) developed a queuing model for passengers and their baggage

with different connecting airport terminal facilities. This study divided the departure

process stations into two types. The first type served single flights, including boarding,

loading, and transit flights. The second type served multiple flights of different

airlines, and included immigration, security, and baggage sorting. The authors

proposed an analytical model to investigate the impact of delay propagation among

these processes at Taoyuan International Airport. The relationships between both

sequential and non-sequential processes were investigated. Consecutive processes

refer to the airline services for multiple flights, such as check-in, while non-

consecutive processes refer to the airport services for multiple flights, such as

immigration, security screening, and baggage sorting. The results suggested that the

non-consecutive processes performed better as they provided more buffer time

between different operations and fewer delays from previous facilities (see Figure 2-

1). Hsu et al. (2014) further explored the control strategies for both flight and departure

process delays. In their study, the characteristics of the departure process at airport

terminals were presented by the “time network” of the departure process in order to

identify the required time for completing the process.


Figure 2 - 1: The relationships between consecutive and non-consecutive processes (Hsu et al., 2014).

Dorton and Liu (2015) developed a similar analytical model representing a

queuing network based on a combination of M/G/1, M/G/2 and M/M/1 servers as seen

in Figure 2-2. This study focused on two different factors. The first type was the

internal factors, such as staffing and equipment, and the second type was the external

factors, such as baggage volume and the alarm rate of security screenings.

Figure 2 - 2: Mathematical model of the SSCP using a Jackson open queuing network (Dorton & Liu,

2015)


Solak et al. (2009) divided the airport terminal system into two main areas:

passageways and processing stations (i.e. check-in and security checkpoints). They

proposed an analytic approximation of the maximum commuter delay in the two areas

of the terminal, including walking time, processing time, and queue length. Their

assumptions included the relationship between width of passageway and flow rates,

the measurement of walking speeds, and the fact that processing time changes over a

day. Further expansion of this theory was made by Sun and Schonfeld (2015), who

presented a strategic capacity planning model for airport systems that considered the

uncertainties in forecasting traffic levels. The key issue for conducting such research

is the facility performance functions “delay levels as functions of capacity utilizations

rate” Sun and Schonfeld (2015, p. 1). This function is nonlinear, which complicates

the solution’s design. To solve this problem, the authors proposed a deterministic total

cost minimisation model as a first phase, and then extended it into a stochastic model

by considering the uncertainties in traffic predictions. An outer-approximation

algorithm was used to solve the mixed integer nonlinear problem.

2.5 SIMULATIONS METHODS

Simulation refers to a group of theories, applications and methods that replicate

a real system of behaviour for evaluation and experimentation. Simulation models are

used to mimic complex systems over time by applying assumptions related to a

system’s operations (Diefenbach, 2010). Simulation is a powerful and flexible tool that

can assist the representation of the assumptions of the model and its conditions in

logical and mathematical relationships (Winston & Goldberg, 2004).

Simulation approaches can be classified based on three characteristics. Firstly,

simulations can be either static or dynamic in nature. Static models are not time

sensitive; events have the same validity if they occur a second apart or a year apart.

Dynamic simulations, which are more common, involve events that are time sensitive,

such as a manufacturing process. Secondly, models can be defined as continuous or

discrete. Continuous simulations represent systems that comprise continuous change,

such as pressure levels or fluid levels, whereas discrete event simulation model

systems comprise events that occur at a specific point in time. Discrete simulations are

effective for modelling parts or people that arrive at specific times and undergo

processes at specific times. The operations of an airport terminal are discrete and

would be accurately modelled by DES. Finally, simulations can be deterministic or


stochastic. Deterministic simulations have no random input, meaning that events

always happen at exactly the same time, for instance fixed appointments. Stochastic

simulations are simulations where at least some of the events occur at random times

(Dorton, 2011).

The simulation approaches can be classified into three sets (Table 2-2):

1. Discrete Event Simulation (DSE) modelling

2. System Dynamic (DS) modelling

3. Agent based simulation (ABS) modelling

Table 2 - 2: Comparison between the three types of simulation (Ma, 2013; Owen, 2013)

Discrete Event

Simulation

System Dynamic Agent based

simulation

Demonstration Demonstrate the system

as queues activities,

processes, schedules

Demonstrate the system

as flows and stocks

Proactive and

autonomous agents that

interact with each other

to achieve their aims

The key of problem Randomness related to

interrelated events and

processes

Problem can be

understood by analysing

the causal reaction effects

Individual agent modules

with directions of their

interfaces.

Mathematics Depending on statistical

distributions.

Depending Mathematical

modelling

Based on Algorithms,

simple probability and

logic

Communication ease True illustration of

system

Showing the model

design and numerical

results perfectly

Very good in illustrating

the behaviour of

individual entities.

Model accuracy Because of heavy

dependence on data, the

model processes accurate

valid static model

The accuracy of the

model is high because of

its heavy reliance on data

Difficult to constructs

but they are accurate

models


The overall approach of DES is to deal with randomness related with

interconnected events that leads to system behaviour. The theory of DES has been

applied to numerous fields to obtain insight into the complexity of the system being

studied. It has also been used to analyse the performance of systems (Dorton, 2011).

According to Wolverine Software Company’s website, DES software “allows you to

place your system under a microscope and explore its operation under laboratory

conditions” (Wolverine Software Company, 2014). Banks (2010, p. 12) also points out

that DES “is the modelling of systems in which the state variable changes only at a

discrete set of points in time.” According to the Winter Simulation Conference (WSC),

a well-known and distinguished conference, there are many applications for DES and

it is seen in many disciplines. Examples of these applications are: healthcare services,

supply chain management, transportation modes and traffic, and military applications.

Other good examples of DES systems include banks, warehouses, and gas stations

(Tarshizi, 2014).

DES has many advantages in real-world systems. First, a DES can describe a

real system’s features and characteristics and allows the developer to make any

changes to the study system. It also enables the testing of variations of the system

without affecting the actual system (Diefenbach, 2010). In addition, a DES can assist

with the definition of individual components of a system and the interaction of

components that actually have an impact on the system (Heizer, 1996). Another

significant benefit of a DES is that it provides a cost-effective decision-making tool,

because it permits the minimisation of risks by developers throughout, as they can

discover the correct decision before they make the wrong one. On the other hand, DES

does have some drawbacks. It can be expensive and lengthy in terms of development

and running. A DES also needs a large amount of computational time. According to

Lapin (1994), to overcome the warmup period and allow for a steady state, the model

should be run for sufficient periods.

There is extensive literature on the simulation models used for both airport

terminal modelling and performance measurement analysis. Simulation models have

often been developed to deal with operational problems, although these models

required a more detailed description and more input data from the particular system

under study (Manataki & Zografos, 2009b; Wu & Mengersen, 2013). Additionally,

simulation models require more time and operation than macroscopic models


(Manataki & Zografos, 2009a). Manataki and Zografos (2009b) suggested that most

of the existing simulation models are either too specific in one processing point or are

general simulation platforms for the integration of more than one processing unit. The

latter is required for a sufficient airport model.

2.5.1 Simulation models for passenger flow

A substantial number of simulation models have focused on the evaluation of

passenger flow. One such example of simulation modelling was presented by Eilon

and Mathewson (1973), who used an agent-based simulation model to measure

terminal facility congestion and passenger processing time. The model used

comprehensive sets of parameters, such as flight schedules, service rates, and resources

to describe passenger flow. It also contained facilitation processes and some passenger

characteristics, such as nationality. Similarly, Takakuwa and Oyama (2003) developed

a microscopic simulation model of an airport terminal in order to assess passenger

flow, with a primary focus on international departures. Their model considered the

influence of variables such as flight schedules, passenger nationality, walking speed,

volumes of bags, and passenger group size to analyse the facilitation process capacity.

The results indicated that the number of passengers missing flights could be decreased

if additional staff were added, and if the first and business class check-in counters were

used to process economy and group-class passengers.

Important efforts were made to use software packages, such as Anylogic and

ExtenSim, for modelling some elements of airport operations. For example, Savrasovs,

Medvedev, and Sincova (2009) analysed the performance of Baggage Handling

Systems (BHS) using a simulation method based on discrete event simulation

including optimisation of resources and determination of waiting times (Figure 2-3).

Curcio, Longo, Mirabelli, and Pappoff (2007) demonstrated the potential capabilities

of the Anylogic software package for simulation of the operations for processing

inbound international passengers including immigration services and other processing

elements/services (Figure 2-4). The limitation of this Anylogic model is its neglect of

the interference of the inbound and outbound passengers, for example, through the

competition for passport control service staff and facilities. Nevertheless, both studies

illustrate significant potential of both ExtendSim and Anylogic for efficiently

simulating airport and terminal operations and processes for inbound and outbound

passengers.


Figure 2 - 3: An illustration of the ExtendSim model used for simulation and optimisation of an airport baggage handling system (Savrasovs, et al. 2009).


In addition, Ma et al. (2011) provided a similar microscopic simulation model

represented by an agent-based model that mainly focused on human factors (i.e.

passenger characteristics). The proposed model was used to study the check-in

operations of passengers and their use of discretionary facilities. Similar research was

presented by Beck (2011), who designed a simulation model for the passenger flow in

a new airport terminal of Heathrow International Airport, both before and after its

opening. This paper discussed a number of factors that had to be considered for the

model, such as the characteristics of the passengers.

Figure 2 - 4: Scheme for processing the inbound international passengers using the Anylogic software

package (Curcio, et al. 2007).

Cheng (2014) developed a model using the simulation method for passenger

flow as it had become an important approach in designing and managing airports. Most

researchers have failed to take into account group dynamics when developing

pedestrian flow models. Therefore, for more realistic passenger flow conditions, this

study included group dynamics. An agent-based model was proposed due to its

feasibility and effectiveness as an approach for investigating the movement of

passengers in airports. Similarly, Fonseca et al. (2014) used a microscopic agent-based

model for the Barcelona International Airport to simulate the flow of passengers,

companions, employees, and vehicles. Specification and Description Language (SDL)


was used for the formal representation of the model. This system was used to assess

the initial airport design, and the dynamic optimisation of the terminal management

and operations. Yamada et al. (2017) modelled Japan’s Fukuoka airport’s international

terminal as a Complex Adaptive System. They constructed passenger flow simulations

dependent on the DES model. The authors concluded that it is possible to attain

simulation input data through discussions with stakeholders deploying simulation.

Hence, it is believed that it is possible to lower model uncertainty by continuously

discussing, predicting, and modelling with the stakeholders.

Another interesting approach was based on the simulation of passenger flow in

the airport environment on the basis of grid-based probabilities for non-deterministic

human motion (Schultz & Fricke, 2011). In this model, human motion was

probabilistically determined from one square element of the grid to another based on

the relative probabilities of the corresponding motions (see Figure 2-5). This figure

also shows the separate modelled paths of different individuals superimposed onto a

photograph of the actual place in the airport (Schultz & Fricke, 2011).

Figure 2 - 5: Grid element scheme for the probabilistic discrete determination of human motion from a

given position to the nearby positions on the grid (Schultz & Fricke, 2011).

Rauch and Kljajić (2006) stated that passenger flow could be defined as a

discrete stochastic process. Therefore, discrete event simulation (DES) is often used

to model such a complex system of constraints from a limited infrastructure capacity,

e.g. the airport terminal (Verbraeck & Valentin, 2002). There is extensive literature on

DES being used to analyse departing passenger flow (Guizzi et al., 2009; Novrisal,

Wahyuni, Hamani, Elmhamedi, & Soemardi, 2013; Rauch & Kljajić, 2006). Guizzi et

al. (2009) developed a simulation model aimed at predicting delays using logical and


rational management in the check-in and security checkpoint areas. They took into

account the available capacity, the volume of passengers based on time of day, and

passenger behaviour. The Rockwell Arena simulation software tool was used in this

study to determine the average queue length and waiting time. In contrast, Rauch and

Kljajić (2006) constructed their model using the General Purpose Simulation System

(GPSS), a simulation programming language. The authors analysed departure

passenger flow, from check-in through to boarding, until just before departure, in order

to identify system bottlenecks and capacity. Key factors such as passenger arrival

patterns, passenger service time, flight schedules, and operating processes were

measured. Similarly, (Novrisal et al., 2013) developed their model to analyse

congestion problems in the departure process at Soekarno-Hatta International Airport

in Indonesia. The model’s objectives were to reduce processing and waiting time in

the system. It was discovered that the number of check-in counters needed to be

increased as they had reached maximum utilisation. Maximum utilisation was reached

at approximately 61% of the total time passengers spent in the queues across the

departure process before boarding (Novrisal et al., 2013).

Manataki and Zografos (2009b) proposed a system dynamics modelling

approach as a mesoscopic model that focused on aggregate characteristics while

working at an intermediate level. Their model’s structure was built based on stock and

flow diagrams (see Figure 2-6).


Figure 2 - 6: Diagram of stock and flow (Manataki & Zografos, 2009b)


The stock was used to model the passenger facilitation process and physical

facilities. Stock refers to the state of the system, and flow refers to the rate of change

of the stock. The developed model evaluated the significant factors that affected the

flow elements. These factors were walking time, processing rate, and the number of

service counters.

2.5.2 Simulation models of security processes

Simulation models have been developed to analyse the security screening

process. In 2014, van Boekhold et al. developed their general microscopic simulation

model using ExtendSim software to assess the impact of security screening

performance, as well as the impact of pre-screening. This study aimed to express the

acceptable wait time thresholds for these processes, and to offer suggestions to

mitigate wait times by adjusting various aspects, such as approach, procedures, and

equipment. The main effectiveness measures considered in this study were average

wait time, average service time, average queue length, and average throughput rate. A

similar study was undertaken by Chitty et al. (2017), who studied flexibility when

developing initial schedules for easing the evolutionary dynamic process of re-

optimisation. The research concluded that evolutionary dynamic re-optimisation can

shorten passenger waiting times. The research used various methods to measure

flexibility, such as MaxLaneCoverage, Unopenable, and AverageShiftLength,

alongside decreasing opening hours of security lanes and waiting times for passengers.

Results demonstrated that passenger waiting times were shortened using this approach

for dynamically and static re-optimised schedules.

Dorton and Liu (2015) investigated the SSCP by applying a DES. The

independent variables of the proposed model included baggage volume carried by each

passenger, and the number of suspect bags that needed manual inspection. A simulator

for construction (SimFC) was applied to model the processes of traveller check-in and

security checkpoints at the Calgary International Airport. A case study was presented

using two different scenarios to examine and analyse flights to USA destinations. The

paper revealed a number of points, the most notable of which was that the passenger

wait times were strongly influenced by the available security (Siadat et al., 2012).

Simulation models have been used to study passenger experiences in the area of

security. Research undertaken by Kim et al. (2013) used several simulations of gate

assignments to study airport gate scheduling in order to improve passenger experience.


The data came from previous studies and a major USA airport hub. The first goal of

the study was to minimise the transit time of travellers in passenger terminals. Transit

time included the time spent between the security checkpoint and the gate, between

the gate and the baggage claim area, and between two gates. The results confirmed

that models could improve traffic flow efficiency in passenger terminals on ramps and

help develop the robustness of gate operations.

Dynamic system management has been considered in combination with a

simulation model in the area of security (Kierzkowski & Kisiel, 2015, 2016).

Kierzkowski and Kisiel (2015) developed a simulation model of airport security

screening counters to estimate the impact of the behavioural characteristics of the

operators and passengers on the reliability of the system. The model’s inputs were the

capacity of the area of systems, which included the entry area, the manual area, and

the loading area, as well as actual time of simulation and unloading time (A.

Kierzkowski & Kisiel, 2015). In their second work, Kierzkowski and Kisiel (2016)

developed a simulation model using FlexSim that aimed to improve the efficiency of

security control with respect to the level of safety. In addition, the authors developed

dynamic management algorithms for the security control system operations schedule

in an integrated approach with a simulation model. The results obtained from the

algorithms showed more improvement compared to static management. In static

management, the average time of an air traveller’s stay in the system was 9.538 min

and required 285 work hours. While in dynamic management, the average time of an

air traveller’s stay in the system was about 7 min and required nearly 162 work hours.

The input data of the proposed model includes:

Security control structures, such as counter characteristics and security

control process distribution.

Flight schedules.

The procedure of the security control process (e.g. the consequences of

triggering the walk through metal detector gate, etc.) (Artur Kierzkowski &

Kisiel, 2016).

Kierzkowski and Kisiel (2017), extended their work and applied fuzzy logic

theory for multi-criteria evaluation to study the efficiency, capacity, and level of

service of the security control system. The study focused on three elements:


Capacity of a security control counter.

Efficiency of prohibited items detection.

The passengers’ evaluation of the system.

A very recent and similar study was conducted by Skorupski and Uchroński

(2018), who presented a fuzzy inference system to evaluate overall efficiency of

prohibited items detection during baggage and passenger security screening. The

FASAS (Fuzzy Airport Security Assessment System) tool assists airport management

in security control. This model was created and deployed in a simulation experiment

to display the efficiency of a method to manage a system of security screening at the

airport (Katowice International Airport). The results revealed the performance of the

screening system can be enhanced by upgrading screening devices and by improving

training session frequency. Overall, the results demonstrate that screening

performance can be substantially improved, however, as the required performance

level improves there is a trade-off with personnel training costs and system throughput.

2.6 OPTIMISATION METHODS

To enable better decision and operational planning, significant optimisation

approaches and a mix of two models have been developed to deal with recent airport

issues (Abdoul Soukour et al., 2013; Bertsimas, Lulli, & Odoni, 2011; Dorton & Liu,

2015; Kalasky, Coffman, De Grano, & Field, 2010; Lin et al., 2015). These approaches

have also been developed to deal with the issue of process optimisation, with the aim

being minimal use of technical resources and minimum waiting time in queues (Artur

& Tomasz, 2017; Bevilacqua & Ciarapica, 2010; Roanes-Lozano, Laita, & Roanes-

Macıas, 2004; Solak et al., 2009).

Research looking at the combination of two or more approaches was undertaken

by Ju, Wang, and Che (2007), who presented a combination of simulation and

optimisation methods. A simulation model was developed to understand the current

bottlenecks in related terminal operations and their causes. It was also used to evaluate

the key performance measures of the processing units. The authors used the

optimisation technique to reassign different resources. Du et al. (2015) presented a

distribution optimisation model to decrease departure delays and alleviate congestion

in transit stations caused by the unsteadiness of the distribution of air travellers in

transit modes. The main purpose of the model was to minimise the average departure


time. A genetic algorithm was used to solve this problem. Solak et al. (2009)

considered terminal operations to be a network system and used a multistage stochastic

integer linear programing model to determine the optimal capacity, taking into account

optimal future expansion and desired LOS. The main objective was to minimise the

maximum delay of each passageway and processing station by considering the

variation in demand as a significant constraint.

Regarding the security issue, Dorton and Liu (2015) described a mix of two

models, including a queuing network and a DES for the security screening system.

These models aimed to analyse the main external factors that influenced SSCP

operation efficiency with respect to both system dependent measures, i.e. throughput

and cycle time. SSCP throughput can be defined as the number of passengers that have

arrived in the system and exited from the system during a one hour interval, while the

cycle time is defined as the amount of time spent in the system. Dorton and Liu’s

independent variables included baggage volume carried by each passenger, and the

number of suspect bags that needed manual inspection.

The issue of staff scheduling is widely researched in operational research. An

early example of research on staff scheduling was carried out by Mason, Ryan, and

Panton (1998), they provided information about optimisation and simulation-based

systems for staff rostering of customs staff at the Auckland International Airport. The

development of an integrated approach deploying simulation, integer programming,

and heuristic descent methods is established to identify new-optimal levels of staffing.

The staffing needs are deployed as input to an integer programming model that

distributes part-time and full-time personnel to every period of the working day. The

authors used a simulation model as the basis of their research. Numerous methods,

involving several heuristic procedures, are described for developing personnel

schedules. The results concluded that these techniques lowered staffing levels, created

good quality rosters and ensured passenger processing targets were met. These

established optimisation and modelling tools are now utilised on a daily basis to

produce new rosters. In the same vein, Andreatta et al. (2014) developed heuristics

algorithms to allocate personnel to each ground handling process of an airport. Their

study divided resources into different sets assigned to the ground handling processes

of a particular flight.


Check-in areas have attracted a lot of attention during the development of

optimisation and hybrid models. For example, Hsu, Chao, and Shih (2012) explored

the dynamic distribution of check-in facilities and dynamic allocation of passengers to

reduce total wait time and better use of facilities. The developed model was

implemented at the Taoyuan International Airport. The results demonstrated that

dynamic distribution of check-in facilities can lower waiting times and improve

service counter use rates. These benefits can be improved through dynamic

passengers’ assignment. Two criteria, service counter utilisation rate and waiting time,

were adopted as indicators for required adjustments for distribution of facilities. The

literature on the problem of check-in facilities also concerns approaches to attaining

more effective check-in procedures. The theories previously applied were queuing

theory, dynamic and integer planning, experimental designs, and system simulation.

Moreover, use of the developed assignment model and dynamic allocation model for

air cargo services, as well as other sectors, to attain higher effectiveness and better use

of facilities, requires further research. Similarly, Xin, Lin, Huang, Cheng, and Chong

Teo (2014) used linear programming to determine the optimal number of check-in

counters over a specified period. They justified their approach by demonstrating

improved effectiveness in human resource utilisation, which resulted in fewer counters

and fewer working hours.

Parlar, Rodrigues, and Sharafali (2013) extend the research further by

developing a stochastic model to compare the impacts of static and dynamic policies.

Their model aims to lower total operating cost of counters while satisfying the needs

of airlines and airport authorities. The approach is simple and easy to implement. The

provided model can effortlessly accommodate realistically sized issues with numerous

passengers. This was justified by comparing the performance of dynamic and static

policies using numerical experiments. The study found that static policy must be

selected over the dynamic policy if the number of passengers is lower than 50 and the

static policy cost is lower than aM +V1(0,0) or if passenger number is above 50. The

static policy makes it effortless to find an optimal counter number when the number

of booked passengers is in the hundreds.

Based on the literature, a very common optimization problem in the apron area

is gate assignment which aims to minimise passenger walking distances from both

check-in to gate and from gate to the baggage claim area. Genç, Erol, Eksin, Berber,


and Güleryüz (2012) employed a hybrid of heuristic and stochastic approaches to

minimise the total duration of un-gated flight. This can be achieved by reducing the

total distance that all passengers walk including:

Connection flight travelling distance

The maximum distance that a passenger needs to walk in total

Similarly, Ding, Lim, Rodrigues, and Zhu (2005) used the Tabu search method

to investigate the problem when the number of aircraft exceeds the number of available

gates. The two main objective functions of this work were to minimise the number of

un-gated aircraft and to minimise total walking distance.

Barnhart, Fearing, and Vaze (2014) developed a multinomial logit model to

address the issue of disaggregate passenger itinerary flows using a small set of

propriety booking data. The authors developed a simplified regression-based approach

for estimating passenger delays. Passenger delays can be caused by three main factors

including flight delays, flight cancellations and missed connections. Delays are caused

by the following factors:

Distribution of flight load factors

Distribution of daily average load factors

Distribution average flight load factors by day of week and time of day

Distribution of percentage of connecting passengers

Distribution of connection time for one-stop passengers.

Recently, Jacquillat and Odoni (2015) developed an approach to interface

tactical capacity utilization which meant optimizing the utilization of airport resources

to process flights over a day. A strategic queuing model of airport congestion was

proposed, which meant planning flight schedules well before the day of operation and

taking into consideration long-term patterns of capacity availability. The main

objective of their study was to model the relationships between flight schedules, airport

capacity and flight delays at the strategic level through understanding of how flights

will be operated at the tactical level. Thus, the authors examined delays over the course

of day as a function of arrival and departure service rates, by means of a stochastic and

dynamic queuing model. The authors also formulated arrival and departure service

rates as a function of flight schedules, operating conditions and observed queue


lengths. The purpose of this model of airport congestion was to quantify the magnitude

of delays and their evolution over a day as a function of flight schedules and airport

capacity. The model is strategic: it uses information that is available before a day of

operations. It may then be used to test the impact of changes in flight schedules or in

airport capacity on flight delays in support of airport congestion mitigation and airline

scheduling. Factors included were weather conditions, landing and take-off ratios and

runway configurations in use.

The research by Mujica (2015) is noteworthy because of the development of a

mix of two models to satisfy the different mandatory restricted policies relating to

airport terminal processing units, such as opening or closing check-in counters for each

flight, check-in starting time, and load balance. In the first phase, an evolutionary

approach was used to improve the initial allocation of check-in counters, taking into

account the policy restrictions. Once the best solutions were found, the author designed

a simulation model to determine which allocation was the most efficient in real life

situations, taking into account significant factors, such as the profiles of the travellers.

Similarly, Yan, Tang, and Chen (2014) addressed the perturbations of the check-in

process in airports by developing a zero-one integer programming model. The types

of perturbations considered in this study, however, occurred due to temporary airport

incidents, such as an airport closure, the crashing of the counter computers, and

temporary power failures. Yan et al. (2014) proposed solution methods that could be

used in the real world to resolve check-in counter reassignment problems. The

variables used were the length of time window and the number of service lines.

An optimisation issue in relation to staff scheduling problems in security services

was resolved by Abdoul Soukour et al. (2013) using a memetic algorithm (MA) with

concepts of an evolutionary algorithm and local search techniques. The algorithm

performs days-off scheduling, shift scheduling, and staff assignment together with all

specifications. Sigurðardóttir (2011) designed a mathematical model using mixed

integer programming, which provided a feasible solution for irregular staff schedules.

The model was tested for three different types of employee datasets, which were solved

using a local search algorithm. The model was able to serve under the conditions of

multiple and changing objectives and goals of staff scheduling and was flexible enough

to handle all the constraints and requirements of staff scheduling in terms of shift

length and shift start time.


At Aarhus Airport, a heuristic method was used by Jensen (2015) to design an

algorithm that creates a staff schedule in a strategic way. The proposed algorithm was

designed for use with specific rules and regulation working hours. It was used to

provide the minimum number of handlers needed to fulfil the demands of each period.

The algorithm was tested to give the actual demand instead of a demand estimate

derived from existing schedules. Tang, Alam, Abbass, and Lokan (2009) presented a

multi-objective constrained resource allocation problem where the first objective was

to maximise the quality of service, and the second objective was to minimise the total

cost. A genetic algorithm was presented to allocate resources among the different

objects in an airport.

For the security domain, Skorupski and Uchroński (2018) presented basic

models that explain passengers with numerous strategies, security personnel with

various jobs and concerned queuing structures. Prior to formal modelling, it first

gathered actual data from Shenyang Taoxian International Airport (STIA). The authors

performed simulation analysis by distributing service times and arrival time intervals,

using three kinds of staff and compared the performance of various network structures.

The paper offers findings regarding establishing the satisfying structure of network

queuing for airport security check-points. It produced a new result demonstrating that

a blend of n M/M/1 systems will perform better than or equal to a M/M/n system when

considering feelings and strategies of passengers. A summary of the models used to

address airport optimisation problems is presented in Table 2-2.


Table 2 - 3: Summary of models used to address airport problems

Modelling

approach

Topic Factors Performance metric Measurement Papers Case Study/ Validation

DES models Passenger flow

analysis

Identifying

bottlenecks

Security system

performance

Flight schedule

Demand fluctuations

Complexity of the system

Baggage volume

Alarm rate

Processing time

Waiting time

Average queue

Processing points utilisation rate

Number of opening counters

Throughput and PAX cycle time

Alodhaibi, Burdett, &

Yarlagadda (2017);

Dorton & Liu (2015);

Gronfula (2014); Guizzi

et al. (2009);

Kierzkowski & Kisiel

(2015); Novrisal et al.,

(2013)

Soekarno-Hatta International

(SHI) Airport.

Naples International Airport

Wroclaw Airport

Microscopic

simulation model

Dynamic

management

Security issues

Resource allocation

Analysing passenger

flow in new terminal

Flight schedule

Human behaviour

Walking speed

Availability of workers

Available capacity

Intensity of passenger flows

Similar to above

Average speed

Average duration of a

passenger’s stay in the system

Number of operator work hours

Beck, (2011);

Kierzkowski & Kisiel,

(2015, 2016); Schultz &

Fricke (2011); Siadat et

al. (2012)

Dresden Airport

Calgary International Airport

Heathrow Terminal 5

Wroclaw Airport

Simulation

System Dynamic

Performance

Evaluation

Passenger flow

Passenger transfer rates

Passenger arrival distribution

Capacity, delay, resource

utilisation,

Level of Service

Manataki & Zografos

(2009b, 2010)

Athens International Airport

Agent based

model

Passenger flow

Group dynamic

Passenger group dynamics

Passenger behaviour

Walking speed direction

Dwell time

Level of service (LOS)

Space Utilisation

Cheng, (2014); Cheng,

V. Reddy, et al. (2014);

Ma (2013); Ma, Fookes,

Kleinschmidt, &

Yarlagadda (2012); Ma

et al. (2011)

Brisbane International Airport

Queuing theory Interacting passenger

flow

Process delays

Performance measurement

Delay cost

Cycle time

Throughput per unit of time

Dorton & Liu, (2015);

Hsu et al. (2014);

Jacquillat (2012);

Jacquillat & Odoni

(2015)

Taoyuan International Airport

John F Kennedy Airport

Fuzzy logic

application

Control passenger

flow

Security efficiency

Efficiency evaluation

Flight Number

Available capacity

Number of available resources

Number of PAX handled

(Cheng, Mu, Zhang, &

Zhang (2014); Artur

Kierzkowski & Kisiel,

(2017); Mu et al.,

Wroclaw Airport

Calgary International Airport


(2014); Skorupski &

Uchroński (2015, 2016)

Optimisation

techniques

Integer

Programming

Mixed integer

nonlinear

Program

Heuristics

Staff allocation

Airport facilities

planning

Efficient traffic flow

Delay approximation

Workforce demand

Passenger load factor

Fluctuated service demand

Flight schedule

Airport size

Flight schedule

Passenger arrival rate

Day-off scheduling

Staff scheduling

Staff assignment

Capacity utilization

Delay levels

Similar to above

Average arrival rate

Maximum flow rate

Abdoul Soukour et al.

(2013); Lin et al. (2015)

Sun & Schonfeld (2016)

Kim et al. (2013); Rodič

& Baggia (2017); Solak

et al. (2009)

----------------

---------------

Hartsfield Jackson Atlanta

International Airport

US hub airport


2.7 SUMMARY OF THE REVIEWED LITERATURE

This chapter reviews the current problems that prohibit airports from efficiently

performing their operations. Passenger flow analysis and control has been the focus of

many publications to address the related issues that have a subsequent impact on

passenger flow, such as limited capacity, efficiency of airport systems, and congestion

problems. The second critical issue that challenges airport authorities is security which

results in an extensive literature regarding the security check process, the capacity of

security screening areas, and dynamic system management. Staff allocation problems

is another issue that has been studied over the past few years, taking into account staff

assignment, workforce demand, and days on-off scheduling.

The literature discussed different types of models that are used to understand and

resolve the recognised problems. These models can be categorised as analytical,

simulation, optimisation, and hybrid models providing decision support capabilities at

all levels of detail: from macroscopic, through mesoscopic, to microscopic. A

macroscopic approach can be used for capacity planning because it can provide

‘‘approximate answers to planning (primarily) and some design issues, with emphasis

on assessing the relative performance of a wide range of alternatives’’(De Neufville et

al., 2013). Additionally, macroscopic and mesoscopic approaches have been employed

to evaluate the effect of instance scheduling and resource allocation on performance

metrics such as queue length and waiting time. Microscopic models have been used to

simulate individuals’ interactions at a higher level of details. Wu and Mengersen

(2013) argued that macroscopic models are insufficient to handle the variability,

complexity and stochastic nature of airport terminals, but microscopic approaches are

difficult to deal with because they require large amounts of data for the high level of

detail.

The models were built based on a comprehensive set of parameters that

characterise the issues, such as flight schedules, processing time, service rates

distribution and number of resources, the facilitation process, and associated passenger

characteristics (e.g. nationality, as it influences which customs lane the passenger can

use).


2.8 KNOWLEDGE GAP IDENTIFIED

Despite the great efforts undertaken in modelling and simulating the issues of

passenger flows, security and staff allocations, there are significant identifiable gaps

in the current knowledge of airport operations. First, while there is considerable

research aimed at modelling and simulating passenger flows at airports, there is limited

research investigating the impacts of different arrival patterns within airports. Second,

there has been little research on development of holistic models that provide an

integrated view of the processes and sub-processes of the whole airport which help

with analysis and evaluation of the various measures of the efficiency of the airport.

According to Zografos et al. (2013), most of the recent tools and models are only

focused on the individual process and address fragmented sections of the decision

making procedures of airports. Third, there have been few attempts to facilitate the

possible integration of airport outbound and inbound processes, including the potential

of the incoming passengers to draw significant personnel resources. Therefore, these

knowledge gaps were formulated as the research questions explained in section 1.2.

2.9 FORMULATION OF RESEARCH SCOPE AND RESEARCH

CONTRIBUTIONS

The motivation for this work was to develop a model capable of studying

passenger flows and staffing requirements at international airport terminals as a single

unit by facilitating the integration of outbound and inbound systems. More

specifically, this would involve:

Development of a simulation framework for outbound passenger flow using

ExtendSim V9.2 simulator software.

Investigating the effect of arrival patterns of departing passengers on the

departure terminal operations.

Development of advanced resource management algorithms to integrate

with the simulation model, including both outbound and inbound processes.

Development of an analytical optimisation framework to perform capacity

planning for strategic planning as the simulation model can only be used at

the operational planning level.

Chapter 3: Simulation Model Framework for the Outbound Passenger Processes at an International Airport 43

Chapter 3: Simulation Model Framework

for the Outbound Passenger

Processes at an International

Airport

3.1 OVERVIEW

In the previous chapter, the background information regarding passenger flow

modelling was presented. To achieve the research aim of developing an overarching

model for the two systems (inbound and outbound), a comprehensive review of the

available literature in the fields of modelling airport terminal operations and

performance evaluation is required. This chapter introduces a generic framework for

an integrated simulation model of departing passenger flows as the first stage of the

overarching model. Figure 3-1 illustrates the common layout of an international airport

terminal and focuses on the standard outbound processes, such as check-in, security,

immigration, and boarding, and the standard inbound processes, including

disembarking, baggage claim, immigration, and quarantine. The first step towards the

development of the final model, however, is a simplified scheme consisting of only

the left-hand side of Figure 3-1 (i.e. without the interaction with the inbound

passengers).

The main objective of this research was to develop a model that can accurately

identify bottlenecks and improve operational efficiency. This model can also evaluate

the effect of an increasing number of passengers on the terminal facilities, a factor that

has made airport systems much more complicated. As a result of this rapid growth in

the number of air travellers and the complexity involved, numerous regulations and

new technologies are being applied to airport operations (Ma, 2013). For example,

flight schedules are frequently changed due to irregular demand. Therefore, a

simulation has been selected as the desirable approach to fully understand the complex

system of an airport. In addition, a wide range of what-if scenarios can be explored

throughout the model to assist in more effective decision-making during airport

terminal operations’ planning, design, and management.

44 Chapter 3: Simulation Model Framework for the Outbound Passenger Processes at an International Airport

Figure 3 - 1: Overview of an airport’s terminal processes, including outbound and inbound processes

3.2 THE CONCEPTUAL FRAMEWORK

In this section, a generic framework for modelling the flow of passengers

through the airport terminal processes is introduced (see Figure 3-2). Each system has

its own particular flow and each system requires a different infrastructure and services.

The development of a model for an airport comprises the following tasks. The first is

the development of a general international terminal system for the outbound traffic

system. The second task is the evaluation of the demand and supply of an airport

terminal. In this step, the entities of the general international terminal system are

established and the characteristics and capabilities of the model are evaluated. The

required information in this step are flight schedules, the structure of the terminal, and

implementation information.


Figure 3 - 2: Airport system model.

The third task comprises the development of an algorithm approach for

managing the resources of the airport system and the fourth task involves validating

and checking this model. The models require validation to ensure their consistency and

credibility and both factors are equally important for any model. The fifth and final

task is the application of the model to various scenarios for further analysis. The model

can be successfully applied to different scenarios and cases related to an airport

terminal, such as arrival distributions, processing time distributions, and the different

internal and external structures of an international airport terminal (Chiu & Walton,

2002). During the fifth step, Key Performance Indicators (KPIs) are monitored as part

of the application of the above scenarios. The KPIs for this model include the average

waiting time, maximum waiting time, average queue length, and maximum queue

length. The basic international airport system model is depicted in Figure 3-2.

As mentioned above, airport systems are extremely complex, therefore, a multi-

disciplinary approach was utilised to understand all the flows and associated features.

The application of simulation techniques is highly beneficial for airport operations

management, especially for an airport’s own standard processes. To obtain a unique

model, the simulation of the system must be constructed according to the specifications

of the specific model that is being studied. The proposed model can determine

bottlenecks in the system and alternative procedures can then be attempted to optimise

the system without negatively impacting it.


3.3 PASSENGER FLOW CHARACTERISTICS

The passengers in an airport terminal can be divided into three types based on

how they are handled inside the terminal: departing passengers, arriving passengers,

and transferring or transiting passengers. Each type of passenger behaves differently

according to why they are using the airport’s facilities. An Australian international

airport terminal has been used as an example to demonstrate a common layout of

airport terminals. According to Ma (2013) definition, departing passengers start with

the check-in process in the international terminal and transfer to their airplane within

the same terminal. These passengers arrive at the airport terminal according to their

flight schedule and normally arrive at least two hours before their departure time.

Departing passengers complete the three main processes: check-in, security screening,

and immigration, and then wait to board the airplane at their specified gate. In contrast,

transferring passengers merely pass through the security screening control and then go

directly to their specified boarding gate as shown in figure 3-3.

Figure 3 - 3: Airport outbound processes (Shuchi, 2016).

The other type of passenger flow is that of arriving passengers, who can be

defined as people who disembark from the aircraft after landing at the airport terminal.

The difference between the process that arriving passengers go through and that for

departing passengers is that the latter is more complex. It involves services provided

to transit passengers and the time taken to complete this process is often significantly

longer than that of the arrival process (Odoni & de Neufville, 1992). Inbound

passenger flow is smoother than outbound passenger flow, although delays can occur

Arrival to the Terminal

Check-inSecurity

ScreeningImmigration and Custom

Boarding


in the inbound system depending on the time it takes to deliver baggage from the

aircraft to the baggage claim area (Ma, 2013).

The current study is predominantly concerned with the airport operations

associated with departing (outbound) passengers. Inbound passenger flow has the

potential to significantly affect the operations concerning outbound passengers. For

example, airport resources may be allocated to services for inbound passengers, e.g. in

the customs area. As a result, additional delays and bottlenecks may affect outbound

passengers because fewer personnel are available to process them.

3.4 OUTBOUND PROCESSES MODELLING

Flight attributes is one of the most vital components of model simulation and

refers to the information necessary for the establishment of the outbound system. There

are three divisions in flight attributes: flight schedules, passenger characteristics, and

boarding characteristics (see Figure 3-4). Outbound passenger attributes can be

generated via the following procedures. The first step involves using flight attributes

to consider the related information, such as flight schedules, check-in types, arrival

methods, and travel class (see Figure 3-4) and storing this in an Excel spreadsheet.

The second step involves developing an algorithm to generate the departing

passengers’ attributes. This is can be done using the Excel programming language

Visual Basic for Applications (VBA). Flight attributes are used as an input for this

algorithm as illustrated in Figure 3-5. The next section discusses the process of arriving

at the airport.


Figure 3 - 4: The input modelling of an outbound simulation model.


Figure 3 - 5: Flowchart for generating outbound passenger attributes.

3.4.1 Arrival at the terminal

Depending on the purpose of travel, airport arrivals are divided into business and

leisure passengers in the developed model. Business passengers are those who are

traveling for work, for instance for a meeting or conference, whereas leisure

passengers are those who are traveling for a holiday or to visit family and friends.

Arrival patterns can be influenced by these two types of passengers (Cheng, 2014;

Manataki & Zografos, 2009b), for example, business travellers tend to use airlines

more frequently and are therefore more familiar with the workings of an airport

terminal and the reliability of the access mode (Ashford, Mumayiz, & Wright, 2011).

The main factor that influences passengers’ time of arrival at an airport is their flight

schedule, which varies from day to day and according to the day of the week.

Robertson, Shrader, Pendergraft, Johnson, and Silbert (2002) stated that depending on

the season, there may be more flights scheduled, e.g. in summer.


Passenger arrival behaviour can be captured by showing the distribution of

passenger arrival times. The distribution shows the number of air passengers and the

time at which they arrive at an airport prior to the departure of a scheduled flight

(Kamyszek, 2014; van Boekhold et al., 2014). The flow rate of passengers arriving for

an international flight provides a universal arrival pattern. In accordance with the

example provided by Ashford et al. (2011), for the accumulative arrivals of passengers

before the scheduled time of departure, all passengers had arrived by one hour before

the scheduled departure time of an international flight. For local flights, however, all

passengers had arrived by 20 minutes before the flight was scheduled to depart (see

Figure 3-6).

Figure 3 - 6: The relationship between departing passengers’ arrival times and the type of flight

(Ashford et al. 2011).

There are many factors that affect an airport’s arrival patterns, e.g. airport ground

access, security issues, and the situation regarding traffic and transportation modes

(Ashford et al., 2011; Manataki & Zografos, 2010; Stefanik, Kandera, & Badanik,

2012). Furthermore, the arrival pattern of an airport will differ at different times of the

day (Stefanik et al., 2012). Rauch and Kljajić (2006) noted that passengers with early

flights generally arrive later than the statistical average. Many scholars hold the view

that the arrival behaviour of international passengers is common for most airports.

Firstly, almost all air travellers arrive one hour before the scheduled departure time of

an international flight. Secondly, the peak hours of the check-in procedure for each


flight normally occur between 100 and 120 minutes before the scheduled departure

time of a flight. Thirdly, business passengers arrive later than leisure passengers.

Lastly, the peak hours earlier on in the day are busier and shorter than the peak hours

in the evening and afternoon (Ashford et al., 2011; Cheng, 2014; Chiang, 2011; Chiu,

2002; Stefanik et al., 2012). Passengers arrive at an international airport terminal via

different transportation modes, e.g. trains, public buses, taxis, and private cars (which

they park at the airport for either a short or long term) (Manataki & Zografos, 2009b).

The model’s input will be discussed in the sections that follow.

3.4.2 Check-in module

In this section, the check-in modelling processes is explained in more detail,

including the check-in characteristics of passengers and the physical environment.

Figure 3-7 shows the main processes used in the simulation. The simulation model

considers three ways that the passengers can check-in. The first is the traditional way,

where the passengers are processed at check-in counters. This method can be divided

into two types: specific check-in and common check-in.

Common check-in refers to counters that can be used by different flights for the

same or different airlines, while specific check-in refers to the counters that can be

used by different flights for just one airline. Another type of check-in is self-service,

which consists of both types of auto check-in: kiosks and online check-in. Most airlines

provide an online check-in service to reduce the processing time for passengers.


Figure 3 - 7: Flowchart of check-in processing at international airport.

The model also considers bag-drop facilities for passengers who checked in

online or for those who check in using self-service kiosks. Delay times at the check-in

area are influenced by significant factors, such as the number of bags for each

passenger. It is assumed that the distribution of the number of bags for each passenger

is uniform (0, 2) (Ma, 2013). The delay time that can occur during check-in can be

calculated by the formula (Park & Ahn, 2003):

𝐷𝑒𝑙𝑎𝑦 𝑡𝑖𝑚𝑒 𝑎𝑡 𝑐ℎ𝑒𝑐𝑘 𝑐𝑜𝑢𝑛𝑡𝑒𝑟𝑠 = 0.2 𝑚𝑖𝑛 ∗ # 𝑜𝑓 𝑏𝑎𝑔𝑠 (1)

The structure of the check-in module is built based on a hierarchical model,

starting with a high level of check-in processes and moving down to sub-processes and

sub-sub processes (see Figure 3-8).


Figure 3 - 8: Module hierarchy of check-in system.

Passengers are advised to be at an international airport 2.5 hours before their

departure time and check-in closes 25-30 minutes prior to the flight’s departure

(Cheng, 2014). In the model, self-service check-in was built in the area located close

to the check-in counters. Sets of check-in counters were then created to serve either an

airline or a group of airlines. These sets consist of a row of 20 counters for check-in

services. Among the 20 counters, there are three counters for business class and

seventeen for economy passengers. The model is dynamic over a set time and can run

various scenarios, such as number of check-in counters needed, which can be

controlled by the input data represented by assigned. The major elements and

processing facilities of the check-in module are summarised in Table 3-1

The above flowchart maps out the data entities and relationships for a check-in

system and then translates this into an ExtendSim model that involves the following

options:

Passengers who check in online and in the terminal.

Passengers with and without baggage.

Economy and business passengers (with business priority).


Passengers with and without baggage problems to be fixed during the check-

in process.

Passengers who only need to drop their baggage off after checking in online.

Table 3 - 1: Summaries of major elements and processing facilities of check-in module.

Elements and processing facilities Values

Check-in type

0 = online check-in

1 = airport counters

2 = self-service

Distribution of number of bags

Uniform (0,2)

Percentage of business class passengers 15%

Percentage of economy class passengers 85% (Cheng, 2014)

3.4.3 Security screening module

In security screening, the main activities are x-rays and carry-on luggage scans.

Passengers place their belongings, including laptops, liquids, and personal items into

separate trays with their take-on bags for scanning. All passengers and their carry-on

bags must be checked (see Figure 3-9). Each passenger joins the security checking

queue and when they reach the head of the queue, they are guided by security staff to

an available counter. Items are placed onto the x-ray machine and passengers walk

through the metal detector. Passengers and carry-on bags must successfully pass the

metal detector and x-ray examination, respectively. There is the possibility of failures

for both passengers and bags, however. If a passenger fails the metal detector check,

they are asked to go through the metal detector again or are selected for a body check.

The required data related to time statistics was collected (Philip J Kirk, 2013a),

such as time spent at security screening processes. On average, the time spent at

security screening is between six and seven minutes and 10% of all passengers fail

when they first walk through the metal detector. Assumption was made that 15% of

the total passengers have been chosen arbitrarily for the random explosive trace check.


Figure 3 - 9: Flowchart of screening checkpoints for processing at the international airport.

The elements of security screening checkpoints are consistent from one

checkpoint to another. These elements are classified as static in nature, such as the

footprint required for a WTMD, TRX, AT, or WBI. Other elements, such as the

passenger queue and composure area, are variable because the characteristics change

based on the space available, airline passenger load factors, and the number of

passengers screened at checkpoints per hour (Transportarion Security Administration,

2009).

3.4.4 Immigration module

The immigration process is the third stage of the departure process and occurs after

the security control process. Figure 3-10 illustrates the immigration operation


processes considering the two methods of document checking-in at Australian

international airports: Smart Gate and immigration counters. An assumption is made

that 30% of passengers used Smart Gate because of the strict rules that have been

enacted by the Australian Border Force (Australian Government, n.p. ). The rules are:

The age of the passenger must be 16 years or older.

Travellers must be from one of the following countries: Australia, Canada,

China, France, Hong Kong, Ireland, Japan, Korea, Macau, New Zealand,

Singapore, Sweden, Switzerland, Taiwan, United Kingdom, or United States

of America

Figure 3 - 10: Flowchart of immigration system at international airport.

The other passengers are manually processed at immigration counters. Like the

other standard processes, check-in and security, passengers join the queue before being

processed by personnel from immigration. In the model, the number of available desks

will be set up based on demand. Peak hours require more available desks to meet

passenger satisfaction. If the queue length exceeds the configured threshold value, an


additional desk is opened, i.e. to open one more desk the number of passengers in the

queue must have increased by 40.

3.4.5 Boarding procedure module

The last stage of the outbound process is boarding passengers onto an airplane

and the airplane leaving the airport. Airlines are responsible for boarding passengers

when the aircraft is ready to depart (Shuchi, 2016). Normally, gates are open 30

minutes before the departure time and passengers should present their boarding pass

and passport to be checked by airline staff at the gate.

In this module, an algorithm was developed to manage the boarding procedures

for all gates. ExtendSim’s C-based language ‘ModL’ was used to code the algorithm.

Each item going in represents a passenger arriving at the gate who immediately passes

through the block and out of the model. To code such an algorithm, it is very important

to understand the boarding procedures and declare the associated variables, including

constant and static variables. Another vital component of flight attribution is boarding

characteristics. The main elements of this component are boarding time, boarding

strategy, and gate number. We created custom blocks for managing the boarding

procedures for all gates. The passenger is tracked within the database global array

called 'boarding'. The counter in the value receives a value every minute so that

operations can be performed to track the queues and boarding status of each passenger

for all gates. For this research, there are two boarding strategies:

The nature of the first boarding strategy is unconstrained. In this strategy, it

is assumed that 20 people enter a plane within 60 seconds.

The nature of second boarding strategy is random. In this strategy, it is

assumed that 15 people enter a plane within 60 seconds.

3.5 EXTENDSIM MODELS FOR OUTBOUND PROCESSES

In this study, discrete event simulations of passenger flow and queuing in

different airport facilities (including between operational sections) were implemented

using the simulation software package ExtendSim. It helps developers to connect

blocks together in order to move items though the system from the beginning of the

processes until the item exits the system (Diefenbach, 2010). This software package

facilitates the simulation of passenger flow in the presence of multiple facilities within


an airport. This assists with the identification of any existing or expected bottlenecks

and passenger processing times, as well as with the optimisation of operational

capabilities and identifying the personnel resources required to minimise processing

times. In addition, this simulation package has the ability to program the assumptions

of a system via a variety of blocks and connections. The simulated arrivals into the

airport terminal can be generated in relation to the distribution that counterparts show

in real-life.

An excellent feature of ExtendSim is its ability to build models with menus of

blocks. These blocks can be pulled onto the model page and linked to other blocks and

their function is to control both the attributes and flow of items in the model. Another

significant feature of ExtendSim is that the developer can modify and even create new

blocks for the user using the ‘ModL’ language to attend to specific cases within the

model. The blocks are activated when entities pass through them, demonstrating the

operations that are performed in the system (Diefenbach, 2010).

The viability of this approach for airport simulation has been confirmed by the

large number of previous publications that have used this software package or a similar

software package to analyse and simulate various airport operations, including the use

of ExtendSim for the simulation and optimisation of airport baggage handling

Savrasovs et al. (2009) and security screening simulations van Boekhold et al. (2014),

the use of the Arena package for the DES of passenger flow in a terminal Guizzi et al.

(2009), the use of the SES/Workbench package for the optimisation of airport

operations Saffarzadeh & Braaksma, (2000), the use of the General Purpose

Simulation System for the simulation of passenger flow and airport capacity Rauch &

Kljajic, (2006), and the use of the AnyLogic package for the simulation of passenger

flows and security issues (Curcio et al., 2007).

At the same time, it has been argued that the real capabilities of the ExtendSim

software package for discrete event and discrete rate simulations and the optimisation

of airport operation processes are far from being fully explored. This is demonstrated

by the significantly more extensive and successful use of this package to simulate and

optimise the even more complex operational processes of hospital emergency

departments and other facilities (Diefenbach & Kozan, 2011; Diefenbach, 2010;

Dorokhov, 2009; Williams, Chambers, Dada, McLeod, & Ulatowski, 2014).


The output of the ExtendSim model includes the total number of passengers

processed by particular workstations and checkpoints in the airport terminal, the

number of staff involved in each operation or service, the number of passengers

currently (and on average) queueing for a particular operation or service, and the wait

time and processing time for all passengers (including the average and maximum wait

time and processing time). This output enables the identification of bottlenecks and

weak operational points that may cause major processing delays.

3.5.1 Hierarchical blocks

The models were built based on a series of hierarchical blocks, including the

arrival of passengers, check-in, security screening, immigration and customs, and

boarding. These hierarchy blocks contain a number of blocks that each have a different

functionality, including blocks that simulate the steps in a procedure (Queue, Activity),

blocks that perform a calculation (Math, Random Number), blocks that store data or

interface with other applications (Read, Write), and blocks that plot the result of the

model (Plotter, Histogram). There are also some tools for interface formation (Popup,

Buttons). The hierarchical blocks that are created can be stored in ExtendSim’s library

and reused again in the same or a different model.

3.5.2 ExtendSim modules description

As explained in the section above, the structure of the model is built around the

basis of a hierarchical model structure. In this context, the proposed model is organised

into two hierarchical levels:

1) The first level of the hierarchy reflects the airport departure system broken

down into a set of the main departure procedures, including check-in,

security, immigration, and boarding.

2) The second level describes the intricate details of the different sub-processes

in the airport terminal. Specifically, the main departure procedures that the

airport terminal model consists of are:

arrival characteristics, including the distribution of arrivals, method of

arrival (car, bus, or train), number of bags, class of travel, and time of travel;


the check-in process, including the type of check-in, e.g. at the kiosk or

online with a bag, business or economy, and the assignment of each flight

to specific check-in counters;

security screening, including x-ray checks conducted in the common

security screening line, x-ray checks for diplomats, and secondary screen

checks (i.e. random checks);

immigration processing, including Smart Gate services and the manned

counter for passport control; and

boarding procedures, including boarding time, waiting time at the gate,

boarding strategy, jetway capacity, and flight capacity.

The simulation involves eight blocks that demonstrate the actual processes in

Australian international airports. Each block is explained in more detail in the

following sections. First, creating the entities in the model representing departing

passengers. These passengers are generated following the normal distribution

according to flight schedules. The data were stored in the Excel spreadsheet and then

linked within ExtendSim by the module called create (see Figure 3-12).

Figure 3 - 12: Input data represented by passenger attributes.


Secondly, the values of the attributes of each passenger are defined in the

ExtendSim schedule table. Every passenger created as a simulation entity is tagged

with a unique attribute, such as arrival time, check-in group, and arrival method. Each

row of the table introduced above refers to Extendsim entities (passengers), including

the related attributes of the outbound processes. Some passengers arrive late to the

airport and are assigned high priority, meaning they can skip the queue if the queue

time is greater that the time until check-in closes.

Block 3 introduces the overview of the check-in system, starting with kiosk

check-in and followed by a hierarchy of blocks of check-in groups. Block 4 simulates

the check-in counters for both business and economy passengers. For the security

screening lanes, each lane includes x-ray machines to scan and check if there are any

dangerous or prohibited items; these are demonstrated in block 5. Block 6 simulates

the random checks where some passengers are selected randomly for further checks.

Block 7 simulates the immigration counters, including the SmartGate option. Finally,

passengers complete the outbound processes and are directed to the boarding

procedures located in block 8 that allow the passengers to exit the system.

Block 1: High level of outbound processes

The first block, ‘Block 1’, presents the outbound processes of international

airports and is hierarchical to simplify the complex processes that contain processes

and sub-processes (see Figure 3-13). For example, as explained in section 3.4.1, the

check-in system consists of three levels of processes: check-in groups for individual

or multiple airlines, check-in type, and check-in counters. The yellow blocks at the top

of Figure 3-13 refer to the input data of the model. ExtendSim data can be classified

into three groups:

Passenger attributes, such as SmartGate users, etc.

Operational data, such as average processing time, number of available

resources, etc.

Flight attributes and related information, including gate number, departure

time, boarding time, etc.


Figure 3 - 13: Block 1: ExtendSim simulation for outbound system.


Block 2: Passengers with high priority

Block 2 consists of two modules. The first module creates passenger entities

using the schedule, while the second module aims to identify passengers who are

waiting at a common check-in queue and may miss their flight. Some flights might

share the same check-in counters because they are from the same airline. Therefore,

an algorithm was developed to allow passengers to skip the queue if the estimated

queue wait time is greater than the time until check-in closes.

Figure 3-14 demonstrates the technique for the proposed module of prioritising

travellers. Step one uses ‘Get block’ to read the flight number and class of travellers

to determine which check-in group they are in and the queue length. The next step

links ‘Get block’ with the developed customs block, ‘Prioritise arrivals’, to run the

algorithm illustrated in Figure 3-15.

Figure 3 - 14: Block 2: Prioritise arrivals.

Where:

𝑋 = the time before check-in counter close

𝑌 = estimated wait time at queue of check-in 𝑗

𝑄 = queue length of check-in 𝑗


Figure 3 - 15: Block 2: Algorithm for assigning passenger high priority.

Block 3: Check-in group simulation model

As there are three types of check-in, a decision block named ‘Select Item Out’

was used to read the check-in option for each passenger. The online check-in option

was assigned a value of 0, while common check-in (airport counters) and self-service

(kiosk) check-in were assigned values of 1 and 2, respectively (see Figure 3-16). After

identifying the check-in option, each item is directed to the associated option. The

reason for using a decision module in the check-in system simulation was because of

the location of a kiosk that comes before the common check-in counters.


Figure 3 - 16: Block 3: Assigning check-in type using the decision module.

Figure 3-17 illustrates the self-service check-in module as the second module of

block 2. In this module, it is assumed that the processing time follows a normal

distribution where the mean is 2 and the standard deviation is 1. The capacity of this

module can be controlled using the ‘maximum items in activity’ box, which based on

the real data was 50 kiosks.


Figure 3 - 17: Block 3: Self-service module.

The last module of block 3 is where passengers are routed to the proper check-

in lane (see Figure 3-18). This can be done by determining which group the flight

belongs to. In the next section, the hierarchical block of check-in airline groups is

explained in more detail.

Figure 3 - 18: Block 3: Hierarchical block for check-in group module.


Block 4: Check-in counters

Each check-in lane has an ID that is linked to the input data, including flight

attributes and passenger attributes. Another reason for giving each lane an ID value is

so the output of the simulation runs can be reported. Compared with other outbound

processes, the check-in system behaves differently because it is operated by different

airline companies within a separate lane, whereas security screening and immigration

are operated by the same agent for all lanes and counters.

Before entering the check-in counter queues, a decision module is used to read

the travel class of the passengers to route them to the proper queue (see Figure 3-19).

Figure 3 - 19: Block 4: Decision module for selecting class of travellers.

A decision module was also used to identify the attributes of check-in type with

respect to bag check-in. The check-in options at the common check-in counters

include:

online check-in with bags

self-service check-in with bags

common check-in with or without bags

Additionally, Figure 3-20 demonstrates the decision modules for the check-in

type and the number of bags for each passenger. Some passengers check-in using self-


service kiosks or online services, but they might have bags they need to check in

manually at check-in counters.

Figure 3 - 20: Block 4: Decision modules for number of bags and check-in type.

The last module of block 4 is the check-in counters module. An activity block is

used to simulate each workstation and how it works based on queue length. This

module consists of three sub-modules. The first module, ‘Get block’, is used to link

the input data stored in the Excel spreadsheet with the activity block to read the delay

time for each passenger based on the number of bags (see Figure 3-21).


Figure 3 - 21: Block 4: Delay time module.

The second sub-module of the check-in counters controls the workstation

dynamically by setting the workstation to ‘active’ or ‘shutdown’ based on the

availability and queue length (see Figure 3-22). This procedure can be done by using

a ‘read’ block to check if the queue length exceeds the threshold and if there is

available staff at each time unit of the simulation. The last sub-module of the check-in

counters uses a ‘writing’ block to report the utilisation of each workstation.

Figure 3 - 22: Block 4: Workstation control module.


Block 5: Security screening checkpoints.

In this section, block 5 of the security screening system is explained in more

detail. The block considers the following modules:

VIP lane option as the first module

Queue system in front of x-ray machines

Workstations refer to (x-ray) machines

Recheck because of the first failure

Control workstation dynamically

Figure 3-23 illustrates the decision module of passenger type at the security

screening checkpoints. This module is used to route passengers to the proper security

screening lane. As 0.05 of passengers are diplomatic, block 5 includes a separate lane

for diplomatic passengers.

Figure 3 - 23: Block 5: Diplomatic decision module.

The passengers are then held in the queues for the x-ray lane for screening

purposes and are processed based on a first-in-first-out (FIFO) queue method (see

Figure 3-24) (van Boekhold et al., 2014). The output of the simulation, e.g.

maximum/average queue length and maximum/average waiting time at screening

machines is stored using a ‘write’ block.


Figure 3 - 24: Block 5: Queue system module.

While Figure 3-25 shows the distribution of the delays at the x-ray machines, it

is assumed that processing time at x-ray machines follow triangular with maximum

0.75, minimum 0.2, and most likely 0.5 minutes (Cheng, 2014).

Figure 3 - 25: Block 5: Processing time distribution module.

Sometimes passengers, or their items, need to be re-checked because of a

triggered alarm. According to Olaru and Emery (2007), 10% of first screenings fail


and the items need to be re-checked. Therefore, the decision module illustrated in

Figure 3-26 was used to reroute items randomly.

Figure 3 - 26: Block 5: Security first failure module.

The last module of block 5 controls the workstations of the x-ray machines.

These processes can be based on demand, as represented by the queue length and the

available personnel (see Figure 3-27).

Figure 3 - 27: Block 5: Workstation control module.


Block 6: Random explosive security check

After completing the x-ray stage, 15% of passenger will be selected randomly

for further inspection. Both diplomatic and non-diplomatic passengers can be chosen

to be inspected (see Figure 3-28).

Figure 3 - 28: Block 6: Probability of random explosive check module.

A decision module is used to appoint attributes to each passenger that is selected

for a random or secondary check (see Figure 3-29).

Figure 3 - 29: Block 6: Random check decision module


As can be seen in Figure 3-30, the activity block is used to simulate the

workstation. The processing time distribution for this block follows the normal

distribution, with a mean of 2 minutes and a standard deviation of 0.2 minutes.

Figure 3 - 30: Block 6: Processing time distribution at random explosive check.

Block 7: Immigration counters and SmartGate system

The logic of block 7 can be separated into four identical sections. Each section

has its own features in Linking, assigning and managing the process of handling

passengers. Figure 3-31 shows the decision module as the first module of block 7,

which is used to identify the SmartGate users and then routes each type of passenger

to the SmartGate or to the common immigration counters.


Figure 3 - 31: Block 7: SmartGate user decision module check.

Figure 3-32 shows an overview of the SmartGate service and the distribution of

the passengers being processed. The module contains three sub-modules: a queue

system module, blocks for the output report, and a workstation module. An activity

block was used to simulate the SmartGate machine using the related parameters, such

as processing time distribution and the number of available machines. ‘Read’ and

‘write’ blocks were used to record the simulation output, e.g. maximum/average

waiting time and maximum/average queue length.

Figure 3 - 32: Block 7: SmartGate processing time distribution.


Other passengers are directed to the standard immigration counters. A queue

block is used as the first block to hold passengers in front of the immigration counters.

People are processed based on the FIFO queue method (see Figure 3-33).

Figure 3 - 33: Block 7: Immigration queue system.

The last module of block 7 controls the operation of the immigration

workstations. In this module, a ‘read’ block is used to link each workstation with the

resource pool (see Figure 3-34).

Figure 3 - 34: Block 7: Immigration workstation control module.


Block 8: Boarding procedures

Block 8 has three modules with different functionalities. As the distance between

immigration and the boarding gates varies due to the location of each gate, a

mathematical block denoted by an equation block was used to calculate the walking

time from the immigration system to the boarding gate (see Figure 3-35). The module

was developed according to the following steps:

Define the value of ‘walking speed’ if we assume that the average walking

speed is between 3-5 k/h based on the age of the passenger.

The destination is assumed to follow a normal distribution with a mean of

𝜇 = 200 and a standard deviation of 𝜎 = 70.

Therefore, the walking time is calculated using the equation

𝑤𝑎𝑙𝑘𝑖𝑛𝑔 𝑡𝑖𝑚𝑒 =(

Distance 𝑚𝑒𝑡𝑒𝑟𝑠

1000)

𝑤𝑎𝑙𝑘𝑖𝑛𝑔 𝑠𝑝𝑒𝑒𝑑∗60 (2)

Figure 3 - 35: Block 8: Walking time distribution for boarding gate module.

The second module of the boarding procedure uses ‘Get block’ to read the

attributes of the flight number assigned to the right gate (see Figure 3-36).


Figure 3 - 36: Block 8: Walking time distribution to boarding gate module.

ExtendSim allows modellers to develop custom blocks to mimic real-life

scenarios. Therefore, an algorithm was developed to simulate the boarding procedures

(see Figure 3- 37). In contrast to the other processes, the boarding process is performed

by the same staff as the check-in process, therefore, there is a need for an algorithm to

share staff between check-in and boarding. The boarding procedure algorithm was

based on the following algorithmic steps:

1. Read each gate status.

2. Select which one is active based on flight attribute ‘boarding time’.

3. Do not select staff who have finished their shift.

4. Check if there is staff available in the pool, otherwise move staff from

associated check-in.

Figure 3-37 shows the algorithm flowchart for the boarding procedures. The

related data are set-up and uploaded at the beginning of the simulation. The input data

for the developed algorithm are ‘time taken to process at the gate’, ‘number of staff at

the gate’, ‘process time to get on the plane’, and ‘delay time on the jetway’.


𝑁𝐴𝑆: Number of airline staff available in the database of global array

𝐴𝑆𝐶: Airline staff for check-in counters assigned 𝑗 = 1,2,3 …

𝐺𝑖: Departure gate 𝑖 = 1,2,3 …

𝐴𝑏𝑔𝑎𝑡𝑒: Airline staff for boarding gate

𝐴𝑆𝑠ℎ𝑖𝑓𝑡__𝑒𝑛𝑑 : End shift of airline staff

Figure 3 - 37: Block 8: Flowchart for boarding procedure algorithms.

3.6 NUMERICAL TESTING

The simulation framework was applied to Brisbane International Airport (BNE).

A model was developed that includes the main characteristics of the BNE with regards

to passenger flow and processing and with respect to a variety of functional areas and

facilities. To validate the model, four different load factors were evaluated. The load

factor is the proportion of an airplane’s seats that are occupied. The load factors

considered were 50%, 60%, 75%, and 100%. Several flight schedules were analysed

to understand their impact on passenger arrival profiles and terminal facilities.


3.6.1 Impact on arrival process

The arrival pattern of passengers and the rate of passenger arrivals is affected by

flight departure times and the destinations of flights. According to Rauch and Kljajić

(2006), passengers with early flights generally arrive later than the statistical average.

The mode of arrival to the terminal depends on the modal split, which is the proportion

of passengers that use private cars, trains, and buses. Figure 3-38 and Figure 3-39

demonstrate the passenger arrival profiles per mode of transport and show the

distribution of passenger arrivals over time (in 10-minute intervals). We assume 75%

of the BNE passengers use private cars, 10% use buses, and 15% use trains.

Figure 3 - 38: Arrivals patterns for 100% flights full.

Figure 3 - 39: Arrivals patterns for 50% flights full.

0

20

40

60

80

100

120

140

160

0

60

12

0

18

0

24

0

30

0

36

0

42

0

48

0

54

0

60

0

66

0

72

0

78

0

84

0

90

0

96

0

10

20

10

80

11

40

12

00

12

60

13

20

13

80

14

40

No

. Pas

sen

gers

Time (min)

Arrival Time (min) Midnight to Midnight

Car Bus Train

0

20

40

60

80

100

0

60

12

0

18

0

24

0

30

0

36

0

42

0

48

0

54

0

60

0

66

0

72

0

78

0

84

0

90

0

96

0

10

20

10

80

11

40

12

00

12

60

13

20

13

80

14

40

No

. Pas

sen

gers

Time (min)

Arrival Time (min) Midnight to Midnight

Car Bus Train


It can be seen that two high peaks occur in the morning at 08:20 and 10:30 and

that two lower peaks occur in the early evening and at night between 17:40 and 18:35

and between 21:20 and 23:50, respectively. This information can be used to assist

airport operational managers with scheduling staff within the terminal and its facilities.

It can also be used to determine the walking time required for the passengers from each

transport station at which they arrive at the terminal.

3.6.2 The impact on terminal facilities

As far as the efficiency of the operational processes at airport terminals is

concerned, Figures 3-40 and 3-41 and Figures 3-42 and 3-43 show the simulation

outcomes for the security screening and immigration processes, respectively.

0

2

4

6

8

10

1

59

11

7

17

5

23

3

29

1

34

9

40

7

46

5

52

3

58

1

63

9

69

7

75

5

81

3

87

1

92

9

98

7

10

45

11

03

11

61

12

19

12

77

13

35

13

93

14

51

no

. of

Pas

sen

gers

Simulation run time (min)

Security Queue Length

Q Diplomatic Q Common Q Random

0

10

20

30

40

16

11

21

18

12

41

30

13

61

42

14

81

54

16

01

66

17

21

78

18

41

90

19

61

10

21

10

81

11

41

12

01

12

61

13

21

13

81

14

41

no

. of

Pas

sen

gers


Security Queue Length

Q Diplomatic Q Common Q Random

Figure 3 - 40: Security queue length 100% flights full.

Figure 3 - 41: Security queue length 50% flights full.


Figure 3 - 42: Immigration queue length 100% flights full.

Figure 3 - 43: Immigration queue length 50% flights full.

What-if scenarios were performed to analyse the queue length for the two

facilities. It is clear that the queue length for security screening decreases by more than

four times if the capacity of the flight is 50% full. For the same condition, the queue

length for immigration sharply decreases from 275 to eight passengers. In addition, the

above figures clearly demonstrate that a severe bottleneck occurs at security screening

and immigration during the day between 11:30 and 16:30. Therefore, it is believed that

the results of the model are quite accurate because they align with the arrival profiles

shown earlier.

0

2

4

6

8

10

15

51

09

16

32

17

27

13

25

37

94

33

48

75

41

59

56

49

70

37

57

81

18

65

91

99

73

10

27

10

81

11

35

11

89

12

43

12

97

13

51

14

05

no

. of

Pas

sen

gers


Immigration Queue Length

Queue Length Common counter Queue Length Smart gate

0

50

100

150

200

250

300

1

57

11

3

16

9

22

5

28

1

33

7

39

3

44

9

50

5

56

1

61

7

67

3

72

9

78

5

84

1

89

7

95

3

10

09

10

65

11

21

11

77

12

33

12

89

13

45

14

01

no

. of

pas

sen

gers


Immigration Queue Length

Queue Length Common counter Queue Length Smart gate


3.7 CHAPTER SUMMARY

At the beginning of this chapter, the three different types of passenger flow

processes were discussed: departure, arrival, and transfer. Each type indicates one form

of passenger flow. The conceptual framework for developing such a holistic model

that considers the entire processes of an airport terminal was also discussed.

This chapter focuses on outbound processes as the first component of an

overarching model for simulating passenger flow at Australian international airports.

ExtendSim simulation of outbound processes consists of eight key modules including

arrival to the terminal, check-in, security screening, immigration and boarding

procedures. Then the process of generating passengers arriving at the international

airport was explained using the flight attributes and the relevant information, such as

flight schedules, passenger characteristics, and boarding characteristics as the main

inputs.

The simulation results demonstrate that flight schedules have a large impact on

passenger flows. The proposed simulation framework and model can be used to predict

ahead of time the effect of different flight schedules and may be used as a feedback

mechanism to improve it before implementation. Taken together, these results suggest

that integrated flight schedule creation and passenger simulation analysis may be an

avenue for addressing some of the issues of passenger flow within airport terminals,

especially for the two most-affected processes: security screening and immigration.

The next chapter demonstrates how the developed outbound simulation model

can be used to analyse the problem of different arrival patterns and their impact on

departure terminal facilities.

Chapter 4: The Impacts of Arrival Patterns on Airport Mandatory Processes 85

Chapter 4: The Impacts of Arrival Patterns on

Airport Mandatory Processes

4.1 INTRODUCTION

Modelling of generic framework simulation models for outbound major standard

processes of an international airport terminal was the focus of work undertaken in Chapter 3.

The traditional layout of the outbound processes, the characteristics of entities data, and the

characteristics of operational procedures were discussed.

This chapter discusses the development of the passenger arrival process model and

demonstrates how the arrival patterns affect the ability of outbound processes to process

outgoing passengers (Joshi, 2008). Estimating the exact arrivals of passengers to the airport is

difficult as the passenger’s arrivals are highly dynamic, and the rates of arrivals change from

time to time. Hence, this study has been conducted to understand such uncertainties by testing

various shapes of distributions in order to provide the best policy that satisfies check-in,

security screening, and immigration processes in terms of minimum waiting time and length

of queue. The pattern of passenger arrivals is considered an important factor in planning

airport-terminal facilities, such as the number of check-in counters and service agents, along

with the operation times of passenger check-in and queue length (Park & Ahn, 2003). The

arrival pattern of passengers and the rate of passenger arrivals are affected by many factors,

including flight-departure times, type of traveller (business/leisure), and flight destination.

According to Rauch and Kljajić (2006) passengers with early flights generally arrive later than

the statistical average. The airport check-in rules are considered significant factors influencing

passenger-arrival patterns at the airport (Manataki & Zografos, 2009b).

Most airports share some common arrival behaviours for passengers in international

terminals. These behaviours are as follows:

90% of passengers arrive at the airport 60 minutes before departure time;

Business passengers arrive later than leisure passengers;

The peak hours of check-in are 100–120 minutes before the departure time; and

In the morning, the peak hours are shorter but busier than in the evening (Ashford et

al., 2011; Cheng, 2014).

86 Chapter 4: The Impacts of Arrival Patterns on Airport Mandatory Processes

The simulation model presented in Chapter 3 will be used to analyse the effect of

passenger arrival variation. This model also investigates the influence of arrival patterns on the

efficiency of airport-terminal processing and focuses on mandatory processes, including check-

in, security checkpoints, and immigration. The outcome of this analysis could help in setting

the time-of-arrival policy for international airports. To determine the most suitable arrival

policy for an airport, two basic parameters are critical: the mean value of time before flight (µ)

and the arrival time before flight (Ω). Two experiments were conducted to investigate the

influence of these parameters on arrival patterns and airport terminals at check-in points.

Finally, we present the strategy for determining the optimum policy for the airport in section

4.5.

4.2 DEVELOPMENT OF PASSENGER ARRIVAL PROCESS MODEL

Passenger arrival time is treated as a random variable. The distribution of passengers

arriving at the check-in counters varies by time of day, day of week, airport, season, and type

of passenger. Other factors, such as the mode of transportation and the security requirements,

are not considered Furthermore, this model explains the behaviours of any passenger (business

or leisure). The flight schedule for a Wednesday was obtained from an international airport and

used as input for the model to determine the following:

Estimate the volume of passengers showing up at the terminal.

The passenger-arrival pattern during the day (24 h).

Several different statistical distributions can describe the arrival profile based on the

given flight schedule. These distributions include the exponential distribution, uniform

distribution, empirical distribution, and normal distribution (Fonzone, Schmöcker, & Liu,

2015; Olaru & Emery, 2007; Schultz & Fricke, 2011). In this study, the normal distribution

was selected to characterise the passenger-arrival pattern based on the test done by Ma (2013)

for a given flight schedule, the whole arrival rate of the proposed data follow normal

distribution. Figure 4-1 depicts the flow chart used to obtain the arrival pattern under normal

distribution from the flight-schedule data.


Figure 4 - 1: Flowchart for modelling passenger arrivals at the international terminal

The modelling procedure for estimating the volume of passengers arriving at the airport

over time is based on the following algorithm:

1. Obtain flight schedule of the airport, including the related flight information that

contains the airline, scheduled time of departure, and number of passengers on each

flight.

2. Select the policy for passenger-arrival time at the international airport. For example,

2 hours, 3 hours, or 4 hours beforehand (scheduled departure time).

3. Determine the relevant distribution, along with critical parameters such as mean and

standard deviation.

Obtain airport flight

Select time of PAX arrive before flight

departure time

Calculate PDF for flight 𝑖

Calculate CDF for flight 𝑖

𝒊 = N

𝑖 = 𝒊 + 𝟏 𝒊 < N

Collect aggregation for all


4. Calculate the probability distribution function (PDF) and the cumulative distribution

function for each flight (𝑖) to determine the number of passengers arriving per time

interval before the departure time.

5. Add the number of passengers arriving at each time interval for each flight to

estimate the incremental total number of passengers arriving for all flights. Thereby,

the aggregate numbers of passengers arriving per time interval can be obtained for

the entire flight timetable.

4.3 CASE STUDY 1: IMPACTS OF DIFFERENT TIME BEFORE DEPARTURE

VALUES

In this case study, a multitude of scenarios has been tested. As explained previously, the

investigation will consider the two critical parameters, the mean value (µ) of arrival

distributions and the policy on time to arrive at the airport (such as 3, 3.5 or 4 hours before

departure time). The experiments are conducted to study the impact of different arrival times

at the airport given fixed µ (see Table 4-1).

The behaviour of Cumulative Distribution Function (CDF), and arrival profiles caused

by varying time before the flight (Ω) for a given mean (µ) time before departure will be

discussed. It also investigates to what extent different policies impact the performance of

terminal processing points.

Table 4 - 1: Selection of Ω values under a fixed µ values

Mean

(min)

Sets of time before departure (in min) under

each mean

60 120 150 180 210 240

90 120 150 180 210 240

120 120 150 180 210 240

150 150 180 210 240

180 180 210 240

210 210 240

For a given µ value, the CDF, and arrivals pattern, were calculated using the proposed

approach presented in section 4.2. This procedure was followed for every µ value listed in the

Table 4-1. The simulation software described previously was used to investigate the possible

effect on the passenger flow performance at terminal processing points by varying the time of


passengers arriving at the airport prior to departure time for a given mean value. The key

performance metrics, such as waiting time, average waiting time, queue length and average

queue length, have been considered for this analysis. These plots are described in the next

sections.

4.3.1 Behaviour of CDF of time before flight.

Figure 4-2 depicts the cumulative distribution function (CDF) for the same policies of a

passenger arriving for flight 𝑖. It is clear to see that passengers are more distributed under the

mean depending on how early they came to the airport, especially for those people arriving 2

hours and 2.5 hours before the departure time (see plots a and b). In contrast, there is less

distribution if passengers arrive 3.5 hours and 4 hours before departure time under the mean

value of 90, 120 and 150 min. This means passengers will experience less rush with these

policies. From Figure 4-2, it can be seen that any time after the mean value will have the lowest

CDF value corresponding to the earliest time before the flight. Also, it shows significant

variation in plots with low mean value and the variation decreasing as mean value decreases.

Figure 4 - 2: CDF of passengers arriving before flights for a given mean (µ): (a) µ = 60 min; (b) µ = 90 min; (c)

µ = 120 min; (d) µ = 150 min; (e) µ = 180 min (f) µ = 210 min.

(a) (b) (c)

(d) (e) (f)


For any given µ value, corresponding CDF curves for the time before flight values, share

a typical value at the mean. From Figure 4-2 it is also clear to see that for any given µ value,

in the time interval µ < t <= Ω, CDF increases as Ω increases and CDF decreases as Ω increases

in the time interval µ > t > 0. Hence, the impact from the Ω under same µ value is similar for

any µ value.

4.3.2 Behaviours of arrival pattern

Figure 4-3 illustrates arrival profiles of departing passengers for different times before

flight Ω under a given mean µ. For any given µ, the arrival pattern is very similar for any Ω

value that has the condition Ω>µ. Despite the merely slight differences for this situation, it can

be observed that the arrival pattern with different Ω values is almost similar, irrespective of the

µ value. However, for the graphs Ω≤µ, the arrival pattern shows variation from other graphs in

certain regions as follows:

i. For µ ≤ 120, considerable increase of PAX of can be seen in Ω = 120 curve, in the

time interval around 12:00 h to 20:00 h.

ii. For µ = 150, considerable increase of PAX of can be seen in Ω = 150 curve, in the

time interval around 07:00 h to 22:00 h.

iii. For µ ≥ 180, slight decrease of PAX of can be seen in Ω = 180 and 210 curves in

graphs (e) and (f) respectively, in the time interval around 14:00 h to 22:00 h.

4.3.3 Results of simulation and discussion

This section presents the data from running the simulation model to show the possible

impacts of varying time before flight under a given mean. Table 4-2 summarises the results of

the simulation model for the proposed scenarios, including four keys of the operational

performance metrics. These metrics are maximum/average queue length and

maximum/average waiting time. The simulation results illustrate that the queue length and

waiting time decrease when the time before a flight increases.


Figure 4 - 3: Departing passenger arrival profiles at airport terminal for different (Ω) under given (µ): (a) µ = 60 min; (b) µ = 90 min; (c) µ = 120 min; (d) µ = 150 min; (e) µ

= 180 min (f) µ = 210 min.


Table 4-2: Detailed output of ExtendSim simulation model for case study 1

Scenarios Check-in Security Immigration

𝜇 values Ω value Max

queue

Ave

queue

Max

waiting

Average

waiting

Max

queue

Ave

queue

Max

waiting

Average

waiting

Max

queue

Ave

queue

Max

waiting

Average

waiting

𝜇 =60 120min 46 2.06 23.58 3.38 111 7.68 12.45 1.87 413 61.48 82.33 20.37

150 min 40 1.04 17.85 1.7 105 4.33 11.14 1.06 321 43.38 64.06 14.42

180min 40 1.14 15.29 1.9 126 7.07 13.04 1.73 344 47.66 68.66 15.91

210min 59 2.02 27.54 3.33 103 5.46 10.85 1.33 237 34.51 47.51 11.43

240min 95 2.95 37.61 4.87 54 1.76 5.36 0.43 220 20.85 43.93 6.86

𝜇 =90 120min 22 0.81 10.67 1.33 210 18.71 21.86 4.57 431 72.58 86.15 23.8

150 min 9 0.15 3.25 0.25 187 16.29 20.19 3.99 376 55.47 75.13 18.32

180min 14 0.31 6.16 0.51 138 9.35 15.11 2.27 353 54.84 70.51 18.34

210min 19 0.42 9.53 0.71 109 5.76 12 1.41 366 60.36 73.09 20.01

240min 34 0.56 14.03 0.92 49 2.88 5.26 0.71 295 47.22 58.88 15.58

𝜇 =120 120min 37 1.12 14.82 1.86 163 13.44 16.9 3.29 443 80.74 88.57 26.61

150 min 11 0.2 4.28 0.33 178 15.38 18.64 3.75 413 68.45 82.47 22.45

180min 7 0.08 3.92 0.124 181 13.34 19.86 3.25 408 70.36 81.69 23.19

210min 27 1.01 12.21 1.7 95 3.93 9.93 0.96 306 46.01 61.2 15.05

240min 16 0.12 6.21 0.2 125 8.3 13.28 2.03 330 49.12 65.77 16.33

𝜇 =150 150 min 11 0.11 4.03 0.19 140 9.85 15.28 2.41 389 65.82 77.69 21.89

180min 12 0.11 4.68 0.18 137 10.29 15.16 2.51 407 72.59 81.33 24.16

210min 4 0.03 1.48 0.04 127 9.63 13.13 2.36 371 62.97 74.22 20.68

240min 8 0.07 3.14 0.11 116 6.13 12.2 1.51 379 66.76 75.96 21.87

𝜇 =180 180min 8 0.04 3.24 0.07 163 11.31 18.19 2.77 359 58.21 71.68 19.63

210min 7 0.07 2.87 0.011 128 9.84 14.06 2.41 393 66.74 78.26 22.22

240min 31 0.7 14.2 1.16 121 7.59 13.06 1.86 382 64.3 76.22 21.21

𝜇 =210 210min 42 2.19 19.98 3.61 151 11.94 15.84 2.91 396 67.79 79.13 22.56

240min 69 3.21 27.59 5.27 78 3.56 8.94 0.087 294 46.93 58.49 15.61


As can be seen from Figures 4-4, 4-5, and 4-6, it is evident that for any µ and Ω

combinations, the overall behaviour of the three queues of check-in, security, and immigration

with different time before flight values are dissimilar for each µ value. However, the maximum

queue length for immigration for the same figures occurred at Ω = 120 min, while the minimum

queue length of immigration processes was found at Ω = 240 under the µ = 60 min. In the

check-in process, the maximum queue length happened at Ω = 240 min with µ = 60 min and µ

= 90 min.

Figure 4 - 4: Queue lengths of different time before flight given µ = 60




From the above figures, it is observed that passengers who arrive with greater or equal to

120 µ values will be processed with minimum waiting in the queues overall outbound

processes. Moreover, there is a direct relationship between the policies of time of passengers

arrival at the airport and the performance of departure processes, where the waiting time in the

queue is decreased if the policy of time to arrive at the airport is increased.

4.4 CASE STUDY 2: IMPACTS OF DIFFERENT MEAN VALUES

The second experiment is focused on the impact of a variable mean value given a fixed

arriving time (such as three hours before departure time) as shown in Table 4-3. A combination

of different arrival policies at an international airport before the departure time was simulated.

The objective of case study 2 is to further investigate the behaviour of CDF, and the arrival

profile caused by different mean µ values for a given time before departure value Ω. For this

study, the following table was used to extract the scenarios. For a given Ω value, the CDF

departing passenger arrivals profile, we follow a similar procedure as that explained in case

study one, the proposed approach is used to calculate CDF and arrival profile. This procedure

was followed for every Ω value listed in Table 4.3.


Table 4-3: Selection of Ω values under different µ values

Time before

flight (min)

Mean (min)

120 60 90 120

150 60 90 120 150

180 60 90 120 150 180

210 60 90 120 150 180 210

240 60 90 120 150 180 210 240

For a given Ω value, the CDF departing passenger arrivals profile, we follow similar

procedure explained in the case study one, the proposed approach is used to calculate CDF,

PDF and arrival profile. This procedure was followed for every Ω value listed in Table 4.3.

4.4.1 Behaviour of CDF

Figure 4-7 illustrates the CDF of different mean values for a given time before the flight.

It can be seen that by having different mean values, the maximum values of CDF are similar,

and all graphs replicate each other. Moreover, for a given time before the flight, the CDF is

increasing with µ value at any time.

Figure 4 - 7: CDF of passengers arriving at airport for flight (i) for a given time before the flight under different

(µ): time of passenger arriving (a) Ω = 120 min; (b) Ω = 150 min; (c) Ω = 180 min; (d) Ω = 210 min; (e) Ω =

240 min

(a) (b) (c)

(d) (e)


4.4.2 Behaviours of arrival pattern

Figure 4-8 illustrates the arrival profile of the departing passengers for a given time

before the scheduled departure time under different values of mean µ. As shown in the figure,

for a given time before the scheduled departure time, the behaviour of arrival patterns are the

same for every mean value. Furthermore, for each plot the peak value is the same for any mean

values. All plots reach their peak in the time interval 05:00 h to 10:00 h, and in the time interval

20:00 h to 24:00 h. In this case, for any graphs Ω ≤ µ, arrival patterns show little variation

compared with what was seen in case study one. In addition, the number of passengers reaches

the possible maximum at the end of three hours’ time.

Figure 4 - 8: Departing passenger arrival profiles at airport terminal for different (µ) given (Ω): (a) Ω = 120

min; (b) Ω = 150 min; (c) Ω = 180 min; (d) Ω = 210 min; (e) Ω = 240 min


4.4.3 Results of the simulation and discussion

From the data collected from the five different simulation scenarios in outbound terminal

processes including check-in, security screening and immigration processes, it was found that

that queue length and waiting time increase as µ increases, especially for the security and

immigration domain (Table 4-4). According to graphs (a) to (f) in Figure 4-9, it is evident that

for any µ and Ω combination, the queue lengths increase from check-in to immigration

processing points. Furthermore, the overall behaviour of the three queues with different µ

values are dissimilar for each Ω values. However, in graphs (a) to (c), the maximum queue

length for immigration increases with µ, until µ = 120 min.


Table 4-4: Detailed output of ExtendSim simulation model

Scenarios Check-in Security Immigration

Ω µ values Max

queue

Ave

queue

Max

waiting

Ave

waiting

Max

queue

Ave

queue

Max

waiting

Ave

waiting

Max

queue

Ave

queue

Max

waiting

Average

waiting

Ω = 120 min

60 46 2.06 23.58 3.38 111 7.68 12.45 1.87 413 61.48 82.33 20.37

90 22 0.81 10.67 1.33 210 18.71 21.86 4.57 431 72.58 86.15 23.8

120 37 1.12 14.82 1.86 163 13.44 16.9 3.29 443 80.74 88.57 26.61

Ω = 150 min

60 40 1.04 17.85 1.7 105 4.33 11.14 1.06 321 43.38 64.06 14.42

90 16 0.288 3.25 0.25 187 16.29 20.19 3.99 376 55.47 75.13 18.32

120 11 0.2 4.28 0.33 178 15.38 18.64 3.75 413 68.45 82.47 22.45

150 11 0.11 4.03 0.19 140 9.85 15.28 2.41 389 65.82 77.69 21.89

Ω = 180 min

60 40 1.14 15.29 1.9 126 7.07 13.04 1.73 344 47.66 68.66 15.91

90 14 0.31 6.16 0.51 138 9.35 15.11 2.27 353 54.84 70.51 18.34

120 7 0.08 3.92 0.124 181 13.34 19.86 3.25 408 70.36 81.69 23.19

150 12 0.11 4.68 0.18 137 10.29 15.16 2.51 407 72.59 81.33 24.16

180 8 0.04 3.24 0.07 163 11.31 18.19 2.77 359 58.21 71.68 19.63

Ω = 210 min

60 59 2.02 27.54 3.33 103 5.46 10.85 1.33 237 34.51 47.51 11.43

90 19 0.42 9.53 0.71 109 5.76 12 1.41 366 60.36 73.09 20.01

120 27 1.01 12.21 1.7 95 3.93 9.93 0.96 306 46.01 61.2 15.05

150 4 0.03 1.48 0.04 127 9.63 13.13 2.36 371 62.97 74.22 20.68

180 7 0.07 2.87 0.011 128 9.84 14.06 2.41 393 66.74 78.26 22.22

210 42 2.19 19.98 3.61 151 11.94 15.84 2.91 396 67.79 79.13 22.56

Ω = 240 min

60 84 2.95 37.61 4.87 54 1.76 5.36 0.43 220 20.85 43.93 6.86

90 34 0.56 14.03 0.92 49 2.88 5.26 0.71 295 47.22 58.88 15.58

120 16 0.12 6.21 0.2 125 8.3 13.28 2.03 330 49.12 65.77 16.33

150 8 0.07 3.14 0.11 116 6.13 12.2 1.51 379 66.76 75.96 21.87

180 31 0.7 14.2 1.16 121 7.59 13.06 1.86 382 64.3 76.22 21.21

210 69 3.21 27.59 5.27 78 3.56 8.94 0.087 294 46.93 58.49 15.61

240 92 6.38 42.42 10.52 207 19.25 22.75 4.7 393 62.09 78.59 20.66


Figure 4 - 9: Queue lengths of different mean value at a given time before flight


Figure 4-9 (b) and (c) show a decrease of the above variable for any µ afterwards.

The corresponding µ value for check-in remains at 60 min until Ω increases up to 210

min. Afterwards, the corresponding maximum queue length occurs µ = 240 min for Ω

= 240 min. The minimum queue length for check-in starts to occur at 90 min at Ω of

120 min. As Ω increases, this value reaches 150 min. The variation of queue length

with changing µ value is not always the same for any time Ω values. Based on the

results of simulation, immigration counters face considerable congestion most of the

time.

It is observed that passengers who arrive ≥ 2.5 hours before the departure time

are more likely to be served with the minimum waiting time, especially at check-in

and security, except if Ω = µ. Another observation is that the maximum queue length

of check-in counters occurred in the case of µ = 60. Overall, the departure processes

are interdependent with each other and impact on each other, which means if the queue

length of the check-in domain is increased, the queue length at the other processes

decreases, hence, the proposed model can be a very useful approach to balance the

performance of terminal operations.

4.5 SELECTION OF BEST TIME TO ARRIVE AT THE AIRPORT BASED

ON THE NORMAL DISTRIBUTION

In the previous two studies, the impact of passenger arrival patterns on airport

terminal facilities was investigated. Two case studies were conducted, the first one

investigated the impact of different arrival times before departure flight having the

same mean values. The second case study examined the influence of different mean

values for a given time before the flight was simulated. In this section, the best policy

of passengers arriving at the airport based on the two basic parameters of time before

flight Ω and mean µ time before departure will be determined. The results of

simulation for any combination of Ω and µ are presented in the table 4-5 below.


Table 4 - 5: Summary of the simulation results


4.5.1 Selection of the best scenario at each process

Figure 4-10 shows queue length and waiting for the check-in processes

considering all scenarios of the time before the flight. It can be observed that scenarios

5 and 26 have the highest values of queue length and waiting time, which means

whenever passengers came at Ω > 180 min.

Figure 4 - 10: Check-in queue length and waiting time for all scenarios

Figure 4-11 illustrates the queue length and waiting time of the security

screening checkpoints. From the graph below, it can be seen that the maximum queue

length and waiting time occurred in scenarios 7 and 26. While the minimum queue

length and waiting time occurred in scenario 19, which means the security screening

will have the optimum performance when passengers arrive 3.5 hours beforehand, with

the mean value of 150 min.

Figure 4 - 11: Security screening queue length and waiting time for all scenarios


Figure 4-12 shows queue length and waiting time at the process of immigration.

It is clear to see that the best scenario that meets both metrics is scenario number 5. It

is suggested that passengers should come to the airport four hours prior to the departure

time, with the mean value of 60 min.

Figure 4 - 12: Immigration queue length and waiting time for all scenarios

4.5.2 Aggregation of all processes

In this section, we will introduce the strategy of selecting the best policy of

passenger arrivals at the airport. This strategy aims to find the optimum policy

regarding the minimum waiting time queue length at each processing point, including

check-in, security checkpoints, and immigration for all 26 simulated scenarios. These

scenarios are the combinations of the two main parameters of Ω and µ (refer to Figure

4-12).

i. Sets of time before departure time (Ω) = 120 min, 150 min,

180 min, 210 min, 240 min

ii. Sets of the mean value of time before flight (µ) =

60 min, 90 min, 120 min, 150 min, 180 min, 210 min, 240 min

The procedure of this strategy is that each processing point (check-in, security

and immigration) was selected with each time having highest priority. Seven different

scenarios have been conducted to provide accurate results that satisfy all possible

scenarios that might occur in the real world, as each airport behaves differently, and

its elements have unique functioning features (Manataki & Zografos, 2010).


Given that check-in = a, security = b, immigration = c, the combinations of

possible conditions are introduced in the following sets:

𝑎 = 𝑏 = 𝑐, 𝑎 < 𝑏 < 𝑐, 𝑎 < 𝑐 < 𝑏, 𝑏 < 𝑎 < 𝑐, 𝑏 < 𝑐 < 𝑎, 𝑐 < 𝑏 < 𝑎, 𝑐 < 𝑎 < 𝑏

Figures 4-14(a) – (g) show the results of aggregation of the waiting time and

queue lengths for the three processing points (check-in, security and immigration) that

were selected in this study. Table 4-6 summaries the seven scenarios under different

priority for each process. Considering such scenarios would assist us in gaining

insights into the real-life scenarios since each airport operates differently and has

different behaviours with different priorities. Figure 4-13(a) demonstrates that all three

processes have the same priority with an equal factor of 33.3%. This factor is

multiplied by the KPIs values of each process. For each scenario, we then compute the

sum of the particular KPI value to reach the total value for all three processes. This

process is repeated for the remaining scenarios listed in Table 4-6.

Table 4- 6: Illustration of different conditions to select the best scenario

Scenario Condition Check-in factor

(%)

Security factor

(%)

Immigration

factor (%)

1 𝑎 = 𝑏 = 𝑐 33.3 33.3 33.3 2 𝑎 < 𝑏 < 𝑐 20 30 50 3 𝑎 < 𝑐 < 𝑏 20 50 30 4 𝑏 < 𝑎 < 𝑐 30 20 50 5 𝑏 < 𝑐 < 𝑎 50 20 30 6 𝑐 < 𝑏 < 𝑎 50 30 20 7 𝑐 < 𝑎 < 𝑏 30 50 20

(a)


(b)

(c)

(d)


Figure 4 - 13: (a-g) the impacts of different arrival patterns based on the priority for each processes

(e)

(f)

(g)


For the above scenarios, it is clear to see that the graphs in Figure 4-13 (b) – (d)

have the same behaviours and patterns when considering immigration is the higher

priority, especially for the Figure 4-13 (b) and (d) graphs. The minimum queue length

value occurred in scenario number 7 when passengers arrive at the airport two hours

beforehand under given mean value 90, yet the minimum waiting time can be seen in

both scenarios 7 and 13.

Figures 4-13 (e) - (g) have slightly different patterns when assumed that check-

in is the higher priority. It is observed that the check-in process has a significant impact

on the system when it is considered as the high priority, which leads to a smaller

number in the queue and shorter waiting time spent in the departure system. Thus, it

can be concluded that investigating arrival patterns of departing passengers is critical

and has relevance for planning, as this would assist airport management to respond

effectively and provide a better quality of service, enhancing passenger satisfaction.

Table 4-7 summarises the best policies that satisfied check-in, security and

immigration in terms of minimum queue length and waiting time.

Table 4-7: Summary of the results of selection of the best policy of time before flight

Case # Condition Best scenario considering

queue length

Best scenario

considering

waiting time

1 𝑎 = 𝑏 = 𝑐 7 13

2 𝑎 < 𝑏 < 𝑐 7 7

3 𝑎 < 𝑐 < 𝑏 7 7

4 𝑏 < 𝑎 < 𝑐 7 7

5 𝑏 < 𝑐 < 𝑎 7 13,16

6 𝑐 < 𝑏 < 𝑎 13 13

7 𝑐 < 𝑎 < 𝑏 7 13

4.6 CHAPTER SUMMARY

This chapter has analysed the influence of different arrival patterns on passenger

processing activities including check-in, security, and immigration using discrete

event simulation in order to address question two. These results provide important

insight into the two important parameters, namely time before flight Ω and mean µ

before the flight and how they impact the performance of an international terminal


system. Observations of the first experiment show that for a given µ under different Ω

values there is a significant influence on the departing passengers’ profiles, especially

during time before flight Ω ≤ µ. However, in the second set of experiments, the

behaviour of arrival patterns was similar for all scenarios. Furthermore, the simulation

results demonstrate how much congestion the airport will incur.

In case study one, it was observed that for a given µ the peak value for the

probability distribution increases when time before flight decreases. Furthermore,

having different times before flight under a given mean caused more of a significant

impact on departing passenger profile, especially when time before flight Ω ≤ µ. In the

terminal facilities, the queue length and waiting time decreased if the time before the

flight increased for any mean value. This impact was seen more in the security and

immigration processes. In contrast, for case study two, the behaviour of CDF and

arrival profile of departing passengers is similar for varying mean µ under for a given

time before flight. Moreover, the queue time and waiting were seen to increase from

check-in to immigration. In addition, the overall behaviour of the three queues with

different µ values are dissimilar for each Ω value.

The scenario of a passenger arriving at the airport terminal four hours beforehand

for a given mean µ value of 60 was found to be the best policy considering queue

length if the check-in and security were selected to be more important factors. In the

rare condition that all the processes are equal, scenario 7 and scenario 13 (passenger

arriving at the same time but under mean value µ 90) seem to be the best scenarios

when considering both queue length and waiting time.

In the next chapter, the simulation model will be extended to include inbound

processes and integrated with an Advanced Resource Management (ARM) approach

as the next phase of the overarching model

.

Chapter 5: A Framework for Sharing Staff between Outbound and Inbound Airport Processes. 109

Chapter 5: A Framework for Sharing Staff

between Outbound and Inbound

Airport Processes.

5.1 INTRODUCTION

In Chapter 3, the first phase of the overarching model was presented. It focussed

on the processes and sub-processes of the outbound system of an international airport

terminal, based on the discrete event theory discussed in sections 3.2 and 3.3. This

chapter will extend Chapters 3 and will introduce an integrated simulation model of

an airport terminal involving both inbound processing points and an Advanced

Resource Management (ARM) model. The model describes the potential interactions

between inbound passengers and outbound passengers, resulting in competing

priorities regarding physical space. This competition results in the need to model and

optimise these interactions under realistic terminal conditions.

It is believed that the development of such a model, including all airport aspects

(i.e. outbound and inbound), has become very important since most of the current

research is predominantly and separately focussed on airport operations associated

with the departure (outbound) passenger or the arrival (inbound) passenger. In

addition, this model provides a clear idea of passenger flows through the entire airport

terminal processing procedures. The objective of developing such a model is to

examine the possible bottlenecks in arriving and departing passenger flow. It also

provides a platform for studying more complex processing behaviour and operational

strategies using the simulation environment, ExtendSim. Consequently, insight into

current and future situations of airport systems will be gained. Section 5.2 discusses

the generic framework for inbound process flow of Australian international airports.

Section 5.3 illustrates the ExtendSim simulation models for terminal arrival processes.

Section 5.4 describes the development of resource allocation management algorithms,

including integrated and non-integrated modules.

110 Chapter 5: A Framework for Sharing Staff between Outbound and Inbound Airport Processes.

5.2 INBOUND PASSENGER FLOWS MODELLING

5.2.1 Outline of inbound flow processes

This section discusses inbound passenger flows at Australian airports. Wu,

Pitchforth, and Mengersen (2014) identified the inbound facilitation procedures

through consultations with airport specialists, including the airport operator,

biosecurity, border protection and customs. Based on Wu et al. (2014), incoming

passenger facilitation processes can be divided into four systems and sub-systems,

including arrival concourse, entry control point, baggage hall and secondary

examination area, as shown in Figure 5-1.

Figure 5 - 1: An illustration of inbound passenger facilitation processes (Wu et al., 2014).

As can be seen from Figure 5-1, these elements consist of both types of

activities: mandatory and discretionary (Wu et al., 2014). Mandatory activities refer

to activities that each passenger must pass through, such as immigration, security

screening, and quarantine; while discretionary activities refer to the optional activities,

such as duty free shopping and restroom usage.

At each process of the inbound system, passengers have different levels of

interactions. For example, at the immigration domain interface, a passenger interacts

with the personnel, equipment, and process, while at the disembarking area, a

passenger interacts with the airline service, personnel, and process (Popovic, Kraal, &

Kirk, 2010). In addition, there are many factors that influence processing time and

processing policies of inbound passengers at different processing points. An example


of these factors is the nationality of the passenger, which can have a great impact on

processing time at immigration checkpoints for two reasons. The first reason is that

passengers from countries like Australia, New Zealand and the USA, can use the new

SmartGate technology at immigration, while other passengers are processed manually

through immigration counters. The second reason is that some counters have stricter

visa requirements which might take extra time, for instance, for customs agents to

process (Pitchforth, Wu, Fookes, & Mengersen, 2015). The purpose of this research is

to determine the major components that are able to define the inbound passenger flow

within an international airport terminal. These components consist of disembarking,

immigration control point, baggage claim, and secondary examination area operated

by quarantine. The other points of discretionary activities, such as restrooms and

restaurants, are omitted from this study.

5.2.2 Inbound process simulation modelling

The process of the simulation model of the inbound system is illustrated in

Figure 5-2. The arriving passenger will disembark from the aircraft when it arrives at

the assigned gate according to flight schedules and other related factors, such as

aircraft size. For the passengers who have a connecting flight, they will be directed to

the security screening checkpoints, and then they will proceed to the gate for another

flight. In this research, it is assumed that transit passengers are within the concourse

only. The main input data of the inbound simulation model is introduced. The input

model consists of three input data points, including flight attributes, inbound passenger

attributes, and the requirement of passenger processing points.

Figure 5-3 demonstrates the modelling of relevant input data of the inbound

simulation system. In the first part of the graph (left-hand side), the inbound flight

attributes data are introduced. As we mentioned earlier in section 3.5, flight attributes

are considered the vital elements of simulation input data. They consist of three types

of data: flight schedule, number of passengers, and passenger’s characteristics. The

inbound flight schedule contains relevant flight information, i.e. scheduled time of

landing, gate number, flight code, and flight airline. The second type of data that are

required to estimate the number of passengers per flight are the size of the flight and

load factor. The load factor varies based on the type of airline and the day of the week

(Manataki & Zografos, 2009b). The last type of input data are the characteristics of


the passengers. Some assumptions have been made in relation to passenger

characteristics.

Figure 5 - 2: Flowchart of the upper level of the inbound process flow model


Figure 5 - 3: The input modelling of the inbound simulation model

5.3 INBOUND EXTENDSIM MODULE DESCRIPTION

In this section, a discrete event simulation was utilised to simulate inbound

process flow using ExtendSim software. The inbound simulation models can be

organised into two hierarchical levels. The first level of the hierarchy reflects the

inbound airport system broken down into a set of the main inbound procedures. The

second level describes the intricate details of the different sub-processes in the airport


terminal. Specifically, the main inbound procedures that the airport terminal model

consists of are:

Passengers disembarking, including generating inbound passengers’

attributes, such as SmartGate users, walking speed, number of bags etc.;

Inbound security screening checkpoint for inbound passengers, assuming

15% of total transfer passengers failed in the first attempt. This module

takes into consideration x-ray checks and secondary screen checks (i.e.

random checks);

Inbound immigration processing points, including SmartGate services and

the manned counter for passport control;

Quarantine processing checking points, including two separate queues to

process arrival passengers: one for passengers who have something to

declare, and another for passengers who have nothing to declare.

The simulation contains seven blocks, which demonstrate the actual practised

process in the inbound system of Australian airports. The first block describes the main

processes of the arriving system around the basis of a hierarchical module design, such

as disembarking passengers, security screening for transit passengers, inbound

immigration, waiting time in the baggage claim, waiting time at duty free, and

quarantine. Each one of these blocks contains many pre-programmed blocks of the

software with different functionalities: blocks setting attributes of passengers, or

setting parameters of processes, programmes or generators; blocks to display

information about the processes; blocks to perform a service, or queues; and other

blocks to perform the mathematics of logical operation (Olaru & Emery, 2007). The

second block discusses the process of creating entities in the model that refer to arrival

passengers – this will be explained later in section 5.3.2. Arrival flights have two types

of passengers: transit and non-transit. Transit passengers refer to those who have a

connecting flight, while non-transit refers to passengers who enter the final destination

of their journey. For transit passengers, block three was developed to simulate the

security screening process, as an interface point between the inbound and outbound

systems to link transit passengers with departing passengers at boarding gates.

The rest of the passengers will be routed to pass through the duty free area, as

illustrated in block four. Passengers reach the first processing point of the inbound


system of immigration for passport checks, as shown in block five. After this step,

passengers go to the baggage claim area to collect their bags, which is presented in

block six. Block seven illustrates the quarantine system, including both the lines of

something to declare and nothing to declare, using the hierarchy block. The second

level of the block simulates quarantine stations to scan and check if there are any

explosive or illegal items.

5.3.1 Block 1: Hierarchy blocks for inbound processes

In order to develop a robust inbound simulation model, the major elements that

represent the system should be extracted. The elements of the inbound system are

disembarkation passenger, security screening for transit passengers, time spent at

duty-free, inbound immigration, baggage collection and quarantine. Figure 5-4 shows

the upper level of the arrival passenger flow model at an Australian international

airport. Hierarchy blocks of ExtendSim were used to easily build a model with several

levels, where each reflects one particular process of passenger handling and its sub-

processes. Each hierarchy block has a unique functionality, hence the investigation of

the particular process, as an observation needs to be made on just one level. The

enumerated hierarchy block above will be explained further in the following sections.

Figure 5 - 4: Block 1: The high level of inbound flow modelling

5.3.2 Block 2: Creating inbound passengers’ entities of an arrived flight

Figure 5-5 illustrates the structure of the block two hierarchy, ‘passengers

disembarking’. The block focusses on the process of creating passenger entities based

on arrival flight scheduling data. This block consists of four modules, each with unique

attributes to identify and route the passenger throughout the map of the model.


Figure 5 - 5: Block 2: Structure of the hierarchical block ‘passengers disembarking’

There are three steps to generate inbound passengers. The first step is that flight

scheduling data were used to set up a flight attributes table, considering information

related to arrival flights, such as flight airlines, arrival time, assigned gate, and flight

capacity (Figure 5-6). The number of passengers and how this relates to other

variables, such as flight size and load factor, need to be examined. The methods used

to determine the number of passengers have been previously examined (Manataki &

Zografos, 2009b; Schultz & Fricke, 2011).

Figure 5 - 6: Inbound flight attributes


The next step is generating inbound passengers’ attributes. This can be done

through the development of an algorithm using an Excel macro visual basic

programme. Figure 5-7 illustrates the flowchart of the algorithm that was used for this

purpose. Unlike outbound passengers’ arrival time patterns, inbound passengers

reached the terminal at one time rather than arriving three hours beforehand.

Figure 5 - 7: Algorithm for creating inbound passenger attributes

Figure 5-8 shows a snapshot of inbound passenger attributes which will be the

input for the ExtendSim simulation model. Each row represents the major information

related to an arrived passenger, including arrival time, flight number, gate number,

number of bags, and walking speed. Based on the given attributes, ExtendSim can read

particular attributes at any processing station and deal with it. For example, since there

are two types of passengers in the immigration system, a decision module is used to

identify the type, and guide them to the proper processing point.


Figure 5 - 8: Inbound passenger attributes

The third step is to link the inbound passenger table with ExtendSim through a

global array database, as explained in Chapter three, section 3.5. The global array can

be defined as a two-dimensional (row and column) array of data that is reachable by

any block in the model. The role of a global array is to share information between

blocks when a direct connection is either impossible or inconvenient, to exchange data,

or to store information that can be accessed by a row and column index. Figure 5-9

clarifies the mechanism for connecting the data source developed in the Excel

worksheet with ‘block two’s’ disembarking passenger.


Figure 5 - 9: Block 2: Mechanism of linking inbound passenger attributes with the ExtendSim model.

In addition, block two considers the variable of a passenger’s walking time

between the gate and the first processing point. Walking time for each passenger can

be calculated by dividing walking speed by the distance between a particular gate and

the immigration processing point. It is assumed the distribution of walking speed is


three to five kilometres per hour depending on the age of the traveller, according to

TranSafety (1997) study. The following figures explain the technique that ExtendSim

uses to calculate walking time. First, Get-Module is used to determine the distribution

of walking speed of specific passengers, as illustrated in Figure 5-10.

Figure 5 - 10: Block 2: Walking speed module

The second step is determining the gate number and the distance between the

gate and the first processing point. As it can be seen in Figure 5-11, Get-Module is

used to read the gate number of each arriving passenger, which is linked to the lookup

table module. The reason for using the lookup is to store the data of the gate number

and the distance between each gate to the first processing point.


Figure 5 - 11: Block 2: Arrival calculating gate distance module

Lastly, an equation module is connected to calculate the value of walking time

and to send this value to the activity module to be processed, as shown in Figure 5-12.

Figure 5 - 12: Block 3: Walking time module


5.3.3 Block 3: Inbound security module

The module of inbound security is developed to process passengers who have a

connecting flight. This module is built based on three levels. The first level

demonstrates the hierarchy block of the x-ray system and the decision module for the

random check queue, as shown in Figure 5-13. According to Maertens and Grimme

(2015), the assumption is made that only 20% of total inbound passengers are

considered transit passengers.

Figure 5 - 13: Block 3: The hierarchy module of x-ray and routing for random check

The second level represents the x-ray system which is further split into two sub-

modules, including the queueing system and workstations, as illustrated in Figure 5-

14. At the inbound security point, the passenger is processed based on the FIFO

method which serves each passenger at a time, but the passenger with the shortest time

will be processed first.


Figure 5 - 14: Block 3: Simulated queue of x-ray check and hierarchy block of workstation

The third level demonstrates the activity block for processing items

(passengers/ongoing bags). The characteristics of processing the objects can be

defined as the distribution of processing items and the capacity of workstations to

process each time unit, as shown in Figure 5-15. As can be seen, the maximum item

in the workstation is one each time, the processing time distribution is from ~Tri (0.2,

0.5, 0.75) where the min is 0.2, max is 0.75, and the most likely value is 0.5.


Figure 5 - 15: Block 3: Characteristics of processing items

As previously explained, the write and read modules were used to read data from

or write data to the ExtendSim database, such as global array or Excel workbooks.

Thus, write modules are linked to queue modules to record each minute of the output

of queue length and waiting time at the inbound security system; this is illustrated in

Figure 5-16.

Figure 5 - 16: Block 3: Storing the outputs of security


After that, the decision module is used to assign passengers that have been

chosen for a secondary security check (at random), as shown in Figure 5-17.

Moreover, Figure 5-17 shows the processing time distribution of passengers selected

for the random explosive trace inspection.

Figure 5 - 17: Block 3: Random explosive decision module

Finally, the queue module is used, followed by the activity module to simulate

the random explosive process. From the graph below, it can be seen the passenger

processes based on the FIFO method, while the distribution of processing time is from

~Erlang (0.1, 0.2), as shown in Figure 5-18.


Figure 5 - 18: Block 3: queuing and processing time characteristics of random explosive check

5.3.4 Block 4: Duty free

After a decision module is used to distribute entities with 20% for transit

passengers, the rest will be routed to the inbound immigration system through duty

free shops. Thus, this block simulates passengers who wish to shop in the airport duty

free. It is assumed that 40% of total inbound passengers will spend some time at duty

free, with the assumption of uniform distribution (1, 24) minutes (Philip J Kirk,

2013a). Get-Module is used to identify the time that each passenger spends shopping

in duty free, as shown in Figure 5-19.


Figure 5 - 19: Block 4: Assign duty free attributes module

5.3.5 Block 5: Inbound immigration and customs module

The system of immigration consists of two subsystems, including common

immigration counters and SmartGate (Wu et al., 2014). Similar to the processes of

immigration in outbound passengers, arrival passengers must join the allocated queues

formed in front of immigration booths, including Australian and international

passengers, to complete the processes required to enter the baggage area. There are

two different queues in the immigration domain because of the nationality of the

passenger. International passengers, who are not allowed to use SmartGate, unlike

Australian, Swedish and New Zealand citizens, have special queues, which are not

usually crowded with an assumption of 30% of total passengers. In immigration

domain, passengers should present their passport and incoming card, and should get

their visa checked by immigration staff to assess its validity. It is noted here that the

immigration area is considered the first bottleneck in the arrival terminal because of

the length of the queue there (QUT, 2010). The major elements of the inbound

immigration processes are illustrated in Figure 5-20.


Figure 5 - 20: Flow chart of inbound immigration checkpoint process

Figure 5-21 demonstrates the logic design of the inbound immigration system.

The logic can be divided into four sections; each section containing different

functionalities and modules. These modules are the decision module, queue module,

common workstation module and SmartGate module.


Figure 5 - 21: Block 5: Logic design of inbound immigration checkpoint process

At the beginning of simulating immigration counters, a decision module was

used to identify the attributes of SmartGate users and direct them to two ends which

both connected to a queue module, as illustrated in Figure 5-22.

Figure 5 - 22: Block 5: Inbound SmartGate user decision module


Another module in block five is common immigration queue characteristics.

International passengers who are not eligible to use SmartGate services will be held in

front of common immigration counters and served based on the FIFS queuing method,

as showed in Figure 5-23. The queue module is linked to read and write modules to

record the results of each minute during the simulation.

Figure 5 - 23: Block 5: Inbound immigration queue module

Then, the activity module is assigned to configure immigration workstations

which consider the characteristics of processing passengers. As can be seen in Figure

5-24, the distribution of processing at each workstation followed a triangular

distribution where the maximum was 1.5, the minimum was 0.5, and the most likely

value was one minute. It is assumed that each counter can process one passenger at a

time. Also, write and read modules are used to record the output of queue length and

waiting time.


Figure 5 - 24: Block 5: Process characteristics of inbound immigration

The last module of block five is the logic design of SmartGate processing points.

Passengers can be serviced based on the availability of self-process kiosks. The

proposed module is flexible because of the ease of adding/removing SmartGate kiosks

or the parameters of processing time distribution; this is illustrated in Figure 5-25.

Figure 5 - 25: Block 5: Logic design of the SmartGate module


5.3.6 Block 6: Baggage claim module

As passengers exit the system of immigration entry point, they are directed to

the right carousel to collect their baggage (QUT, 2010). Also, passengers may search

for trolleys, which can be found in the luggage hall. The passengers can also get

information about the appropriate carousel by looking at the screens located in the

luggage hall or by inquiring with officers. Figure 5-26 demonstrates the decision

module to identify passengers who have bags that need to be claimed.

Figure 5 - 26: Block 6: Baggage claim decision queue module

Passengers can perform two different discretionary activities in the baggage hall,

including the disposal of quarantined items in designated bins and the use of

restrooms. It is assumed that the distribution of the number of bags = uniform (0, 2),

where the delay time is assumed to be the normal distribution for a given mean value

(µ) of 10 minutes and standard deviation value of three minutes, as shown in Figure

5-27 (Kirk, 2013; Ma, 2013).


Figure 5 - 27: Block 6 baggage claim delay time module

5.3.7 Block 7: Inbound quarantine module

The last stage of the inbound facilitation process is quarantine. In this area, the

incoming passenger card should be presented by all passengers to the customs marshal.

Based on the items written on the incoming card, the customs marshal decides whether

the passenger exits the airport or needs to go to a quarantine or customs check.

Passengers are subjected to a Care Quality Commission (CQC) check if they have

declared one or more of the following:

Food items;

Wooden objects, plants, herbs, seeds, or traditional medicines;

Animals, animal parts, or animal products;

Soil, or articles that been in contact with soil;

Have visited one of the rural areas, been in contact with animals, or been in

the farms outside Australia in the past 30 days;

Have visited Africa or South America in the previous six days.

Passengers who need to go to a quarantine check will join the special queue for

that. Here, the CQC officer calls each passenger and asks them to load their luggage


onto the x-ray machine to scan. The CQC officer will see if the luggage needs to be

opened to do more checking after passing it through the x-ray machine; then, they will

assess the items and decide if these items are permitted or prohibited. If these items

are permitted, the passenger can exit this area; however, if the items are prohibited,

they will be confiscated and the passengers will either be directed to the exit or held

for further interrogation. Figure 5-28 shows the quarantine procedures.

Figure 5 - 28: Flow chart for quarantine process

On the other hand, if passengers have declared one or more of the following,

they will be subjected to a customs check:

AUD $10,000 or more in Australian or foreign currency equivalent;

Duty- and/or tax-free items worth AUD $900 or more purchased or obtained

overseas;

Declaration of prohibited or restricted items, such as medicines, steroids,

firearms, weapons, or any kind of illicit drugs;

Possession of goods/samples for business or commercial use;


Possession of more than 2250 ml of alcohol or 250 cigarettes or 250 g of

tobacco products.

When the customs officer calls the passenger to proceed to the customs counter,

passengers will be asked to put the luggage in the x-ray machine to be checked and

scanned. After that, the customs officer will check whether the luggage needs to be

opened or not and if it needs to be opened, the officer will carry out the search. The

passengers declaring AUD $10,000 or more in Australian or foreign currency

equivalent will be asked to fill in the ‘money movement form’. Those who have

obtained duty- and tax-free items worth more than $900, or more than 2250 ml of

alcohol, 250 cigarettes, or 250 g of tobacco products, will be asked to pay customs

fees or have some of the items confiscated by the customs officer. Passengers deciding

to pay the customs fees will proceed to the cashier for the payment, and then continue

to the exit. In case of a passenger’s refusal to pay the customs fees, the customs officer

will detain the items according to policy and then direct the passengers to the exit

(QUT, 2010). It is noted here that the second bottleneck is the waiting time to undergo

customs/quarantine checks (QUT, 2010).

The quarantine system is built based on two identical sections, where each

section reflects the process and sub-processes of the system. Figure 5-29 shows the

logic chart of block seven, the system including the routing decision module.

Figure 5 - 29: Block 6: Logic chart of quarantine module


To begin with, a decision module is used to determine an attribute of travellers

who need to declare what they bring into Australia. Then the passengers will be split

up depending on their given attributes and directed to the proper line to be get

processed, as shown in Figure 5-30.

Figure 5 - 30: Block 7: Inbound declaration decision module

Each sub-system has its characteristics of holding and processing passengers.

Thus, two separate lanes are created linked with activity modules. Figure 5-31

illustrates the queue module of the declaration lane where passengers are processed

based on the FIFO queueing method. The write and read module is linked with the

queue module to record the output data, such as waiting time and queue length at

processing points.


Figure 5 - 31: Block 7: Declaration queue module

Passengers have been classified into two types during the declaration process.

The first type is the passenger who brings permitted items to the country and

announces them on the incoming card. The activity modules are used to simulate the

declaration process, where the delay time is assumed to follow uniform distribution

between one to five minutes, as shown in Figure 5-32.

(a)

(b)


Figure 5 - 32: Block 7: Inbound immigration queue module

Another type is the passenger who breaks the law by bringing illegal goods into

the country without announcing them. The decision module is used to assign such

passengers and direct them to the next processing point for further inspection. Figure

5-33 shows the secondary inspection process.

Figure 5 - 33: Block 7: Inbound Quarantine queue module


On the other side, there are some other passengers who have nothing to declare.

However, they still need to have their baggage scanned through the x-ray machine.

Therefore, the same model has been configured with the assumption that passengers

are processed depending on the sort of FIFO queue method, as illustrated in Figure 5-

34.

Figure 5 - 34: Block 7: Nothing to declare queue module

Figure 5-35 illustrates the process characteristics in the nothing to declare lane

system. It is clear to see that the maximum items in activities is 1 each time.


Figure 5 - 35: Block 7: Quarantine workstation for nothing to declare line

5.4 INTEGRATED INBOUND AND OUTBOUND PROCESSES

At this stage, the outbound and inbound simulation models can be run

simultaneously. This can be done by simulating the major processes of both sides and

developing passenger attribute modules using flight schedules of arrival and departure

flights, as explained previously in Chapter 3 (section 3.5) and section 5.2.2. The model

can give a broad picture of the status of entire terminal operations and how they impact

each other. It can also be very supportive in identifying any bottlenecks in the system.

However, because of the randomness of the events and that each system works in

isolation, further improvement can only occur by development of the ARM approach.

The development procedures for such a model are discussed as follows.

5.4.1 Advanced resource management (ARM)

This section discusses the development of an advanced resource management

model. Figure 5-36 shows the overview of passenger flow at the international terminal


building, as well as departure processes from check-in to boarding, and arrival

processes from de-boarding to baggage claim and exiting the airport. Many issues

mean that managing airport staff can be difficult (Wu & Mengersen, 2013). One of

these issues is that each process is operated by a different stakeholder, for instance,

check-in and boarding is performed by the related airlines and the ground handlers

(Philip J Kirk, 2013b; Manataki & Zografos, 2009a). Another important issue of

managing terminal staff is that each airport domain behaves differently regarding

skills required to perform particular workstations. Chuin Lau (1996, p. 93) argued that

‘manpower scheduling is concerned with the scheduling of manpower resources to

meet temporal operational requirements in ways that satisfy the goals and policies

imposed by the management, labour union and the government’. Also, each process

has its own rules for allocating staff and ways for sharing the staff resources between

related areas. Given the complexity of an international terminal system, algorithms

have been developed to manage staff dynamically across the international terminal

(Ip, Chung, & Ho, 2010).

Algorithms can be divided into two different levels:

Algorithms for managing staff that are working on one system

(outbound/inbound) and can be shared with another process in the same

system, such as airline staff responsible for operating check-in, and could

share with boarding and quarantine;

Algorithms to control staff that work on a particular process that is located

on both sides (outbound/inbound), such as immigration and security (see

Figure 5-36).


Figure 5 - 36: Flowchart framework for ARM model

5.4.2 Mechanism of development of algorithms

The algorithm can be developed as follows:

1. Develop the simulation models, including the physical structure of

outbound and inbound processing checkpoints, as explained in sections

3.5.2. and 5.2.3.

2. Create a global database (see Figure 5-37) for each processing point for the

following reasons:

To store the attributes in a pair of dynamic arrays, one-dimensional to store

the name of each attribute, the two dimensional to store the value of each

attribute name;


To link the parameter table, including number counters, staff ID, the start

shift, and the end shift with ExtendSim.

3. Upload input data, represented by (1) staff attributes, (2) given rules

‘threshold of the queue of particular processes’, and (3) allocation methods.

These input data will be explained in more details in the next sections.

4. Provide a dynamic link between the dialog item (data table or parameter)

and the data source of a global array, as shown in Figure 5-38. Two blocks

named read and write are used to import and export information between

data source and the ExtendSim simulation model and vice versa. The

function of the read block is to read data from a data source to be used in a

model, while the function of the write block is to write data from an

ExtendSim model to a data source ‘global array’.

Figure 5 - 37: Resource allocation dialog for global array


Figure 5 - 38: Dynamic link between parameter tables and ExtendSim

5.4.3 The logic of development algorithms

The proposed approach behaves as dynamic management where the staff can be

controlled and allocated based on a given rule, such as queue length threshold. The

logic that has been followed to develop such algorithms is categorised into three

options. The first option is adding staff to a particular workstation. That can be done

by looping through across the queues of processing points and reading the status of

the queue; if the queue length exceeds the threshold, check if there is available staff in

the resource pool and add them to the first available spot. The second option is

removing the staff from the processing point and shutting down the workstation. This

can be done if the queue length is less than a given threshold.

The last option is more complicated because the staff need to be shared between

related processes. This option is considered after the loop through the staff resource

pool ‘global array’ databases is done and no available staff are found. Then, we need

to share resources between areas based on the queue threshold for each process; for

example, airline staff will be shared between economy and business counters with

more priority to the business class queue. The purpose of developing such algorithms

is to operate airport terminal processes at the optimum level.


5.4.4 Input data of integrated model

1. Staff attributes

Since each process has different characteristics, staff attributes can be classified

into general and specific. As seen in Figure 5-39, there are four types of staff to operate

airport operational elements in this model. The first type is airlines that are responsible

for operating check-in and boarding domains of the outbound system (Philip J Kirk,

2013b). The second type is security which is operated by the airport owner. In

Australia, security is contracted out to expert companies to perform this terminal

domain. The third type is immigration and custom staff who are responsible for

controlling Australian customs and passport checks; this is the responsibility of an

Australian government agency. The last type is quarantine staff for border protection,

which are operated by distinct stakeholders of a government agency.

Figure 5 - 39: Staff attributes for the ARM model

2. Rules given for managing staff

The proposed approach allocates staff based on two different techniques. The

first type is a schedule-based method. In this method, the staff will be allocated and

used all the time to observe what the results are in terms of queue lengths and waiting

time. The second type is demand-based allocation. Unlike the first type, demand based

methods allocate and reallocate staff based on queue lengths rules. Additionally, in

this method, the staff will not only be shared between two processes, but also with the

Staff

Attributes

Type of staff (Airline, Security, Immigration,

Quarantine)

Shift time (start shift, end shift)

staff ID ActivityAttributes flightNum

AssignedStatus

Availability

•0= unavailable•1= available

•2= available, assign first


upcoming passengers to the particular processing point every 30 minutes. Regarding

the rules of queue threshold, action will be taken if queue length reaches X value, as

explained above.

5.4.5 Categories of algorithms

In this section, we will provide further explanation of the development

algorithms, including the aims and objectives, the functionalities, and the conditions

of the algorithms. The algorithms were developed to meet the behaviour of airport

terminal operations that occur in real life and have been classified into two categories

as explained below.

Algorithms developed for non-integrated processes

This section discusses the algorithms that were developed for the airport

terminal elements that are located on only one side, without integration between

outbound and inbound processes. An example of these elements is check-in/boarding

and quarantine. Figure 5-40 illustrates the flowchart of the airline staff dynamic model.

Airline staff is responsible for operating check-in and boarding domains. Also, they

allow passengers to drop their baggage and submit their passports and tickets to be

processed. The operating procedures of airline staff dynamic allocating is based on the

following algorithmic steps:

Step 1: Loop through the economy and business queues and read the queue

status.

Step 2: Check if the queue length exceeds the queue threshold and if there is

available staff in the resource pool, assign them to the first spot. If not, and the

queue length is less than the minimum limit, then remove the staff and shutdown

the workstation.

Step 3: (When step 1 and step 2 are not applicable), share staff between economy

and immigration counters, if there is no available staff in the resource pool. This

procedure is done based on the given policy and always prioritises business

class. The mechanism of moving staff is that if the business queue length

exceeds the queue threshold, such as five PAX, and there is more than one staff

member at economy counters, then move the staff from economy to business

counters and vice versa.


𝑄𝐵 : Queue length at business check-in counters

𝑄𝐸 : Queue length at economy check-in counters

𝑇ℎ𝑄𝐵 : Threshold for business queue

𝑇ℎ𝑄𝐸 : Threshold for economy queue

𝑁𝐴𝑆 : Number of airline staff available in the database of global array

𝐴𝑆𝐶 : Airline staff for check-in counters assigned 𝑗 = 1,2,3 …

𝐴𝑆𝐵 : Airline staff for business counters

𝐴𝑆𝐸 : Airline staff for economy counters

𝐴𝑏𝑔𝑎𝑡𝑒 : Airline staff for boarding gate

G : Departure gate

𝐴𝑆𝑠ℎ𝑖𝑓𝑡__𝑠𝑡𝑎𝑟𝑡 : Start shift of airline staff

𝐴𝑆𝑠ℎ𝑖𝑓𝑡__𝑒𝑛𝑑 : End shift of airline staff

Figure 5 - 40: Flowchart algorithm for airline staff allocation module


The airline staff is also the operators of boarding. Thus, the integrated module

to share staff between check-in and boarding is developed. Before developing the code

of this algorithm, a global array database was created for boarding procedures and to

define and declare the related variables, such as flight code, boarding time, boarding

strategy, and gate number. As we can see in Figure 5-41, the first step of the algorithm

is to check the gate status by looping through all gates and seeing which one is active

based on the departure time. The second step is that if the gate is active, and there is

available staff in the global array database, assign the staff who are not at the end of

the shift.

Figure 5 - 41: Flowchart for integrated module for boarding procedure

Quarantine module

Quarantine is another domain that is located only in the inbound area. In this

domain, staff is shared between declare and nothing to declare lanes. The developed

module supports the two options of the resource allocation technique, including

schedule-based and demand-based methods. Figure 5-42 shows the flowchart

algorithms of quarantine staff management. The procedures for the resource

management approach of quarantine processing can be done via the following steps.


First step is to loop through the quarantine database and check if staff member 𝑖 is

available and not finished his/her shift. Second, every minute of simulation, we check

the demand column of the declare and non-declare processes and see if the queue

exceeds the threshold and staff need to be added or removed. If step one and two are

not applicable, we need to swap staff between both lanes based on the policy provided.

Lastly, for every minute, the code reads quarantine queues statuses and, for example,

if the queue length of the declaration lane exceeds the maximum limit, staff is moved

from nothing to declare to the declaration lane, and vice versa. The proposed

simulation allows for the determination of the following quantities, where:

𝑄𝐷𝑒𝑐 : Queue length to declare at quarantine processing point

𝑄𝑁𝑜_𝐷𝑒𝑐 : Queue length at nothing to declare queue at quarantine

𝑇ℎ𝐷𝑒𝑐 : Threshold of declaration queue at quarantine

𝑇ℎ𝑁𝑜_𝐷𝑒𝑐 : Threshold for nothing to declare queue at quarantine

𝑁𝑅_𝑠𝑡𝑎𝑓𝑓 : Number of quarantine staff available in the database of global

array

𝑅𝑅_𝑆𝑡𝑎𝑓𝑓 : Quarantine staff assigned 𝑗 = 1,2,3 …

𝑅𝑠𝑡𝑎𝑓𝑓_𝐷𝑒𝑐 : Quarantine staff for declare lane

𝑅𝑠𝑡𝑎𝑓𝑓_𝑁𝑜_𝐷𝑒𝑐: Quarantine staff for nothing to declare lane

𝑅𝑠ℎ𝑖𝑓𝑡__𝑠𝑡𝑎𝑟𝑡 : Start shift of quarantine staff

𝑅𝑠ℎ𝑖𝑓𝑡__𝑒𝑛𝑑 : End shift of quarantine staff


Figure 5 - 42: Flowchart algorithm of quarantine staff management module


Algorithms for integrated processes

Security staff management module

This section discusses the integrated model that is used to manage staff working in the

processes that are located on both sides of an international terminal. An example of these

processes is immigration and security screening. As explained earlier, security is operated by

the owner of the airport, but in Australia, security is contracted to third-party companies to

process passengers. The developed algorithm considers a VIP lane in the security system to

process diplomatic passengers. Moreover, the disembarking passengers from arrival flights

who have connecting flights have to pass through inbound security before reaching their

outbound boarding gate. Thus, the security staff need to be shared, not only between two lines

in the outbound system, but also with inbound security to process transit passengers. The

algorithm of the security sharing resource is presented in Figure 5-43.

𝑄𝑑𝑖𝑝 : Queue length at diplomatic security screening checkpoint

𝑄𝑛𝑜𝑛_𝑑𝑖𝑝 : Queue length at non-diplomatic security screening checkpoint

𝑄𝑖𝑛_𝑆𝑒𝑐𝑢𝑟𝑖𝑡𝑦 Queue length for inbound security

𝑇ℎ𝑑𝑖𝑝 : Threshold for diplomatic security screening checkpoint

𝑇ℎ𝑛𝑜𝑛_𝑑𝑖𝑝 : Threshold for non-diplomatic security screening checkpoint

𝑇ℎ 𝑖𝑛_𝑆𝑒𝑐𝑢𝑟𝑖𝑡𝑦 Threshold for inbound security screening checkpoint

𝑁𝑆_𝑆𝑡𝑎𝑓𝑓 : Number of security screening staff available in the database of global array

𝑆𝑆_𝑆𝑡𝑎𝑓𝑓 : Security staff assigned 𝑗 = 1,2,3 …

𝑆𝑆_𝑑𝑖𝑝 : Security staff for diplomatic lane

𝑆𝑆_𝑛𝑜𝑛_𝑑𝑖𝑝: Security staff for non-diplomatic lane

𝑆𝑆_𝑖𝑛𝑏𝑜𝑢𝑛𝑑 : Security staff for inbound lane

𝑆𝑆𝑠ℎ𝑖𝑓𝑡__𝑠𝑡𝑎𝑟𝑡 : Start shift of security staff

𝑆𝑆𝑠ℎ𝑖𝑓𝑡__𝑒𝑛𝑑 : End shift of Security staff


Figure 5 - 43: Flowchart 1-2 of security resource allocation management


For the demand-based allocation method, the procedure of moving and managing

security staff is summarised based on the following algorithmic steps:

i. Create a global array database to declare variables related to staff attributes, such as,

start shift, end shift, staff availability. Also, global array is used to exchange interface

data with the internal data structure needed for ModL programming.

ii. Read the queue statuses of both sides, inbound and outbound, every one minute of a

simulation run and record which queue status is exceeded.

iii. Determine the available staff members who are not in the end of their shift and assign

them to the area where the queue has reached its maximum limit, and remove staff

if the queue is less than the minimum limit. If steps one and two are not applicable,

move to step four.

iv. Share the staff based on the given rules and priority. According to Odoni and de

Neufville (1992), the departure process, which sometimes involves services

provided to transit passengers, typically requires a significantly longer time than the

arrival process. The queue length of outbound security will be observed; if the queue

exceeds the maximum then move staff from inbound to outbound security screening

processing points, and vice versa, as shown in Figure 5-44.

Figure 5 - 44: Flowchart 2-2 of security resource allocation management


Immigration inbound and outbound module

Figure 5-45 illustrates the flowchart of immigration resource management. The same

logic is also proposed for immigration resource allocation. As previously explained, priority is

always given to the outbound system because the time of departure is scheduled, which means

that passengers might miss their flight if they do not arrive on time. The procedures for

allocating staff work as follows:

1. Check the available staff and unassigned staff from finished shifts.

2. Read the queue statuses for both sides and add/remove staff based on the policy

provided by airport operation management.

3. Move staff between outbound and inbound because the queues exceed their limit and

there is no staff available. This can be done by continuously looking at the outbound

immigration queue and observing that if the queue length at outbound immigration

exceeds the given policy and there is more than one staff member at inbound

immigration, then move staff from inbound to outbound, and vice versa, where:

𝑄𝐼𝑚𝑚𝑖_𝑖𝑛𝑏𝑜𝑢𝑛𝑑 : Queue length for inbound immigration processing point

𝑄𝐼𝑚𝑚𝑖_𝑜𝑢𝑡𝑏𝑜𝑢𝑛𝑑 : Queue length for outbound immigration processing point

𝑇ℎ𝐼𝑚𝑚𝑖_𝑖𝑛𝑏𝑜𝑢𝑛𝑑 : Threshold for inbound immigration queue

𝑇ℎ𝐼𝑚𝑚𝑖_𝑜𝑢𝑡𝑏𝑜𝑢𝑛𝑑 : Threshold for outbound immigration queue

𝑁𝐼𝑚𝑚𝑖_𝑠𝑡𝑎𝑓𝑓 : Number of Immigration staff available in the database of global array

𝐼𝑚𝑚𝑖_𝑆𝑡𝑎𝑓𝑓 : Immigration staff assigned 𝑗 = 1,2,3 …

𝐼𝑚𝑚𝑖𝑠𝑡𝑎𝑓𝑓_𝐼𝑛 : Inbound immigration staff

𝐼𝑚𝑚𝑖𝑠𝑡𝑎𝑓𝑓_𝑜𝑢𝑡: Outbound immigration staff

𝐼𝑚𝑚𝑖𝑠ℎ𝑖𝑓𝑡__𝑠𝑡𝑎𝑟𝑡 : Start shift of immigration staff

𝐼𝑚𝑚𝑖𝑠ℎ𝑖𝑓𝑡__𝑒𝑛𝑑 : End shift of immigration staff


Figure 5 - 45: flowchart algorithm of immigration resource allocation


5.5 CHAPTER SUMMARY

This chapter introduced the extension of an international terminal simulation model of

inbound flow processes. The model considered the main inbound processes and transit

processes, including passengers disembarking, security screening for transit passengers,

immigration, and quarantine. The ExtendSim software was utilised to model the inbound flow

processes, and these processes were built on the basis of hierarchy blocks.

Furthermore, this chapter presented a novel integrated simulation model, ARM, for an

entire international airport terminal model. The development model, for the first time, clearly

presents the dynamic resource allocation approach to the overarching model, which has helped

the model to be more accurate and mimic the real-life scenarios in the international airport

terminals. The staff can be utilised based on the selected method, either schedule-based or

demand-based, and can be moved from land-side to air-side according to the given policy

reflected by the queue thresholds.

In the next chapter, a case study is conducted to demonstrate the capability of the

simulation model and how well it reflects real-life scenarios. The case study is conducted by

obtaining actual data from King Khalid International Airport (KKIA), Riyadh, Saudi Arabia.

158 Chapter 6: Case Study - Validation of the Simulation Model

Chapter 6: Case Study - Validation of the

Simulation Model

6.1 INTRODUCTION

This chapter demonstrates that the proposed simulation modelling approach can

accurately represent an actual airport terminal. As discussed, airport terminals are a

dynamic environment involving several independent and interconnected sub systems

that are in communication with each other such as X-ray inspection, check-in, and

passport control facilities (Yamada et al., 2017). This chapter’s validation activities

were confirmed by comparing simulation results with the real data provided by King

Khalid International Airport (KKIA) – Riyadh, Saudi Arabia. Several runs were made

to determine the stochastic variability of the model. In this case study, four types of

simulation outcomes including the average and maximum waiting time in the queue

and the average and maximum cycle time at each departure facility were considered.

The given data only concerns departure processes including demand

characteristics and operational characteristics such as the airport flight schedule,

processing time, and number of service counters for processing points. Hence, the

validation processes take into consideration the outbound simulation model discussed

in Chapter 3, without the interaction with the inbound passengers

The King Khalid International Airport (KKAI) is the second largest airport in

the Kingdom of Saudi Arabia after King Abdul-Aziz in Jeddah. It is located 25 km

north of the capital city of Riyadh (Almuharib, 2014). The airport contains four

international passenger terminals, of which Terminals 1 and 2 are currently in use.

Terminal 1 is used for all international flights excluding those operated by Middle East

Airlines, Air France in addition to Saudia and Flynas use Terminal 2. Figure 6-1

displays an aerial view of the layout of the airport including terminal 1 and 2, which

are the first two from the top.

Chapter 6: Case Study - Validation of the Simulation Model 159

Figure 6 - 1: The terminals and runways of the King Khalid international airport.

According to collected statistics, KKIA has observed a steady increase in the

numbers of passengers traveling by air over the last 10 years. In 2016, the airport

handled more than 23.4 million passengers, as shown in Figure 6-2. Passengers can

access the airport by three transportation modes: the first type is a private car with

55.5% of the total of travellers and the second is a taxi with 42.2%. The third type is

the Saudi Public Transit Company, known as SAPTCO; this company transported only

2.3% (Alhussein, 2011).

Figure 6 - 2: Passenger movement numbers at KKIA from 2005-2016 (Statista, 2019)


Both terminals have the same layout, consisting of two levels where level one

deals with the inbound processes and level two with the outbound processes. By

comparing the design of KKAI terminals with Brisbane International Airport,

departing passengers can reach the departure level through elevators that take the

passengers from the basement level via the arrival level. The outbound processes start

with the security check of luggage by an X-ray machine.

The second process is dropping bags onto a conveyor belt to transfer them to

assigned flights and get checked in. When passengers complete the check-in process

they must pass through passport control counters prior to undergoing the security

screening process. Passengers then enter a departure holding area to await flight

boarding. Figure 6-3 illustrates in greater detail passenger flow types in KKIA. This

case study focuses on the international departure processes.

Figure 6 - 3: Scheme of passenger flow types at KKIA terminals (Kloosterziel et al., 2009).


6.2 KKAI OPERATIONAL DATA

To validate the developed simulation model presented earlier in this thesis,

operational data were taken from KKAI including flight schedule, processing time,

waiting time and physical resources. Flight schedule information is provided for each

terminal, including the number of flights, departure time and the number of passengers

for each flight. In Terminal 1 for instance, the number of flights is 38, the total number

of passengers is 5208 while the total number of passengers in Terminal 2 is 7501 per

day. Based on the given flight schedules and the total number of passengers travelling

per day for each terminal, this airport is classified as a small terminal and a medium

terminal respectively.

Other required input data of the airport’s terminal model are the data concerning

the service process and the physical resources of outbound processes. The KKIA

operations management was contacted to provide information associated with airport

terminal physical and processing time at each terminal facility. In KKIA, the

processing time and waiting data were collected by the KKIA operational staff through

observation of 15 passengers at each facility. The processing time data were analysed

using Extendsim statistical fit analyser in order to understand the processing time

distribution function for each service as the input for the simulation model. Figures 6-

4 and 6-5 illustrate the analysis of processing time data for Terminal 1 and Terminal

2.

To provide further analysis, Figures 6-4(a) and (b) present the scatter of the input

data and the fitted density of check-in facility of Terminal 1, where the x-axis of Figure

6-4(a) and (b) graphs represent the processing time in minutes. It can be observed that

the maximum processing time is 6.26 minutes while the mean (𝜇) 𝑖𝑠 2.15 minutes.

The processing time at check-in follows the Weibull distribution expression of

WEIBULL (0, 1.55, 2.4). Similarly, Figures 6-4(c) and (d) illustrate the scatter of the

input data and the fitted density of security screening facility of Terminal 1. The time

passengers need to be processed at security screening is illustrated as an Inverse

Weibull with a sample mean of 0.721 and standard deviation of 0.584. Additionally,

the distribution of passport control was found to follow Inverse Weibull as well, with

the sample mean of 0.521 and standard deviation of 0.409. In this airport, the greatest

amount of lost time incurred by passengers was at check-in, with the average


passenger waiting time in this domain being 15.95 minutes; while the average waiting

time for security and immigration is 1.25 minutes and 1.76 minutes, respectively.

The time that Terminal 2 passengers spent in being processed is different for all

facilities as shown in Figure 6-5. The time taken to process passengers at the check-in

facility is presented in Figure 6-5(b). The time is illustrated as triangular distribution

expression with a sample mean of 2.16 minutes and the standard deviation is 1.18.

Figure 6-5(c) presents the processing time input data for the security screening facility

of Terminal 2. The time passengers spent at security screening follows the same

distribution patterns as Terminal 1, which is Inverse Weibull with a sample mean of

0.22 minutes. Finally, the processing time at immigration has a triangular distribution

(0, 2.23, 0.15) as shown in Figure 6-5(f).


Figure 6 - 4: Processing time distributions for departure processes of Terminal 1 of KKIA

Check-in processing time, 15 samples mean = 2.152, SD= 1.490

Security screening processing time, 15 samples mean = 0.721, SD= 0.584

Immigration processing time, 15 samples mean = 0.521, SD= 0.409

Security screening processing time Inverse Weibull (0, 1.59, 2.52).

Immigration processing time Inverse Weibull (0, 1.99, 2.6).

Check-in processing time Weibull (0, 1.55, 2.4).

a

b

c

d

e

f


Figure 6 - 5: Processing time distributions for departure processes of Terminal 2 of KKIA

Check-in processing time Triangular (0, 5.11, 1.02).

Security screening processing time, 15 samples mean = 0.22, SD= 0.019

Immigration processing time, 15 samples mean = 0.76, SD= 0.51

Security screening processing time Inverse Weibull (0, 1.66, 9.35).

Immigration processing time Triangular (0, 2.23, 0.149).

Check-in processing time, 15 samples mean = 2.16, SD= 1.181

a

b

c

d

e

f


6.3 MODEL APPLICATION AND SIMULATION PROCESS

The developed simulation model is initialised and customised using the

available real data from the KKIA. The model is modified in terms of the operational

policies followed, and the physical structure of the airport terminals. Therefore, this

section presents the model application outcomes that concern the flows of departing

passengers through outbound processes in order to demonstrate the capabilities of the

model.

Figure 6-6 provides a visual representation of the KKIA departure flow

processes. Every processing point of the outbound system is represented by a module

in the model and each service facility contains a sub-process to perform for different

categories of passengers. Due to this complexity, too many processes are presented in

order to simplify the simulation process.

As for any international terminal, check-in process starts 3 hours before the

departure time where there are two lines to complete check-in, one at the left side of

the main entrance and the other one located at the right-hand side of the terminal. In

the model, each line of check-in service has 13 counters, three for business travellers

and 10 counters for economy travellers. The number of self-service check-in kiosks is

10 for each terminal, as shown in Figure 6-7.

In the simulation model, the default input parameters related to operational

policy were set based on the common practice of the airport provided by the

stakeholder. For example, a new economy check-in counter will be opened if the queue

length exceeds 20 passengers and is it closed if the queue has less than 5 passengers,

while the new business counter is opened when the number of passengers waiting in

the queue is three or more and closed if the queue is zero. Also, the priority for

passengers who arrive late at the airport check-in and are afraid to miss their flight is

considered in the model. After bags are checked-in and passengers receive their

boarding pass, they enter the second mandatory process of passport control, conducted

by immigration.

At the immigration process, there are 14 desks and six self-service kiosks

available for each terminal to process passengers. In the simulation model, seven desks

are available all the time and additional desks are opened if the queue is longer than


30 and closed when the queue is less than 5. Additionally, the input requirements can

be easily altered in order to evaluate ‘what-if’ scenarios (Sargent, 2013).

Figure 6 - 6: Flowchart of KKIA departure flow processes (researcher’s illustration)


In KKIA terminals, the security screening process is closely followed by the

immigration procedure. All passengers and carry-on bags will be checked. At each

terminal, there are six X-ray machines and five metal detectors. In the model, not all

the machines are available most of the time. It is assumed that three machines are

available all the time and additional ones are opened each time the queue length is

increased by 30 and closed if the number of passengers is five or less. Moreover, based

on similar airports, another assumption is made in regard to the percentage of the bags

that fail the X-ray; requiring the passengers to be requested to unpack their items,

which is 10% (Cheng, 2014; Kirk, 2013).

The performance metrics that are used to compare the outputs of the model with

real data are maximum and average waiting time and maximum and average cycle

time. Cycle time can be defined as the total amount of time including the waiting time

at a processing point and the process time. In the simulation model, the cycle time can

be calculated at each process as follows:

Use a block named “set block” in front of processing point.

Creating the attributes value, i.e. Immigration Cycle time.

use a block named information block to read the static information

regarding timing attributes as shown in Figure 6-8.


Figure 6 - 7: Process of calculating cycle time

6.4 SIMULATION RESULTS AND ANALYSIS

This section discusses the simulation results of the two terminals. In order to

demonstrate the model capacities, both terminals have been considered where each

has different input data such as flight schedule and processing time distribution as

shown in Table 6-1. Hence, the simulations were run independently, and the length of

each run is 1440 minutes. Four modules of the simulated international terminal model

will be presented for each simulation run. These modules are the passenger arrival, the

check-in process, the passport control facility and the security screening processes.

This section is structured as follows: section 6.4.1 presents the results of terminal one,

section 6.4.2 discusses the results of terminal two and section 6.4.3 provides a

comparison between the actual data with the simulation results.


Table 6 - 1: Summary of model default parameters at the KKIA international airport.

Parameters Values

Basic parameters:

Time to open check-in counters.

Time to start boarding to the flight prior to departure time

180 min

25 min

Processing parameters:

Open Economy check-in counter if the queue increases by

Open Business check-in counter if the queue increases by

Open security counter if the queue increases by

Open immigration counter if the queue increases by

Close Economy check-in counter if the queue is

Close Business check-in counter if the queue is

Close security counter if the queue is

Close immigration counter if the queue is

Passenger failure rate at metal detector

Percentage of passenger preforming self-service in Immigration

Passengers’ characteristics:

Percentages of business class passengers

Percentages of passengers using self-check-in

Percentages of passengers performing traditional check-in

20

3

30

30

5

0

5

5

10%

30%

15%

12%

80%

6.3.1 Description of Terminal 1 results

Since the passenger arrival pattern at the airport is essential, this is controlled by

the airport flight schedule. Based on the split of the model, passengers may arrive at

the airport using private car, taxi and public bus. Figure 6-9 shows the passengers’

arrival pattern per type of mode, demonstrating arrivals distribution over time (in 10

min time intervals). As discussed earlier, this arrival profile is based on the assumption

that 55.5% of KKIA customers use their own private car, 42.2% of the passengers use

taxis and only 2.3% of airport users would use the Saudi Public Transit (SAPTCO).

This figure illustrates a transition of the flight schedule as main input for the simulation


model. It includes the total number of outbound passengers that show up at the airport

during the day.

Figure 6 - 8: Arrival pattern and profile of Terminal 1 passengers.

There are three severe peaks that happen early in the day (e.g. 02:49-04:00 and

04:23-04:34 am), and another severe peak late in the day, i.e. 21:56-22:52 pm. There

are two smaller peaks, one in the morning (i.e. 07:15-08:00) and another one in the

afternoon (i.e. 15:36-16:00).

Figure 6-10 demonstrates simulation results referring to the three main outbound

processes (check-in, security screening and immigration). Check-in is managed by

airline policies. For example, different check-in rules can be applied by airlines based

on flight types, e.g. whether the flight is international or local. Moreover, travellers

are classified into those who operate by traditional check-in counters or self- service.

An additional classification exists, between economy passengers and business

class passengers. Figure 6-10(a) and (b) show the results of simulation over the time

period (day) where the x-axis demonstrates time in minutes (start from 0 to 1440

minutes) which corresponds to the day of simulation for the check-in facility. Figure


6-10(a) and (b) also illustrate the accumulation of passengers’ patterns during the day,

such as the number of passengers waiting in queue as well as the waiting time of

passengers in minutes, and both are displayed in the y-axes of the figures. It can be

observed that the maximum number of passengers waiting in front of economy

counters is 73 while the maximum number of business passengers is 4. Moreover, the

maximum time passengers wait in the economy queue is 48.77 minutes and the

maximum time business-class passengers wait is 3.63 minutes. However, the average

waiting time is 8.60 min, and 0.2 min for economy and business respectively.

The second important departure process is security screening; passengers have

to scan their items twice; one scan is located before the check-in process and the

second comes after the passport control facility. Figure 6-10(c) and (d) show the

results of simulating the security screening process in accordance with six available

security screening lanes and four random checks. The maximum waiting time is more

than 13 minutes, while the average waiting time is 2.67 minutes. Figure 6-10(c) and

(d) graphically present the accumulation of passenger patterns over time,

demonstrating several peaks occurring over the day with the period of these peaks

concentrating around 02:55-03:25 and 16:56- 17:23 and the maximum number of

passengers in the queue being 70. Finally, Figure 6-10(e) and (f) show the simulation

results for the immigration process, which is responsible for passport control. Figure

6-10(e) and (f) displayed patterns of accumulation of travellers represented by the

number of passengers waiting in queues and the maximum and average of waiting

times. The average waiting time at passport control is 2.2 minutes, while the maximum

waiting time exceeds 10 minutes. Several peaks can be seen and they mainly happened

in the same period as the peaks for security screening.


Figure 6 - 9: a, b Terminal 1 check-in process results; c, d Terminal 1 security screening process results; e, f Terminal 1 immigration process results

0 360 720 1080 14400

6.666667

13.33333

20

26.66667

33.33333

40

46.66667

53.33333

60

66.66667

73.33333

80

Time (min)

No.PaxPax In Queue for Check-in

Economy _Q_Leng Busines_Q_Lengt

0 360 720 1080 14400

6.25

12.5

18.75

25

31.25

37.5

43.75

50

Time (min)

WaitingTime(min)Waiting time for Check-in

Economy _Waiting BusinessWaiting

0 360 720 1080 14400

8.75

17.5

26.25

35

43.75

52.5

61.25

70

Time (min)

No.PaxPax in queue for security screeing

13.86407 370.3981 726.932 1083.466 14400

1.147892

2.295784

3.443676

4.591568

5.73946

6.887352

8.035244

9.183137

10.33103

11.47892

12.62681

13.7747

Time (min)

waiting time(min)waiting time for Security check

security waitng…

0 360 720 1080 14400

3.25

6.5

9.75

13

16.25

19.5

22.75

26

29.25

32.5

35.75

39

Time (min)

No.PaxPax in queue for immigration

Immig_queue_L SmartGate_Q_L

12.49861 369.374 726.2493 1083.125 14400

1.275665

2.55133

3.826994

5.102659

6.378324

7.653989

8.929654

10.20532

Time (min)

Waiting time(min)WaitingTime for immigration

Immig_Waiting

e a c

b d f


6.3.2 Description of Terminal 2 results

This section first discusses passenger arrival patterns at Terminal 2 of KKIA.

The second part of this section will discuss the simulation results for the main

departure facilities. Figure 6-11 presents the departing passenger arrival profile and

the distribution of entering Terminal 2 of KKIA. As explained earlier, the passenger

arrival pattern is controlled by the airport flight schedule. Two main peaks can be seen

during the day; one occurs early morning around 05:46-06:38 and another one occurs

in the afternoon between 13:04-14:48. Another three smaller peaks were observed;

two of them happened in the morning 08:43-09:05 and 10:40–11:10, and the last one

occurred in the evening around 21:57-22:27.

Figure 6 - 10: Arrival pattern of Terminal 2 passengers and entering Terminal 2 distribution.

Figure 6-12(a) and (b) illustrate simulation results during the day for the check-

in process, including the volume of passengers waiting in queue and waiting time spent

in queue.. The maximum waiting time reaches 14.29 minutes while the average

waiting time is 4.13 minutes. Furthermore, the maximum number of passengers

waiting in queue was shown at 360 minutes of x-axis of simulation time to be 43,

which aligns with the first peak of the departing passenger arrival profile.


Similar to the Terminal 1 security screening process, passenger baggage needs

to be inspected twice. Figure 6-12(c) and (d) presents the simulation results including

the number of passengers waiting in queue and the waiting time in minutes. The

maximum waiting time is 12.26 minutes, although the average waiting time is 1.77

minutes. There was a severe peak in the early hours of the day around 05:46-06:38,

lasting about a half-hour with a maximum passenger waiting time of 12.29 minutes.

Figure 6-12(e) and (f) illustrated the results of simulation for the passport control

process of Terminal 2. Two severe peaks happen in the early hours of the day around

02:30-03:00 and 05:46-06:38 and the number of passengers in the queue reaches its

maximum of 176 in the second peak. The average waiting time of passengers in front

of passport control is 5.25 minutes and the maximum waiting time is 17.18 minutes.

These results are aligned with the arrival patterns illustrated in Figures 6-9 and 6-11.


6.855429 365.1416 723.4277 1081.714 14400

1.786762

3.573524

5.360286

7.147047

8.933809

10.72057

12.50733

14.29409

Time (min)

WaitingTime(min)Waiting time for check-in

Economy _Waiting BusinessWaiting

0 360 720 1080 1440-1.453061

13.38027

28.21361

43.04694

57.88027

72.71361

87.54694

102.3803

117.2136

132.0469

146.8803

161.7136

176.5469

Time (min)

No.PaxPax in Queue for immigration

Immig_Q_length

19.25026 374.4377 729.6251 1084.813 1440-0.1258835

1.306041

2.737965

4.169889

5.601814

7.033738

8.465662

9.897587

11.32951

12.76144

14.19336

15.62528

17.05721

Time (min)

WaitingTime(min)Waiting time for immigration

Immig_waitingTi Immig_SmartGate

21.30821 375.9812 730.6541 1085.327 14400

1.021284

2.042568

3.063852

4.085136

5.10642

6.127704

7.148988

8.170272

9.191555

10.21284

11.23412

12.25541

Time

waiting time(min)Secruity screening waitingTime

security waitng…

e a

0 360 720 1080 14400.2234043

5.473404

10.7234

15.9734

21.2234

26.4734

31.7234

36.9734

42.2234

Time

No.PaxPax In Queue for check-in

Business_Q_leng Economy _Q_Lengt

0 360 720 1080 14400

4.5

9

13.5

18

22.5

27

31.5

36

40.5

45

49.5

54

Time (min)

No.PaxPax in queue for security screening

Security Q_leng

c

b d f

Figure 6 - 11: a, b Terminal 2 check-in process results; c, d Terminal 2 security screening results; e, f Terminal 2 immigration process results


6.3.3 Results analysis and discussion

In this section, we will discuss the validation process of the developed model

using empirical and statistical validation. Statistical validation is made by comparing

the simulation results with the observed data from the airports, as collecting real data

from airports is time consuming and an extremely intensive job (Livingstone, Popovic,

Kraal, & Kirk, 2012). Additionally, there is a limitation in the required data related to

the average waiting time and the total of cycle time at each outbound processing point

(Cheng, 2014). Because of the limited actual statistical data available for specific

airport terminals, past studies could only validate limited outcomes of the simulation

with available data of airports in terms of passenger flow. This is done to demonstrate

that how simulation model can utilised to analyse the passenger flows in an

international airport as a response to question one.

As explained earlier, two different terminals of KKIA were simulated, each with

different flight timetables as well as different time distributions. Table 6-2 compares

the simulation results and real data of Terminal 1 obtained at each processing point.

There is high variation between the average waiting time of the simulation and actual

data, while the variation is slightly less in the average cycle time, because of less

processing time, especially in the check-in and security screening facilities. However,

the model can provide better results and reflect the actual situation in the immigration

process. Since this terminal operates international flights, there are variances in

policies performing each flight and its passengers; for instance, there might be an extra

check and some might have stricter security check-ups (Ma, 2013). Furthermore, this

terminal provides an extra baggage check before check-in, which require more time

compared with Australian airports.

For further analysis, the same processes have been conducted for Terminal 2 of

KKIA. Table 6-3 illustrates the comparison between the actual data and simulation

results in terms of the average waiting time and total cycle time at each processing

point of Terminal 2


Table 6 - 2: Comparisons of waiting time in queue and cycle time at check-in, security and immigration

between the simulation data and real time data of Terminal 1.

Domain

Terminal 1

Waiting time (min) Cycle time (min)

Actual Simulation Relative error Actual Simulation Relative

error

Check-in

Max 57.88 48.77 18.67% 64.14 56.31 13.91%

Average 15.95 8.60 85.46% 18.103 11.38 59.6%

Security

Max 3 13.77 78.21% 4.54 20.22 77.54%

Average 1.25 2.67 53.18% 1.97 3.87 49.10%

Immigration

Max 7 10.21 31.4% 17.4 28.81 40%

Average 1.76 2.2 20% 2.76 2.84 2.81%

The results of the average waiting and average cycle times at both check-in and

security screening facilities are slightly better in terms of less variation compared with

Terminal 1 results. Though in the case of the immigration facility, the variation is high

compared with the results of Terminal 1.


between the simulation data and the real time data of Terminal 2.

Domain

Terminal 2


Actual Simulation Relative error Actual Simulation Relative

error

Check-in

Max 9 14.29 37.02% 18.37 17.24 6.55%

Average 3.68 4.13 10.90% 6.16 6.54 5.81%

Security

Max 3 12.26 75.53% 10 17.70 43.50%

Average 0.88 1.77 50.28% 1.55 2.02 23.27%

Immigration

Max 5 17.18 70.90% 9.16 18.29 49.92%

Average 1.56 5.24 70.22% 2.78 6 53.67%

It is believed that the developed model would be more appropriate for such

airports as Brisbane International Airport, since comparisons of the results of the

simulation model with the observed data collected from Brisbane International Airport


by Kirk (2013), were within acceptable differences (Table 6-4). According to Cheng

(2014), with a benchmark showing differences between the average actual time and

the average of simulated time being less than 2 minutes, the model would reflect the

actual situation except for the average cycle time of the security screening facility.

Despite limitations of the model applicable to KKIA airport, given the flight schedule,

the simulation model can provide valuable information for airport management about

any potential congestion as well as the peaks aligning with the departing passenger

arrival patterns shown in Figures 6-9 and 6-11. In order to address the limitations of

the developed simulation model to be more applicable in KKIA international airport,

further improvement is required. For example, other processes such as an extra

baggage check occurred before check-in processes should be considered


between the simulation data and the real time data at Brisbane International Airport.

Domain

Brisbane Airport


Actual

(Kirk, 2013) Simulation

Relative

error

Actual

(Kirk, 2013) Simulation

Relative

error

Check-in

Max 42.81 41.68 2.71% 53.56 46.34 15.58%

Average 12.88 12.06 6.80% 16.65 16.25 2.46%

Security

Max 17.09 26.21 34.80% 21.06 27.8 24.24%

Average 3.75 4.02 6.72% 6.88 4.7 46.38%

Immigration

Max 15.46 25.68 39.79% 18.58 26.17 29%

Average 4.8 5.67 15.34% 6 6.83 12.15%

Ma (2013) argued that the validation of passenger flow at airports depended on

three main elements: passenger flow speed within the terminal, immediate occupancy

by passengers at particular areas in the terminal, and the routing decision by

passengers. Thus, to have more accurate validation, video recording might be used to

collect relevant data such as the waiting time and the cycle time at each terminal

facility. Video cameras can obtain samples and volumes of samples more efficiently

and accurately, because they often record a full day, week or even a month of data.


6.5 CHAPTER SUMMARY

In this chapter, two case studies of passenger flow simulation in the airport of

King Khalid International Airport in Riyadh have been demonstrated. Due to the

available data, the main focus was the outbound process including check-in, security

screening and immigration. The first study was conducted at Terminal 1 of KKIA with

its own flight schedule. It could be clearly seen from the departing passengers’ arrival

pattern graph that there are several peaks occurring during the day. The second study

was undertaken at Terminal 2 of KKIA. Four categories of outputs were generated;

average/maximum waiting time and average/maximum cycle time. In both studies, it

can be observed that bottlenecks at departure facilities occurred in conjunction with

peaks in departing passenger arrival patterns. It could be clearly seen that simulation

results align with the arrival pattern.

However, by comparing the results of a simulation model with actual data of

both terminals, results showed that there is great variation in the check-in and security

average waiting times of Terminal 1, around 85% compared with the real scenario, and

this variation decreased to 59% in the cycle time results for the same facility. The

results of the immigration facility had the lowest variation at about 20% and 2.81% for

the average waiting time and cycle time, respectively. On the other hand, the

simulation results of Terminal 2 showed that check-in average waiting time and

average cycle time have lower variation around 10% and 5.81%, respectively.

Contrary to the results of the immigration facility of Terminal 1, the variation in the

results of average waiting time and average cycle time is high at about 70.22% and

53.67%, respectively.

Comparison of the results of the proposed simulation model with actual data

showed that the model is more applicable for local airports, such as Brisbane

International Airport, than external airports such as KKIA. The results of the model

might be improved by changing the input requirements and running ‘what-if’

scenarios; since airports are exposed to external effects and the developments of air

traffic, it is better to repeat and validate the results of the study frequently (Rauch &

Kljajić, 2006). According to Sargent (2013), there is no particular group of

experiments that can be applied easily to find the model correctness. Conversely, the

model provided sufficient accuracy for local international airports such as Brisbane,

where results demonstrated that the simulation reflects the actual situation. It is known


that the accuracy of the model for a particular case does not guarantee that the same

model can be valid everywhere in its related domain (Sargent, 2013).

Chapter 7 of this thesis presents the integrated simulation model within the

advanced resource management (presented in Chapter 5) and demonstrates how the

proposed model can be used to investigate the effect of different staff allocation

techniques on both sides of the airport terminal (outbound and inbound) processes.

The effects of these staff allocation methods can be understood by comparing the

simulation results under different settings or scenarios.

182 Chapter 7: Application of Advanced Resource Management (ARM)

Chapter 7: Application of Advanced

Resource Management (ARM)

7.1 INTRODUCTION

Passenger satisfaction has become a significant concern for modern airports. In

daily operations, having efficient and effective resource allocation methods in place

can significantly improve the performance of international terminal operations and

enhance passenger satisfaction. Despite recent research on developing resource

allocation models for one airport system domain (e.g. inbound or outbound), there is

little research on the management and allocation of staff in entire international

terminals. For example, if more personnel are assigned to immigration counters for

passport checking from another terminal domain, the process can be faster and high

levels of service will be offered to passengers. This movement of personnel, however,

might lead to a large number of passengers accumulating in the other domains which

could negatively affect levels of service. Thus, there is a need for a decision support

tool for airport terminal planning and operations management to significantly enhance

the efficiency of the overall system (Manataki & Zografos, 2009).

The Advanced Resource Management (ARM) algorithms presented in Chapter

5 are added to the simulation model to study the variable and complex environment of

operational policy. The ARM was developed to be an integrated system used for

arranging resources, identifying the proper resource and allocating them throughout

the model (Imagine That Inc, 2013). It was used to investigate the influence of

different staff allocation techniques on both sides of the airport terminal (outbound

and inbound) processes, by doing this question three will be answered. The effects of

these staff allocation methods can be understood by comparing the simulation results

under different settings or scenarios. This is done by analysing the results obtained

from allocating staff according to scheduled based allocation (static approach) and

demand-based allocation (dynamic approach) under the same conditions and input

variables. The overall objective of the developed ARM is to enhance passenger

satisfaction through reasonable wait time processing at the lowest cost possible

(minimal staff hours). The analysis of the proposed approach is based on processing

Chapter 7: Application of Advanced Resource Management (ARM) 183

activities including check-in, inbound and outbound security control, inbound and

outbound immigration, and quarantine and boarding procedures.

Moreover, this chapter discusses an integrated analytical model for the

optimisation of resources of airport terminal. The objective of the developed

mathematical model is to minimise the cost of waiting time in queues by determining

where the additional resources should be placed. In this mode, there are different

resources types, each with different amounts, hence, another aim of the proposed

model is to obtain resources at minimum total cost.

Section 7.2 demonstrates staff management techniques including the general

inputs of the experiments. Section 7.3 analyses the influence of static and dynamic

allocation methods on the performance of the international terminal. Section 7.4

presents the best policies for dynamic allocation at airport activities. Finally, the

analytical model is discussed and introduced in section 7.5.

7.2 OVERVIEW OF AIRPORT RESOURCE MANAGEMENT

The focus of this thesis is the effective management of passenger flows inside

the airport terminal by improving resource allocation. For example, opening new

counters at the security control and adding extra staff to prevent long queues. This

research aims to identify the best possible techniques and policies of staff allocation

to ensure the desired balance between demand and service quality. Since the terminal

is a highly complex system, with two significant types of passenger flows processes,

the development of a decision-support tool for managing overall staff allocation

processes is challenging. To simplify this issue, the model has been divided into three

sub-models. The first sub-model simulates outbound passenger flow processes as

illustrated in Chapter 3. The second sub-model simulates inbound passenger flow

processes as explained in Chapter 5. The last sub-model involves developing novel

algorithms for resource allocation for both inbound and outbound passenger processes.

It is expected that this model will provide more accurate outcomes in representing

flows of arriving and departing passengers and service processes. The model can be

used to explore a wide range of what-if scenarios which helps with more effective

decision-making in airport terminal planning, operations and management. By doing

this question one can be addressed.


7.2.1 Model demonstration

The proposed model is developed to accurately mimic real-world situations

since terminal operations are dynamic and involve a range of services, e.g. check-in,

passport control, boarding pass control, security screening. To demonstrate the

model’s capability, a set of experiments was conducted to analyse staff allocation

methods under the same conditions and input data. The following phases were

performed:

i. Two experimental scenarios, each with a total of 81 simulation runs, were

conducted to test the variable of staff number for each processing station.

Experiment 1 considers the base case (static method) of allocating staff,

while experiment 2 allocates the terminal staff dynamically, i.e. staff will

be assigned when and where needed.

ii. To obtain insight into stochastic variations, the simulation was run more

than 200 times for each method (static and dynamic) under the same input

data.

7.2.2 General input data

For comparative purposes, the experimental scenarios were conducted under the

same conditions. The first type of data is the flight schedule for both types of flight

departure and arrival. The second type of data relates to the operational characteristics

of the system, for example, the processing distribution at the various airport processing

points and the number of checking points available. Table 7-1 summarises the

operational input data.


Table 7 - 1: Summary of common operational input data for the experiments

Airport

domain

Flow Direction of

PAX

Staff Processing time

Check-in Outbound 8-10 Delay time at check counters =

0.2 min ∗ # of bags

Security

screening

Outbound/inbound 3-5 ~ Tri (0.2, 0.5, 0.75)

Immigration Outbound 6-8 ~ Tri (0.5, 1, 2)

Boarding Outbound Serviced by

airline staff

15 PAX /min

Baggage

collection

Inbound - ~ Norm (10, 3)

Quarantine

Declare

No-

declare

Inbound 12-16 ~ Uniform (1, 5) for declaration and

~ Tri (0.5, 1, 0.75) for nothing to

declare

7.3 SIMULATION RESULTS AND ANALYSES

This section discusses the results of the initial experiments conducted to

demonstrate the developed dynamic staff allocation and to ensure reliability of the

results. Since there are four types of staff including airline, security, immigration and

quarantine, the simulation was run with various staffing levels for each process to

ensure that the terminal is operated with all possible scenarios as shown in Table 7-1.

The results were collected from 81 multi-simulation runs for the two types of

allocation methods: (1) allocating staff according to static base, (2) allocating staff

dynamically—staff allocated when needed. For each simulation run the KPIs

considered in this study were recorded for each processing point, both outbound and

inbound. The KPIs include maximum/average queue length, maximum/average

waiting time, number of late flights/average time delay for late flights and total staff

hours. The output of each scenario was recorded using the ExtendSim database global

array then exported to Excel for analyses as illustrated in Figure 7-1.


Figure 7- 1: Example of ExtendSim database for output data

The airport terminal is operated as one single system with multiple elements

located in the different types of outbound and inbound passenger flow processes. For

this study, staff types were grouped into four different groups: airline staff,

immigration staff, security staff and quarantine staff. The number of scenarios was set

based on the Cartesian product of the staff sets which is defined as (Imrich, Klavžar,

& Rall, 2008):

𝐴 × 𝐵 × 𝐶 × 𝐷 = (𝑎, 𝑏, 𝑐, 𝑑)|𝑎 ∈ 𝐴, 𝑏 ∈ 𝐵, 𝑐 ∈ 𝐶, 𝑑 ∈ 𝐷.

It is assumed that set A = airline staff = 8, 9, 10, set B = Security staff = 3,

4, 5, set C = Immigration staff = 6, 7, 8, set D = Quarantine staff = 6, 7, 8. Since

each set has three elements, the Cartesian products of these four sets have 81 ordered

pairs. The same order of pairs will be used for both experiments for two reasons, (i) to

understand significance of the dynamic allocation approach and how it improves the

efficiency of all processes, and (ii) to obtain insights into the best policy for airport

operations and management.

7.3.1 Static Allocation Base Case method

This section discusses the simulation results from the Static Allocation Base

Case (SABC) method. The SABC method has many variations and is projected to


provide shorter waiting time for passengers as the allocation procedures of SABC are

not controlled by adding and removing staff polices discussed in section 5.4.2.

Inefficient use of airport resources and potentially significant financial losses due to

greater operating hours represented by staffing hours usually result from this method.

7.3.1.1. Check-in and boarding

The effects of SABC on check-in and boarding were investigated with respect

to the average passenger waiting time in queues and the total staffing hours at check-

in counters to serve passengers over 24 hours. Figure 7-2 demonstrates the average

waiting time at check-in counters and the total staffing hours for all 81 possible

scenarios. There is fluctuation in economy passenger average waiting time values over

the first 28 scenarios when there are only eight staff available. The average waiting

time then stabilises for the remaining scenarios. The average waiting time spent at

check-in lies between 5-7.6 minutes. There was also a slight increase in the number of

staff working hours at check-in for the same scenarios, where the average and

minimum working hour’s values are 889 and 772 hours, respectively.

Figure 7- 2: Check-in average waiting time using the SABC method

Figure 7-3 presents the status of boarding procedures employing the SABC

method. Two variables were considered: the number of delayed flights and average

time delay for these flights. In general, the time from the start of boarding to flight

departure is about 30 minutes (Cheng, 2014). Hence, any flight departing more than


30 minutes after the original time is considered a delayed flight. This figure also

demonstrates considerable fluctuation in the number of delayed flights and the average

delay time. The number of delayed flights reached the peak of 18 flights in the 19th

simulation run with an average delay time of 28 minutes.

Figure 7- 3: Influence of the SABC method on boarding procedures

7.3.1.2. Security domain

Figure 7-4 illustrates the impacts of SABC on security screening processes. It

outlines the average waiting time and the total staffing hour status of outbound and

inbound security processes. Similar patterns were found as for boarding procedures,

especially outbound average waiting time and staff hours. There was significant

fluctuation in the outbound average waiting time—reaching a maximum of 21.5

minutes if there are 3 staff members available and a minimum of 1.5 minutes when

there are 5 people available. The inbound average time remained stable for all

scenarios. The maximum total staffing hours stabilised at 120 hours if there are 5 staff

members available and has dropped one and half time to 73 when the system is

operated with only three people.


Figure 7- 4: Security screening average waiting time using the SABC method

7.3.1.3. Immigration domain

This section analyses the impacts of the SABC method on immigration

operational performance. Figure 7-5 demonstrates that the average waiting time for

passengers to be processed by outbound immigration staff reached its highest point of

3.5 minutes in scenarios 1-10, 28-36 and 55-64 when the total number of staffing hours

was 287 hours with 3 staff members assigned to serve passengers. There was high

variation in the average waiting time at outbound immigration due to the irregular flow

rate caused by other processing points such as check-in or security. The average

waiting time of inbound immigration remained stable given the same number of

available staff.


Figure 7- 5: Immigration average waiting time using the SABC method

7.3.1.4. Quarantine domain

Figure 7.6 illustrates the performance of the quarantine system in terms of the

average waiting time. There were consecutive peaks in average waiting time for

quarantine processing when only six staff members were available. The maximum

average waiting time for declaration stations was 57.51 minutes while the shortest wait

time was 23.21 minutes. For non-declaration processing stations, the maximum

average waiting time was 55.88 minutes and the shortest time was 5.25 minutes.

Figure 7- 6: Quarantine average waiting time using the SABC method


7.3.2 Dynamic resource allocation method

This section describes the results from using the dynamic resource allocation

method and compares the results with the SABC method results discussed in section

7.3.1. Applying the dynamic allocation method enhances the efficiency of airport

terminal operational processes. As it can strike a balance between demand and staffing

hours—staff will be allocated to a particular workstation based on the given airport

policy.

7.3.2.1. Check-in and boarding domain

Figure 7-7 displays the average waiting time status at check-in facilities when

the dynamic based approach is used. The dynamic resource allocation method has a

significant influence on the performance effectiveness of the check-in domain. The

average waiting time is a little longer than the SABC, this is because of the given

policy including adding and removing staff based on the queue threshold. The

economy average waiting time decreased gradually from around 16.5 minutes to 13.5

minutes by increasing airline staff from 8 to 9 people and from 13.5 minutes to 11.5

minutes by increasing staff from 9 to 10. Further analysis of the policy of allocating

and reallocating staff will be investigated in section 7.4.

The total staffing hours stabilised with a peak of 326 hours while the lowest

operating hours was 258 hours. These staff operating hour levels are half that from

using the SABC method.

Figure 7- 7: Check-in average waiting time using the dynamic resource allocation method


Figure 7-8 illustrates the impacts of the dynamic resource allocation technique

on boarding time. The dynamic method provided better results in regards to fewer

delayed flights. There was little variation in the number of delayed flights over the

first 28 scenarios when eight airline staff members were allocated to boarding

processes. When more staff were added, the number of delayed flights experienced a

slight drop followed by considerable variation. The highest number of delayed flights

was 13 with 8 staff members available. The maximum delay time for delayed flights

was around 18 minutes with a minimum of 4 minutes.

Figure 7- 8: Influence of dynamic resource allocation method on boarding procedures

7.3.2.2. Security domains

Figure 7-9 shows the average waiting status versus the operation time at security

screening for both outbound and inbound flow processes. Applying the dynamic

allocation approach resulted in stabilised average waiting times for outbound and

inbound security. The average waiting times for outbound and inbound security were

around 5-8 minutes and 7-9 minutes, respectively. The maximum number of staffing

hours was 95 hours and the minimum as 46 hours – both lower than when using the

SABC method (section 7.3.1.2). These results are based on the given conditions, such

as adding and removing staff, if the queue reaches its upper or lower limit. Results can

be improved through performing what-if analysis scenarios to select the best security

system operation policy.


Figure 7- 9: Security screening average waiting time using the dynamic resource allocation method

7.3.2.3. Immigration domain

Figure 7-10 illustrates the influence of the dynamic allocation approach on

immigration processes. There was moderate fluctuation in the average waiting time in

outbound and inbound immigration queues ranging between 6-8 minutes on both

sides. The highest value of staffing hours is 76 hours, while the minimum value is 63

hours. These numbers for both outbound and inbound processes are much lower than

those generated using the SABC method (section 7.3.1.3). Further analysis of the

impacts of different sharing policies will be presented in section 7.4.

Figure 7- 10: Immigration average waiting time using the dynamic resource allocation method


7.3.2.4. Quarantine domain

In the quarantine domain, dynamic and static based scenarios have an almost

similar impact. Figure 7-11 demonstrates that there is congestion in the non-

declaration processing points. The longest average waiting time for declaration is 59

minutes if there are only 5 staff members available. If the upper limit number of staff

(8) was utilised, the average waiting time decreased sharply to below 30 minutes.

Figure 7- 11: Quarantine average waiting time using the dynamic resource allocation method

7.3.3 Comparison of overall impact of static and dynamic allocation models

After analysing each processing point individually in sections 7.3.1 and 7.3.2,

the dynamic approach was found to provide better results because it mimics real-life

scenarios including adding and removing staff and sharing staff between related

processes.

This section compares the results of total average waiting time and the total staff

hours using both methods of allocating staff as shown in Figures 7-12 and 7-13. SABC

results show considerable variation in regard to the outbound average waiting time

compared with the dynamic allocation method under the same inputs, especially, the

number of personnel available. The dynamic approach can significantly impact the

total staffing hours (time spent serving passengers), halving the hours produced by the

SABC method. The proposed dynamic algorithms also have fewer variations

associated with the total staff hours where the minimum number of staffing hours is


625.78 hours and the maximum is 777.15 hours. In comparison, the SABC minimum

number of working hours is 1432 hours and the maximum is 1781 hours.

After employing the dynamic resource allocation method, the total average

waiting time in the outbound system is stable over all scenarios with a maximum of

24.52 minutes and a minimum of 17.38 minutes. The most interesting finding was that

the check-in process plays a significant role in the performance of the entire outbound

system in the case of the dynamic allocation method. According to the literature, the

check-in process is considered the major factor causing delays and congestion at

airport terminals with more than 60% of the total time spent at check-in (Guizzi et al.,

2009; Ma, 2013; Park & Ahn, 2003; Schultz & Fricke, 2011).

Figure 7- 12: Static allocation method results

Figure 7- 13: Dynamic resource allocation method results


7.3.4 Identify the variation of static and dynamic allocation methods

In order to understand the variations of both methods of allocating airport

terminal staff, the model was run over 200 times using the same operational inputs

data described in section 7.2.2 and with a fixed number of staff. Figure 7-14

demonstrates the mean average waiting time and variance using SABC methods while

Figure 7-15 shows the mean average waiting time and variance for the dynamic

resource allocation method. The variance of the SABC has is higher than that of the

dynamic resource allocation method. In addition, the variations in the variance of the

dynamic method occurred around the mean average waiting time while that of the

SABC did not.

Figure 7- 14: Variation in the SABC method

Figure 7- 15: Variation in the dynamic resource allocation approach


However, having the same input, results showed that SABC has lower mean

and variance values, around 4 and 8, respectively, for the first 13 scenarios then

dropped slightly for the rest of the scenarios (mean 2.587, variance 5.853). The mean

and variance of the dynamic allocation approach were 8.223 and 10.206, respectively.

7.4 DYNAMIC APPROACH DEMONSTRATION

The dynamic algorithms presented in section 5.4 have been used to mimic real-

life personnel planning for an international terminal. To demonstrate the approach

capability, a set of experiments using “what if” scenarios were conducted considering

various policies in regard to adding/ removing staff and sharing staff between two

processes. These experiments can demonstrate how the developed dynamic models

improve the performance of the system and how it can be optimally operated under

different operational policies. This investigation can provide a deeper understanding

of best strategy to enhance system performance.

Initially, the queue threshold controls the adding and removing staff rules at each

terminal element. The second phase of sharing staff between related processes is

governed by priority conditions. This is transformed into model input as described in

section 5.4.3. In the next two sections, two more features of the dynamic approach will

be discussed. First, section 7.4.1 demonstrates the impacts of adding and removing

policies on international terminal facilities, followed by section 7.4.2 that investigates

different policies for sharing staff.

7.4.1 Adding and removing staff polices for non-integrated processes

Figure 7-16 illustrates the influence of staff adding and removing policies on

check-in performance, including business and economy counters. Different policies

have been used to evaluate the best policy for adding and removing employees at

check-in. This experiment started with initial operational policies, for example, to add

more staff at business and economy counters when the queue threshold is 5 and 10

passengers, respectively, and to remove one staff member when queue length becomes

2 for business and 10 for economy passengers. Hence, scenarios are combinations of

the following sets of policies:

i. Sets of adding one staff at business check-in if queue length =

1, 2, 3, 4, 5, 6, 7, 8


ii. Sets of removing one staff at business check-in if queue length =

0, 1, 2, 3, 4, 5, 6, 7

iii. Sets of adding one staff at economy check-in if queue length =

10, 20, 30, 40, 50, 60, 70, 80

iv. Sets of removing one staff at economy check-in if queue length =

5, 10, 15, 20, 25, 30, 35, 40

According to Kazda and Caves (2015), the satisfactory length of queue time at

economy check-in is 12 minutes and 3 minutes for business class. Kirk (2013)

observed that 50 of the 71 passengers observed spent the longest amount of time at

check-in with an average time of 17 minutes. Individual times ranged from 2 to 54

minutes.

Figure 7-16 shows that the satisfactory level can be reached in the first three

scenarios with the rule of adding one economy staff member if the queue threshold is

10, 20 and 30, and the policy of queue threshold of other terminal domains is fixed

with the initial operational policies. It occurred again for scenarios 35 and above with

the policy of adding and removing staff for business counters at thresholds of 8 and 0,

respectively, and for economy thresholds of 30 and 5, respectively. The lowest

economy maximum and average waiting times were 48.016 and 7.917 minutes,

respectively, in the first scenario with the initial operational policies discussed earlier.

The lowest maximum and average waiting time in business class was 1.78 and .012

minutes, respectively, for scenario 51 (adding and removing staff set at 1 and 0 for

business class, respectively, and 10 and 40, respectively for economy). Hence,

selection of the best policy is the set of rules that meet the acceptable waiting times

for both business and economy counters.


Figure 7- 16: Check-in facility results of adding and removing staffing policies

As explained in Chapter 5, another non-integrated process of an international

airport terminal domain is the quarantine domain located in inbound flow processes.

In quarantine, 65 different policies were tested to understand the best policy for

optimal quarantine operation in both lanes of quarantine sub-processes; declaration

and nothing to declare lane. The initial operational policy of adding quarantine staff

for this experiment is 10 and 30 passengers for declaration and nothing to declare

lanes, respectively. The initial policy for removal is 10 and 10 passengers for

declaration and nothing to declare lanes, respectively. The same number of policies

used for check-in were used for this experiment. However, there are some differences

regarding rules for the number of passengers waiting in the queue. The following is

the sets of policies:

Sets of adding one staff at declaration if queue length =

10, 20, 30, 40, 50, 60, 70, 80

Sets of removing one staff at declaration if queue length =

5, 10, 15, 20, 25, 30, 35, 40

Sets of adding one staff at nothing to declare if queue length =

10, 20, 30, 40, 50, 60, 70, 80

Sets of removing one staff nothing to declare if queue length =

5, 10, 15, 20, 25, 30, 35, 40

Figure 7-17 demonstrates that the waiting time at each lane is similar for the first

half of the experiments. This means a quarantine desk for declaring is opened when


the number of passengers waiting is greater than 10, 20… 80 and other policies are

fixed with initial values for the first seven scenarios. From scenario 8 to 15, the

selected policy is that a declaration lane desk is closed when the number of passengers

waiting is greater than 40, 35… 5. What is interesting is the variability in queuing time

in the second half of the experiment when the number of passengers waiting at nothing

to declare lanes is greater 80 for opening a desk and 10 for closing the counter. The

best policy to enhance quarantine performance was scenario 51 with (i) the opening

and closing of declaration system desks at queue thresholds of 10 and 40, respectively,

(ii) opening and closing counters of nothing to declare at thresholds of 20 and 5,

respectively. The maximum/average waiting time in the declaration system were 60.31

minutes and 13.17 minutes, respectively. The maximum/average waiting time for

nothing to declare was 27.86 minutes and 5.46 minutes, respectively.

Figure 7- 17: Quarantine facility results of adding and removing staffing policies

7.4.2 Sharing staff policy for integrated processes

This section discusses the results of the simulation referring to staff sharing

policies. Immigration is discussed first as it is located on both inbound and outbound

sides. Staff is shared between the two sides with priority for outbound immigration.

The criteria for selecting the best policy is based on Kirk (2013); the author found that

the average time spent in the immigration domain is between 6 and 7 minutes. Thus,

any policy within this range or lower is acceptable. From Figure 7-18, it is clear to see


that the average waiting time for outbound immigration is in the range of 5 to 7

minutes for the first 16 scenarios and the last nine scenarios. The policies that can

provide acceptable waiting times are summarised in Table 7-2.

Figure 7- 18: Outcomes of simulated immigration staff sharing rules

Results suggest that to decrease the number of passengers waiting in the queue,

the number of staff hours is increased. Waiting times for both sides exhibit similar

behaviour for the first nine scenarios including maximum and average waiting time.

From scenario 11 there was a sharp rise in the maximum waiting time until it reached

peak waiting time in scenario 41. The swapping policy of scenario 41 is that if the

number of passengers waiting at outbound immigration is greater than 100 and the

number of passengers waiting at the inbound domain is less than 50, staff are moved

from inbound to outbound processing. If the number of outbound passengers is 20 or

less, staff are moved from outbound to inbound processing.


Table 7 - 2: Summary of eligible sharing polices

The second integrated process is security screening since it is located in both

outbound and inbound process flows. The same policies applied to the immigration

domain have been used for this investigation. The results are compared with the actual

data collected by Kirk (2013) to understand the acceptable queue time at security.

Based on Kirk (2013), the average queue time at security screening was 3.75 minutes

while the maximum queue time was 17.09 minutes. Hence, the best rules will be those

where the resulting wait times are equal to or lower than these limits.

Figure 7-19 demonstrates the outcomes of the simulation considering the

performance metrics of the 65 sharing policies including the maximum/average

waiting time of inbound and outbound security screening and total staffing hours. The

maximum/average waiting time of outbound security in the first nine scenarios and in

the last 19 scenarios are acceptable policies. The average waiting time ranged between

3.80 to 4.86 minutes in the first nine scenarios and between 3.71 to 5.7 minutes, in the

last 19 scenarios. The maximum waiting time ranged between 17.18 to 22.89 minutes

for the first nine scenarios and between 16.47 to 30 minutes for the last 19 scenarios.

Scenario 5 is considered the best policy since it provides the minimum waiting time

for both outbound and inbound security with average waiting times of 3.92 minutes

and 6.70 minutes, respectively, and maximum waiting times of 17.18 minutes and

25.42 minutes, respectively.


Figure 7- 19: Security screening facility results of sharing staffing policies

7.5 CHAPTER SUMMARY

This chapter demonstrated numerical experiments involving the ARM technique

for outbound and inbound flow processes. The first experiment is the base case

scenario which allocated staff based on static allocation methods. It clearly showed

how the static allocation method can provide less queueing times for all terminal

processes only if the upper limit of employees is chosen. However, the static allocation

method has a significant impact on operating time which caused long staffing hours.

The second experiment represented a scenario of allocating airport employees

dynamically. The model provides better outcomes in representing flows of both

passenger types departing and arriving service processes. It also able to manage the

operations significantly better by allocating the staff if needed which balanced queuing

time and operating hours of staff. The results of this investigation show that the total

staffing hours is halved or sometimes 65% lower using the dynamic allocation

approach. Dynamic allocation also has less variation than the static method and the

variance values are close and occur around the mean. Experimental results

demonstrated that dynamic allocation method can be significantly influenced by queue

threshold values in regard to adding/removing staff and sharing staff between

integrated processes.

Chapter 8 will discuss the development of an analytical optimization framework

for capacity planning. The opportunity of integrating the simulation model within an

analytical optimization framework is also demonstrated to decrease the cost of time

spent in the queues of the airport terminal.

204 Chapter 8: An Analytical Optimization Framework

Chapter 8: An Analytical Optimization

Framework

In recent years, airports have faced many challenges such as the continuing

growth in passengers which imposes significant strains on the air travel global

infrastructure that is expected to keep up with the increasing passenger flows. This

includes the capacity of the airports and their ability to process the increasing numbers

of passengers with high efficiency and minimum delay. At the same time, the required

expansion of the airport capacity might be limited by the available resources (e.g.

limited available land), environmental impacts and lengthy approval processes

(Barnhart et al., 2012). In addition, extension of the major airport infrastructure is

typically time-consuming and costly, which raises the need for the development of

smart systems and methods to improve airport performance within the available

infrastructure limitations.

Additionally, the airport terminal is a complex system and stochastic in nature

since it involves multiple stakeholders each performing a different facility in terminal.

It also has many interactions between different actors (Wu & Mengersen, 2013). Due

to the complex structure of airport terminals, the development of an analytical

optimization framework for studying passengers flow in airports under uncertainty of

future demand is difficult. These difficulties and challenges have led to studies of

overall terminal capacity planning problems. Previous studies have generally focused

on one element of the terminal or have not accounted for expandability (Solak et al.,

2009).

Therefore, the purpose of this chapter is to propose a mathematical approach for

capacity planning. This model will determine the expansion capacities for different

processes of the airport terminal. The objective function of the proposed model is to

minimise the cost of used resources and the total waiting time. A number of technical

constraints exist.

Chapter 8: An Analytical Optimization Framework 205

8.1 PROBLEM DESCRIPTION AND FORMULATION

This section describes and defines the variables and parameters used to

formulate the mathematical problem. The purpose of this model is to determine where

additional resources should be placed in order to reduce the cost incurred in wait times.

Different resource types cost different amounts, so the objective is also to acquire

resources at minimum total cost.

8.1.1 Model notation

Indices

𝑝, 𝑟, 𝑘, 𝑡, f process, resource, passenger types, period, shift

Sets

𝑃, 𝑅, 𝐾, 𝑇, F processes, resources, passenger types, periods, Shifts

Parameter

𝑁𝑟,𝑝, 𝑁𝑟 Maximum number of resources of type 𝑟 in process 𝑝 and across all processes

𝐶𝑟,𝑝 Cost of providing a resource of type 𝑟 in process 𝑝

𝑉𝑘,𝑝 Cost incurred per unit of waiting time for passengers of type k in process 𝑝

𝐵 Total budget available for capacity expansion

𝑛𝑟,𝑝 Current number of resources of type r in process p

𝜏𝑘,𝑝 Expected time taken to serve passenger 𝑘 in process 𝑝

Decision variable:

𝑁𝑟,𝑝 Number of additional resources of type r in process 𝑝

𝑊𝑘,𝑝 Total waiting time incurred in process 𝑝 for passengers of type 𝑘

The model is as follows:

Minimize ∑ ∑ 𝐶𝑟,𝑝𝑁𝑟,𝑝 𝑟𝑝 + ∑ 𝑘 ∑ 𝑉𝑘,𝑝 𝑊𝑘,𝑝𝑝 [Resource cost + cost of waiting] (1)

Subject to:

𝑁𝑟,𝑝 ≤ 𝑁𝑟,𝑝 ∀𝑝 ∈ 𝑃; ∀𝑟 ∈ 𝑅 [Upper bound] (2)

∑ 𝑁𝑟,𝑝 ≤ 𝑁𝑟 ∀𝑟 ∈ 𝑅𝑝 [Upper bound] (3)

∑ 𝐶𝑟,𝑝𝑁𝑟,𝑝 ≤ 𝐵 ∀𝑝 ∈ 𝑃 𝑟 [Budget constraints] (4)

𝑁𝑟,𝑝 ≥ 0 ∀𝑝 ∈ 𝑃; ∀𝑟 ∈ 𝑅 [Positivity] (5)

(𝑊𝑘,𝑝) = 𝑆𝐼𝑀𝑈𝐿𝐴𝑇𝐸(𝑁𝑟,𝑝, 𝜏𝑘,𝑝) [Calculation of waiting time via simulation] (6)

The objective function (1) has two components: (i) the cost of

purchasing/acquiring additional resources of type r in process p, and (ii) the total

passenger waiting time converted to a dollar value. Constraints (2) and (3) ensure that

the additional resources of type r do not exceed the maximum number of resources.


Constraint (4) restricts spending to a particular budget. Constraint (5) restricts the

decision variable 𝑁𝑟,𝑝 to be positive.

8.2 SIMULATED ANNEALING

To solve this model a meta-heuristic approach is advocated as constraint 6

cannot be handled using mixed integer programming without the application of a

simulation model. Of the different meta-heuristics, Simulated Annealing was chosen.

The simulating annealing (SA) algorithm is a meta-heuristic, effective and simple

optimisation algorithm used for the solving of probabilistic and non-linear

optimisation problems. As for any type of meta-heuristics algorithm, it is developed

by simulating and modelling one of nature’s phenomena (Burdett, 2015; Mohammadi

& Safa, 2016). The SA is coded in C++ and the simulation model is used for evaluation

purposes. Figure 8-1 demonstrates the details of the SA algorithm.

The role of simulated annealing is an iterative approach that is able to escape

local optima. This is done by starting with an initial solution and, during the iteration

loop, it moves to a neighbour solution. If the neighbour solution is better than the

current solution, the algorithm moves to it; otherwise the solution will be accepted as

the current solution with a probability 𝑃, which is presented as follows:

𝑃(𝑓) = 𝑒−1 ∗( ∆𝑓

𝑇 ) (7)

Where ∆𝑓 is the difference between the current solution and the neighbouring

solution of the objective function, and T is the temperature. At every temperature, a

selected number of perturbations are evaluated. SA requires several parameters (i.e.

primary temperature, the cooling rate, the number of function evaluations at every

temperature and the final temperature) to implement simulated annealing (Amaran,

Sahinidis, Sharda, & Bury, 2016).

At early stages the temperature is high and many non-improving moves are

accepted. SA time goes by, solutions are only accepted if a strict improvement occurs.

With the slow reduction in temperature, the worst solutions have less probability to be

accepted. According to Amaran et al. (2016), who stated that because of the

exponential form, the acceptance of neighbourhood points is more likely at high

temperature, there is lower probability as temperature is decreased.


Figure 8- 1: General steps of the simulated annealing

Therefore, several series of preliminary experiments were conducted to

determine an appropriate starting temperature. It is evident from the graph below, that

the best parameter values of this problem are as follows: temperature (T) = 15000,

cooling rate (α) = 0.015, and total number of iterations = 600.

Set initial parameters

Generate initial sequence

Generate Neighborhood

Stop

Adjust temperature

Stopping criteria

StoppingCriteria

Is it accepted?

No

Set the solution as the best

Assess New Solution

No

No

Yes

Yes

Yes


T =150 T =1500 T =15000

α =

0.1

5

α=

0.0

15

α=

0.0

015

Figure 8- 2: Selecting the best initial parameters


8.2.1 Simulated annealing algorithm description

Phase 1: Create initial solution

SA may be initialised with a randomly created soliton or via some

heuristic/constructive algorithm. However, because of the resource limitation

constraints, some of the generated solutions will not be feasible. These solutions

should be corrected via a corrective algorithm. Algorthim 1 is used to initialise a set

of solutions.

Algorithm 1: Create initial population

1 For (each Shift);

2 For (each process 𝑝1);

3 do

4 𝑁𝑟1,1,𝑝1

= random number between (1, Maximum number of resources of type 𝑟1,1

available in process 𝑝1 );

5 𝑁𝑟1,2,𝑝1



6 X = 𝑁𝑟1,1,𝑝1+ 𝑁𝑟1,2,𝑝1

; economy and business counters;

while x ≤ number of available 𝑟1,1 𝑎𝑛𝑑 𝑟1,2 ;

7 End

8 𝑁𝑟2,𝑝2 = Uniform (1, 𝑁𝑟2,𝑝2);

9 𝑁𝑟𝑛,𝑝𝑛 = Uniform (1, 𝑁𝑟𝑛,𝑝𝑛);

10 End

Phase 2: Create new solutions

In general, creating a new solution depends on local search improvement

algorithms and the control strategy or general optimisation. In this case, creating a new

solution can be done by randomly changing the number of resources for one process

selected randomly from the current solution. The creating of the new solution

algorithm is illustrated in algorithm 2.

Algorithm 2: Creating new solution 1 If (change in process 𝑝1);

3 while x ≤ number of available 𝑟1,1 𝑎𝑛𝑑 𝑟1,2 do

4 𝑁𝑟1,1,𝑝1



5 𝑁𝑟1,2,𝑝1



6 X = 𝑁𝑟1,1,𝑝1+ 𝑁𝑟1,2,𝑝1

;

7 Else If (change in 𝑝2)

8 𝑁𝑟2,𝑝2 = Uniform (1, 𝑁𝑟2,𝑝2);

9 Else

10 𝑁𝑟𝑛,𝑝𝑛 = Uniform (1, 𝑁𝑟𝑛,𝑝𝑛);;


Phase 3: Assess New Solution

In this step, the goodness of the new solution is evaluated. Algorithm 3

demonstrates the assessment procedures. The generated solution will be simulated to

measure a performance matrix, such as the average waiting time at each processing

point. Also, the best cost will be selected by comparing it with the current cost. The

acceptance probability for calculating the function of simulated annealing is:

SA_ Probability= 𝑒−1 ∗(

𝐶𝑜𝑠𝑡𝑛𝑒𝑤−𝐶𝑜𝑠𝑡𝑐𝑢𝑟𝑟𝑒𝑛𝑡𝑇𝑒𝑚𝑝𝑟𝑒𝑡𝑢𝑟𝑒

)

Algorithm 3: Assess New Solution 1 Function Local Search();

2 New cost = simulation ();

3 If (New cost < Best cost);

4 Update Best solution;

5 Best cost = New cost;

6 Else

7 If (New cost < Current cost);

8 Update Current solution;

9 Current cost = New cost;

10 Else

11 P=Random number between (0 , 1)

12 SA_ Probability= 𝑒−1 ∗(

𝐶𝑜𝑠𝑡𝑛𝑒𝑤−𝐶𝑜𝑠𝑡𝑐𝑢𝑟𝑟𝑒𝑛𝑡𝑇𝑒𝑚𝑝𝑟𝑒𝑡𝑢𝑟𝑒

)

13 If(P> SA_ Probability)

14 Reject the new solution;

15 Else

16 Accept the new solution

Phase 4: Stop criteria

Finally, the condition of stopping the SA algorithm is based on the given

maximum number of iterations. Algorithm 4 is an illustration of the main loop of the

stopping criteria algorithm.

Algorithm 4: main loop

1 Parameter initialisation

2 Function Create initial population()

3 Simulation ();

4 do

5 Function refinement ();

6 Simulation ();

7 Function Assess New Solution();

8 x++;

9 while (x< maximum number of iterations)


8.3 NUMERICAL TESTING AND ANALYSIS

In this section, the integrated meta-heuristic simulation, SA algorithm and the

simulation model were tested. The solution chromosome should simultaneously reflect

two main characteristics:

Number of resources for each process, such as check-in resources (i.e.

economy, business counters), security screening resources and immigration

resources (i.e. common counters, SmartGates)

The number of assignment resources for each shift.

In this numerical investigation, there are three types of process and five types of

resources that were considered. Process type one, the check-in process, has five

separate lines, each with eight counters, two for business and six for economy. Process

type two is security screening with five lanes. Process type three is immigration, which

has eight common counters and 10 SmartGates. It is assumed that there are three

periods during the day to which these processes are assigned to be operated. Figure 8-

3 is a snapshot of the simulation outputs. This work showed that how simulation model

can be used within analytical optimisation framework to expand the capacity of the

airport terminal.

Figure 8- 3: snapshot of simulated annealing results

The cost of waiting time is considered based on the given policy of acceptable

queue time in a particular process. This is named the cost of inconvenience, as it


exceeds given acceptable average waiting time. For example, passengers at the check-

in process can be classified as business and economy, each with different queue time

limits. Kazda and Caves (2015) argued that the average waiting time should not be

higher than 12 minutes for economy class and 3 minutes for business. The summary

of input data used in this study is listed in Table 8-1.

Table 8- 1: Summary of the input data

Domain of the airport Values

Check-in:

Cost for opening new check-in counter is

The acceptable average waiting time for Economy passengers is

Cost for inconvenience at check-in for economy is

The acceptable average waiting time for Business passengers is

Cost for inconvenience at check-in for Business is

20$

12 minutes

15$

3 minutes

25$

Security screening:

Cost for opening new security screening desk is

The acceptable average time that normal passengers should wait is

Cost for inconvenience at Security screening for normal passengers is

The acceptable average time that Diplomatic passengers should wait is

Cost for inconvenience at Security screening for Diplomatic passengers is

15$

5 minutes

15$

2 minutes

20$

Immigration:

Cost for opening new Immigration desk is

Cost for opening new SmartGate is

The acceptable average waiting time at common counter is

Cost for inconvenience at common counters is

The acceptable average waiting time at SmartGate is

Cost for inconvenience at SmartGate is

15$

10$

7 minutes

20 $

0.5 minute

10$

Two different methods of creating new solutions were used to generate a starting

solution. The first is creating a new solution randomly and the second is local search.

The random search method initialises SA with a randomly created solution, while the

local search initialises SA via constrictive algorithm by changing one solution

chromosome and then refined by SA. For each method, 10 runs were repeated with the

same parameters. The results of the runs are presented in Table 8-2 and Table 8-3. The

general parameters used for both methods are: temperature (T) = 15000, cooling rate

(α) = 0.015, and the maximum number runs is 1500.


The first column refers to the number of better solutions, where the average

number of better solutions obtained by random search approach is 7.9 and the local

search approach is able to find 7.8 better solutions on average. Column two presents

the total average waiting time spent in the airport terminal process. It is evident that

local search provides lower waiting times compared with the random approach. It

could reduce the total average waiting time by 44.05%. It also has shorter run time,

with an average of 15.27 minutes compared to 21.05 minutes for the random approach.

However, from the results presented in both tables, it can be clearly seen that the

random search method reduces the objective function value by 12.11%. The mean 𝜇

value of the objective function obtained from the random approach is $1998.3 whereas,

𝜇 of objective function value of the local search is $2256.

Table 8- 2: summary of simulated annealing results using random search technique

Random search

Run # # of better solutions

Total Ave waiting time

Objective function value

Run time (min)

Run1 6 118.73 2031 21.05

Run2 5 58.44 2001 21.16

Run3 10 126.39 2060 23.48

Run4 9 39.57 2038 23.43

Run5 6 64.74 1973 20.24

Run6 9 148.18 1874 20.55

Run7 8 112.714 1913 17.36

Run8 10 86.67 1997 21.55

Run9 7 127.8 2052 17.36

Run10 9 131.25 2044 24.36

𝝁 7.9 101.4484 1998.3 21.054

𝝈 1.7 34.725498 58.99160957 2.256671


Table 8- 3: summary of simulated annealing results using local search technique

Local search

Run # # of better solutions

Total waiting time

Objective function value

Run time (min)

Run1 7 38.88 2336 13.7

Run2 10 92.31 2195 14.02

Run3 9 80.49 2182 15.4

Run4 10 94.97 2042 15.48

Run5 8 22.25 2180 17.22

Run6 8 28.89 2250 22.22

Run7 8 72.6 1994 12.53

Run8 8 85.47 2531 17.27

Run9 4 86.45 2638 12.21

Run10 6 45.6 2215 13.27

𝝁 7.8 64.82 2256.3 15.27

𝝈 1.72 26.45 186.49 2.88

From the 10 replications of both random search and local search, solution

numbers 4 and 6 from the random search and local search were selected as the best

solutions, for two reasons. The first reason is that the value of the objective function

is closer to the mean value of all the objective values. The second reason is that the

chosen simulation runs provide the minimum total average waiting time in the airport.

Figure 8-4 demonstrates the best solution given by the random search for the

solution run number 4. The optimal solution for this simulation run in regards to

opening additional resources for check-in is summarised in Table 8-4. The average

waiting time is 1.91 minutes for business class and 34.5 minutes for economy class

passengers.

Table 8- 4: Check-in additional resource results using the random technique

Line 1 Line 2 Line 3 Line 4 Line 5

Bu

sin

ess

Eco

no

my

Bu

sin

ess

Eco

no

my

Bu

sin

ess

Eco

no

my

Bu

sin

ess

Eco

no

my

Bu

sin

ess

Eco

no

my

Shift 1 1 4 2 3 2 4 1 4 1 4

Shift 2 2 2 1 1 1 1 1 2 2 4

Shift 3 2 4 1 6 1 6 1 2 2 4


For the security screening checkpoints, the opening resources are 5, 2 and 3

control checkpoints for shift 1, shift 2 and shift 3, respectively, having an average

waiting time of 3.04 minutes. The opening common immigration process is 4, 3 and 5

counters, while for SmartGates there are 8, 9 and 7 kiosks for the three shifts, having

average waiting times of 0.11 and 0.02 minutes for the common immigration desks

and SmartGate kiosk, respectively. The total cost of opening all resources is $2038

given the total average time spent in the queues is 39.57 minutes.

Iteration

0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500

obje

ctiv

e fu

nctio

n va

lues

0

10000

20000

30000

40000

50000

Bes

t so

lutio

n

0

2000

4000

6000

8000

Objective

Best solution

Results of simulation run 4 of random approach

Iteration

0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500

Tota

l Ave

wai

ting

time(

min

)

0

500

1000

1500

2000

Bes

t sol

utio

n

0

100

200

300

400

500

waiting_time for all runs

Best solution

Figure 8- 4: SA optimisation results using the random method of creating new solutions

Figure 8-5 illustrates the optimal solution provided by the SA algorithm using

local search for creating a new solution. In this simulation run, the solution is

characterised from the total cost of $2250 given the total average waiting time spent

in the system is 28.89 minutes. The optimal solution for this simulation run is opening

a check-in resource based on the detailed information listed in Table 8-5. By adding


this resource, the average waiting time at the check-in process will be 1.56 minutes for

business class passengers and 7.03 minutes for economy, a reduction of 20.17% for

business and 132.29% for economy compared with the random search method.

Table 8- 5: Check-in additional resource results using local technique

Line 1 Line 2 Line 3 Line 4 Line 5

Bu

sin

ess

Eco

no

my

Bu

sin

ess

Eco

no

my

Bu

sin

ess

Eco

no

my

Bu

sin

ess

Eco

no

my

Bu

sin

ess

Eco

no

my

Shift 1 1 4 2 3 3 5 2 5 1 5

Shift 2 2 7 1 2 1 3 4 3 1 4

Shift 3 2 1 2 2 2 2 4 1 2 1

For the process of security screening checkpoints, the best solution can be found

if opening 4, 4, and 5 security control checkpoints for shift 1, shift 2 and shift 3

respectively. By opening these number of resources at the security screening process,

the average waiting time is 7.92 minutes. Finally, the immigration process should open

5, 4 and 3 common immigration counters and 10, 10 and 9 SmartGates for the three

shifts, in order to get the optimal solution, having the average waiting of 0.10 minutes

and 0.012 minutes for the common counters and SmartGates, respectively.


Iteration

0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500

obje

ctive

func

tion

value

s

0

2000

4000

6000

8000

10000

12000

Best

sol

utio

n

0

500

1000

1500

2000

2500

3000

Objective

Best solution

Results of simulation run 6 of local search approach

Iteration

0 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500

Tota

l Ave

wai

ting

time(

min

)

0

500

1000

1500

2000

Best

sol

utio

n

0

20

40

60

80

100

120

waiting_time for all runs

Best solution

Figure 8- 5: AS optimisation results using the method of creating new solution using local technique

8.4 CHAPTER SUMMARY

This chapter discussed the development of a mathematical approach to capacity

planning. The objective of the model is determining where additional resources should

be located to decrease the cost of time spent in the queues of the airport terminal. Since

the proposed problem is probabilistic and non-linear, a meta-heuristic approach is

advocated because constraint 6 cannot be handled using mixed integer programming

without the application of simulation.

Two different approaches for creating new solutions were used in this study. The

first one is creating a new solution using the random technique and the one is creating

a new solution by using the local search technique. The random technique decreased

the objective function value by 12.11% however, the local search technique can reduce

the total average waiting time by 44.05%. It also has shorter run times, with averages

of 15.27 minutes compared to 21.05 minutes for the random approach.

218 Chapter 9: Conclusion

Chapter 9: Conclusion

9.1 INTRODUCTION

This chapter outlines the conclusions of this thesis. It first discusses the primary

outcomes and how they contribute to the body of knowledge of the field of passenger

flows modelling and airport operational planning. It then provides a summary of the

key research activities and answers the research questions presented in Chapter 1. It

finishes by making recommendations for airport operations management and the

directions of future research.

9.2 SUMMARY AND DISCUSSION

This research has resulted in the development of a holistic model based on the

combination of simulations, ARM algorithms and analytical optimisation approaches.

This model is a decision support tool for airport operators to make well-versed

decisions for efficient airport operation. Application of the model facilitates the

operational planning of integrated inbound and outbound flow processes and

determines the impacts on passenger flow and congestion.

Initially, relevant literature was reviewed to select suitable approaches. Then, the

major processes of outbound and inbound systems were mapped out based on the

business processes models developed by Mazhar (2015) for Australian international

airports. The second phase involved development of the simulation models for

outbound and inbound passenger flow processes using ExtendSim software. The

data/input requirements of the simulation model were categorised into three basic

categories:

Flight schedule such as departure time, boarding time, the size of the

aircraft, and the name of the airline.

Passenger characteristics, e.g. the percentage of passengers travelling by

business class, the percentage of passengers using SmartGates, etc.

Operational characteristics such as distribution of processing time, and

characteristics of the counters for processing facilities.

Chapter 9: Conclusion 219

The final phase was the development of an optimisation approach for capacity

planning. By applying the developed optimisation algorithms, optimum resource

allocation was obtained. The purpose of the capacity planning model is to determine

where additional resources should be placed to reduce the cost of waiting as well the

obtained resources. The key research activities are summarised below.

Chapter 2 reviewed the existing research on recent issues related to passenger

flow, security and staff allocation in a complex environment such as airport terminals.

It then reviewed models used to resolve such problems. These models can be classified

as either ‘analytical’, ‘simulation’, or ‘hybrid’ models, providing decision support

capabilities at all levels of detail: from macroscopic, through mesoscopic, to

microscopic. The literature review highlighted the improvements made in our

understanding of passenger flow modelling to date. Despite the extensive research on

airport issues, the literature suggests that an aggregate model, providing integrated

views associated with the performance of processing points and facilitating both

outbound and inbound processes, is still needed. From the state-of-the-art and state of

practice reviewed, suitable methods for addressing the primary aim of the research

were selected. The simulation method represented by DES was the best technique for

this research when associated with modelling of the system operations and could

capture system behaviour at the macroscopic level (Furian, Neubacher, Vössner,

O’Sullivan, & Walker, 2014; Siebers, Macal, Garnett, Buxton, & Pidd, 2010). The

DES was selected as it can provide a true presentation of the system and can deal with

the analysis of passenger flow problems via a top-down modelling approach.

Moreover, the DES has the capability to model uncertainty and non-linearity

(Sachidananda et al., 2016).

In Chapter 3, the conceptual framework of a holistic simulation model for an

international terminal was introduced. The proposed framework consists of three

major modules including a simulation model for outbound processes, a simulation

model for inbound processes and algorithms for overall management of airport

resources. Due the complexity of the holistic model, Chapter 3 developed the first

phase as a simulation model for departure passenger flows. This phase focused on

modelling outbound standard terminal operations including check-in counters, security

screening, immigration and custom and boarding. The model was built around


hierarchy modules to provide insights into the interactions between processes and sub-

processes.

The developed framework for outbound passenger flow was illustrated by a

series of experiments in Chapter 4. Different arrival patterns of passengers were

implemented in outbound process simulations. The simulation outcomes provided a

better understanding of the behaviour of passenger airport access which could lead to

reduced waiting time and possible congestion by increasing the number of working

stations (i.e. number of check-in counters) at peak times.

Chapter 5 presents the integrated simulation model by combining a complete

inbound simulation model and advanced resource management algorithms to describe

the interaction between passenger behaviour and outbound/inbound processes. The

inbound simulation model was concerned with passenger disembarking including

generating inbound passenger attributes (i.e. SmartGate users, walking speed, number

of bags, etc.), inbound security screening, immigration, baggage claim and quarantine.

We developed an algorithm to generate the attributes of passengers based on data taken

from flight scheduling, i.e. arrival time of flight, gate number and capacity of flight.

The significant contribution of this chapter is the development of advanced resource

management (ARM) algorithms that can dynamically allocate/reallocate or add and

remove airport personnel based on given queue threshold policies. In ARM, the

processes were categorised into two categories:

Non-integrated processes including check-in and quarantine

Integrated processes such as security screening and immigration processes.

Chapter 6 demonstrates the capability of the developed simulation model to

accurately reflect an actual airport terminal. The validation processes were confirmed

by comparing simulation results with the real data provided by King Khalid

International Airport (KKIA). Four types of simulation outcomes, including the

average and maximum waiting time in the queue and the average and maximum cycle

time at each departure facility, were considered. Results showed that the developed

model is more applicable for Australian airports such as Brisbane International Airport

than an external airport, such as KKIA. Since each airport proposes different

operational processes, such as security (Akgun, Kandakoglu, and Ozok (2010), the

model results can be further improved and have more accurate validation as follows:


By collecting data using the video camera technology. This technology can

provide samples and large volumes of samples more efficiently and

accurately, because they often record a full day, week or even a month of

data

By changing the input requirements and running ‘what-if’ scenarios; since

airports are exposed to external effects and the developments of air traffic

Chapter 7 provided an integrated view of all the airport terminal processes using

the devised ARM strategies for allocating and reallocating terminal staff. This chapter

compared the static allocation and developed dynamic allocation methods in terms of

the total average waiting time and total staffing hours. Based on the given assumption

of allocating the staff if needed, the developed ARM model decreases the total staffing

hours is some terminal facilities, such as check-in, by up to half compared with Static

Allocation Base Case method results. It also balanced the averaging waiting time and

operation hours since the staff is only allocated if needed. The ARM can be a decision

support tool and efficiently used to support and model real-world airport staff

allocation planning problems.

Chapter 8 discussed how the physical resources of airport terminals can be

optimised. An analytical model was developed and integrated with the simulation

model to determine where the additional resources should be placed to minimise

waiting time costs. This model also aimed to obtain resources at minimum total cost.

9.3 RESEARCH CONTRIBUTIONS

The motivation for this work was to develop a model capable of studying

passenger flows and staffing requirements at airport international terminals as a single

unit by facilitating the integration of outbound and inbound systems. Previous research

was limited by the focus on individual processes or fragmented areas of decision

making procedures of airports. This limited focus produced a knowledge gap in

relation to the terminal system as a whole and how the processes in different sectors

influenced the operation and management of the entire system.

The primary contribution of this thesis is the development of a holistic model

that integrates the major inbound and outbound flow types to analyse passenger flow

issues in an international airport. By facilitating and integrating inbound and outbound

processes, an integrated view of overall airport operations can be achieved (Yamada


et al., 2017; Zografos et al., 2013). The following sub-sections briefly outline the

significant contributions of this thesis to the body of knowledge on passenger flow

modelling within airports.

9.3.1 Framework for airport outbound passenger flow modelling

Based on objective one, a framework for outbound passenger flow modelling

was developed using DES and built using ExtendSim V9.2 simulator software (see

section 3.5). Excel macro visual basic programming was used to model and generate

model inputs including information related to flight schedules, passenger

characteristics and boarding characteristics. The most important feature of this model

is its ability to predict the effect of different flight schedules which can be used as a

feedback mechanism for improvements before implementation. Results show that

flight schedules have a large impact on passenger flows. Integrated flight schedule

creation and passenger simulation analysis may help address some of the issues of

passenger flow within airport terminals, especially for the two most-

affected processes: security screening and immigration.

9.3.2 Investigating the effect of arrival patterns of departing passengers on the

departure terminal operations

As discussed in objective two, a model for passenger arrival procedures was

developed using MATLAB to estimate the volume of travellers arriving at the airport

over time with respect to different types of distribution functions. The model can

obtain the total number of passengers per time interval for all flights. This model was

integrated with the outbound simulation model to study the impacts of different arrival

patterns on departure flow processes.

Simulation results suggested that different arrival patterns can significantly

affect the performance of operational processes in the airport terminal taking into

account the maximum/average number of passengers waiting in the queues and the

maximum/average waiting time in front of working stations. The policy of the time to

open check-in counters under a given mean value has a significant impact on the

departing passenger arrival profile, especially when the policy of time to show up at

the airport equalling the mean value µ is applied. As a result, potential congestion and

longer waiting times might occur in airport processing activities including check-in,

security and immigration. This can lead to significant congestions and delays which

could lead to passengers missing flights and poor passenger experience. The passenger


arrival distribution patterns are a major factor in planning airport-terminal facilities,

such as the number of check-in counters and service agents, along with the operation

times of passenger check-in and queue length (Fayez, Kaylani, Cope, Rychlik, &

Mollaghasemi, 2008; Park & Ahn, 2003). It can also affect the performance of entire

outbound terminal operations and other related services. For example, based on the

results of Chapter 4, the best time to arrive at the international airport is 4 hours before

the flight (given the mean value before flights is 60 minutes) which could lead to

shorter queues and waiting times at main outbound processes and more time for

discretionary activities and retail which is desirable for airport retail operators.

9.3.3 Advanced resource management strategies

As mentioned in objective three, the developed ARM algorithms were integrated

with the simulation model (see section 5.4) and used to manage operations by

allocating staff if needed, leading to a balance between acceptable waiting times in

queues and staff operating hours. Our ARM approach is significant because it can be

used in other domains to manage resource allocation with slight modifications.

Compared with the SABC results, dynamic allocation could halve the total staffing

hours in some processes, such as check-in up, since the staff will be allocated if needed.

Additionally, the dynamic allocation method can be influenced by queue threshold

values in regard to adding/removing staff and sharing staff between integrated

processes. This approach can also be used to reduce the number of delayed flights and

the total operating hours as explained in section 7.3.2. The developed approach is

significant because it can be used to manage entire airport systems and provide better

results than existing approaches. For example, the results obtained from our model

demonstrated better improvement than Kierzkowski and Kisiel (2016), especially in

regard to the total average waiting time and staffing hours of security screening as

summarised in Table 9-1.


Table 9- 1: Comparisons of developed ARM results with Kierzkowski and Kisiel (2016)

Compared

study

Total average waiting time (min) Total staffing hours (hours)

ARM model (Artur Kierzkowski &

Kisiel, 2016)

ARM model (Artur Kierzkowski

& Kisiel, 2016)

Static case 9.25 9.538 96 285

Dynamic case 7.14 7 65.23 162

9.3.4 Development of a novel holistic model for facilitating outbound and

inbound processes

As discussed on objective four, further expansion of the simulation model was

done using ExtendSim to facilitate the integration between inbound and outbound

processes. The model can be run with either elements or both flow processes

simultaneously. Each element has its own input data of departing and arriving flights

and outgoing and incoming passenger attributes (see section 3.4 and section 5.2.2).

Further interaction between inbound and outbound processes can be investigated by

using the ARM approach within the developed simulation to improve passenger flows.

The proposed model also provides flexibility in changing the order of the

processes as each airport is operated differently. For example, in Australia, the security

screening process of Brisbane International Airport is located before immigration,

while in Perth, security comes after immigration (Mazhar, 2015; Shuchi, 2016).

Furthermore, the model can be changed and modified at the operational level to

characterise passenger flow with respect to a set of parameters involving flight

schedules, processing time, and passenger characteristics (i.e. business and economy,

declaration lane the passenger can use). Statistics for aggregate performance matrices

of outbound and inbound facilities can then be collected.

9.3.5 Strategic and operational planning techniques

As stated in objective five, the last thesis contribution takes the form of strategic

planning. Most existing research has considered the strategic level (Solak et al., 2009;

Sun & Schonfeld, 2015), while other research has focused on the operational planning

level only (Fayez et al., 2008; Manataki & Zografos, 2009b; Schultz & Fricke, 2011).

The Capacity Planning Model (CPM) can provide strategic planning while the

proposed simulation model which can be used for the operational planning level. This


thesis has contributed to the body of knowledge by enabling two levels of planning,

operational and strategic.

9.3.6 Practice contribution

The developed holistic model can immediately be of practical use in two

different ways. The first significant practical contribution is to evaluate the efficiency

and performance of airport operations. This would enable airport operations managers

to identify the potential bottlenecks. The model also supports what-if and trade-off

scenario planning for evaluating changes in operational policies. The model can be

used to deal with sudden problems or any unexpected congestion situations.

Additionally, the proposed simulation model can be used to investigate several related

factors that might affect the performance of the terminal operations. Examples of these

factors are new security regulations and new requirements, resource availability,

arrival and departure patterns, new technology, randomness and variability

characteristics throughout the system.

Moreover, the model outputs of the current research can be treated as a decision

support tool. The proposed model has the potential to improve the overall performance

and efficiency of terminal operations, but only if it is integrated into appropriate ARM

strategies. It can quantitatively forecast and compare the effect of new procedures and

counterpart regulations and decisions on the operational performance of terminal. For

example, decisions associated with passenger arrival time to the airport based on the

departure time of the scheduled flight and queue length threshold to open and close

checking counters or allocate and reallocate personnel.

9.4 LIMITATIONS AND FUTURE RESEARCH DIRECTIONS

Through the course of this research, there were two major limitations observed.

First, there was a lack of access to the detailed data related to operational facilities.

This was due to the recent strict regulations associated with security issues set by

government. This limitation resulted in difficulties in developing, validating and

calibrating the modelled passengers’ flows within an international terminal. The

accuracy of the interaction between passengers and terminal operations and the

decision-making procedures affect the model’s reliability. Information, such as

processing time distribution, service rate and the acceptable waiting time in queues for

different terminal processes, also significantly affect system performance and


passenger satisfaction. Due to the difficulties of accessing this data, the proposed

model has been simplified either by utilising available data collected by previous work

or by making experimental assumptions where needed.

The second limitation is that the proposed model in this thesis is developed for

a specific Australian international airport. This is an issue as most airports facilitate

their passengers in different ways (Fernandes & Pacheco, 2002). For example,

international airports commonly open check-in counters 3 hours prior to departure

while domestic airports open check-in counters just 90 minutes beforehand (Cheng,

2014; Schultz & Fricke, 2011). Therefore, passenger arrival distribution patterns can

vary between local airports and international airports. The simulation model of

passenger flows considers all aspects of international airports. Hence, it can be adapted

widely. Since domestic airports are considered a subset of international airports, the

proposed model can be utilised to simulate all scenarios that may occur in domestic

airports after adjustment of the processes in the model based on empirical data.

To address these limitations, the following future research directions are

recommended:

Determine the influence of arrival patterns on resource-allocation management

including both outbound and inbound systems of an international terminal. This

can be done by considering different distributions functions to determine the

primary inputs of departing passenger arrival profiles. This is likely to provide

significant new outcomes about the expected impacts of such inputs on the

effective allocation of resources. Ultimately, the efficiency of all airport terminal

operations can be improved.

Determine how to employ the developed models to larger airports with two or

more terminals. For example, Dubai International Airport has three separate

terminals—Terminal 3 alone has four concourses each with 26 gates including

five A380 gates and a total capacity of 60 million passengers. Similarly, further

research can be conducted to study the problem of passenger flow within airports

dealing with international and domestic passengers in the same terminal, such as

the Gold Coast Airport.

As previously mentioned, the holistic framework proposed in this thesis was

developed based on the available data (Airport of the Future Project undertaken


by QUT) and available literature. The key weakness of this evaluation technique

is the lack of real-world data or scenarios. Therefore, further research is needed to

validate the developed model by comparing its results with the real-life scenarios

to increase model accuracy and robustness as well as meet practical requirements.

Bibliography 229

Bibliography

Abdoul Soukour, A., Devendeville, L., Lucet, C., & Moukrim, A. (2013). A Memetic

Algorithm for staff scheduling problem in airport security service. Expert

Systems with Applications, 40(18), 7504-7512. doi:

http://dx.doi.org/10.1016/j.eswa.2013.06.073

Adacher, L., Flamini, M., Guaita, M., & Romano, E. (2017). A model to optimize the

airport terminal departure operations. Transportation Research Procedia, 27,

53-60. doi: https://doi.org/10.1016/j.trpro.2017.12.151

Akgun, I., Kandakoglu, A., & Ozok, A. F. (2010). Fuzzy integrated vulnerability

assessment model for critical facilities in combating the terrorism. Expert

Systems with Applications, 37(5), 3561-3573. doi:

https://doi.org/10.1016/j.eswa.2009.10.035

Alhussein, S. N. (2011). Analysis of ground access modes choice King Khaled

International Airport, Riyadh, Saudi Arabia. Journal of Transport Geography,

19(6), 1361-1367.

Almuharib, T. M. (2014). Service Quality Improvement Through Lean Management

at King Khalid International Airport in Saudi Arabia. (PhD), Plymouth

University. Retrieved from https://pearl.plymouth.ac.uk/bitstream/handle/

10026.1/3137/2014Almuharib10373820PhD.pdf?sequence=1&isAllowed=y

Alodhaibi, S., Burdett, R. L., & Yarlagadda, P. K. D. V. (2017). Framework for Airport

Outbound Passenger Flow Modelling. Procedia Engineering, 174, 1100-1109.

doi: https://doi.org/10.1016/j.proeng.2017.01.263

Amaran, S., Sahinidis, N. V., Sharda, B., & Bury, S. J. (2016). Simulation

optimization: a review of algorithms and applications. Annals of Operations

Research, 240(1), 351-380. doi:10.1007/s10479-015-2019-x

Andreatta, G., De Giovanni, L., & Monaci, M. (2014). A Fast Heuristic for Airport

Ground-Service Equipment–and–Staff Allocation. Procedia - Social and

Behavioral Sciences, 108, 26-36. doi:

https://doi.org/10.1016/j.sbspro.2013.12.817

Artur, K., & Tomasz, K. (2017). A model of check-in system management to reduce

the security checkpoint variability. Simulation Modelling Practice and Theory,

74, 80-98. doi: http://dx.doi.org/10.1016/j.simpat.2017.03.002

Ashford, N., Mumayiz, S. A., & Wright, P. H. (2011). Airport engineering: planning,

design, and development of 21st century airports (4th;4; ed.). Hoboken, N.J:

Wiley.

Australian Government. (n.p. ). Going through the border. Retrieved from

https://www.homeaffairs.gov.au/Trav/Ente/Goin/Departing/departureSmartG

ate

Backlund, A. (2000). The definition of system. Kybernetes, 29(4), 444-451. doi:

doi:10.1108/03684920010322055

Banks, J. (2010). Discrete-event system simulation (5nd ed.). Englewood Cliffs, N.J:

Prentice Hall.

Barbo, W. A. (1967). The use of queueing models in design of baggage claim areas at

airports. Retrieved from http://trid.trb.org/view.aspx?id=24641

Barnhart, C., Fearing, D., Odoni, A., & Vaze, V. (2012). Demand and capacity

management in air transportation. EURO Journal on Transportation and

Logistics, 1(1-2), 135-155.

https://doi.org/10.1016/j.proeng.2017.01.263

https://www.homeaffairs.gov.au/Trav/Ente/Goin/Departing/departureSmartGate

https://www.homeaffairs.gov.au/Trav/Ente/Goin/Departing/departureSmartGate

http://trid.trb.org/view.aspx?id=24641

230 Bibliography

Barnhart, C., Fearing, D., & Vaze, V. (2014). Modeling passenger travel and delays in

the National Air Transportation System. Operations Research, 62(3), 580-601.

Bazargan, M. (2007). A linear programming approach for aircraft boarding strategy.

European Journal of Operational Research, 183(1), 394-411.

Beck, A. (2011). Case study: modelling passenger flows in Heathrow Terminal 5.

Journal of Simulation, 5(2), 69-76. doi:10.1057/jos.2011.4

Bertsimas, D., Lulli, G., & Odoni, A. (2011). An integer optimization approach to

large-scale air traffic flow management. Operations Research, 59(1), 211-227.

Bevilacqua, M., & Ciarapica, F. E. (2010). Analysis of check-in procedure using

simulation: a case study. Paper presented at the Industrial Engineering and

Engineering Management (IEEM), 2010 IEEE International Conference on.

Blaž Rodič, A. B. (2010). Dynamic Airport Ground Crew Scheduling. Paper presented

at the Proceedings of the 21st Central European Conference on Information

and Intelligent Systems.

Budesca, G. C., & Juan, A. A. (2014). Optimization of aircraft boarding processes

considering passengers' grouping characteristics. Paper presented at the

Proceedings of the 2014 Winter Simulation Conference.

Burdett, R. L. (2015). Multi-objective models and techniques for analysing the

absolute capacity of railway networks. European Journal of Operational

Research, 245(2), 489-505. doi: https://doi.org/10.1016/j.ejor.2015.03.020

Bureau of Infrastructure and Transport and Regional Economics (BITRE). (2012). Air

passenger movements through capital and non-capital city airports to 030–31.

Retrieved from https://www.bitre.gov.au/about/index.aspx

Cheng. L. (2014). Modelling airport passenger group dynamics using an agent-based

method. (Dissertation/Thesis), Queensland University of Technology.

Retrieved from https://eprints.qut.edu.au/72999/1/Lin_Cheng_Thesis.pdf

Cheng, L., Reddy, V., Fookes, C., & Yarlagadda, P. K. D. V. (2014 ). Agent-based

modelling simulation case study: Assessment of airport check-in and

evacuation process by considering group travel behaviour of air passengers,

Zurich.

Cheng, L., Yarlagadda, R., Fookes, C., & Yarlagadda, P. K. (2014). A review of

pedestrian group dynamics and methodologies in modelling pedestrian group

behaviours. World, 1(1), 002-013.

Cheng, S., Mu, Q., Zhang, H., & Zhang, Y. (2014). A Fuzzy Decision Tree Model for

Airport Terminal Departure Passenger Traffic Forecasting. Paper presented at

the CICTP 2014@ sSafe, Smart, and Sustainable Multimodal Transportation

Systems.

Chiang, P.-N. (2011). A decision support system for assessing conveyance options and

modeling passenger flow in airport terminals. (Dissertation/Thesis), ProQuest

Dissertations Publishing. Retrieved from

http://qut.summon.serialssolutions.com

Chitty, D. M., Yang, S., & Gongora, M. (2017). Considering flexibility in the

evolutionary dynamic optimisation of airport security lane schedules (Vol.

2018-, pp. 1-8): IEEE.

Chiu, C.-Y. (2002). Impacts of new large aircraft on passenger flows at international

airport terminals. (3108486 Ph.D.), The University of Texas at Austin, Ann

Arbor. Retrieved from http://gateway.library.qut.edu.au/login?url=http://

search.proquest.com/docview/305505909?accountid=13380

Chuin Lau, H. (1996). On the complexity of manpower shift scheduling. Computers

and Operations Research, 23(1), 93-102. doi:10.1016/0305-0548(94)00094-O

https://www.bitre.gov.au/about/index.aspx

https://eprints.qut.edu.au/72999/1/Lin_Cheng_Thesis.pdf

http://qut.summon.serialssolutions.com/

Bibliography 231

Correia, A. R., & Wirasinghe, S. C. (2013). Modeling airport landside performance

Modelling and Managing Airport Performance (pp. 1-26): John Wiley & Sons.

Curcio, D., Longo, F., Mirabelli, G., & Pappoff, E. (2007). Passengers’ flow analysis

and security issues in airport terminals using modeling & simulation. Paper

presented at the European Conference on Modeling & Simulation, Praga-

Repubblica Ceca.

De Neufville, R., & Odoni, A. (2003). Airport systems. planning, design and

management. New York, NY: McGraw-Hill, Incorporated.

De Neufville, R., Odoni, A. R., Belobaba, P. P., & Reynolds, T. G. (2013). Airport

systems: planning, design, and management. New York: McGraw-Hill

Education.

Diefenbach, M. (2010). Application of operations research techniques to improve

efficiency in the emergency department. (Dissertation/Thesis), Queensland

University of Technology. Retrieved from

http://qut.summon.serialssolutions.com/2.0.0/link

Diefenbach, M., & Kozan, E. (2011). Effects of bed configurations at a hospital

emergency department. J of Sim, 5(1), 44-57.

Diefenbach, M. L. (2010). Application of operations research techniques to improve

efficiency in the emergency department. (PhD Thesis), Queensland University

of Technology, Brisbane, Australia.

Ding, H., Lim, A., Rodrigues, B., & Zhu, Y. (2005). The over-constrained airport gate

assignment problem. Computers & Operations Research, 32(7), 1867-1880.

doi: http://dx.doi.org/10.1016/j.cor.2003.12.003

Dorokhov, O., Zolotaryova, I., Dorokhova, L. (2009). Discrete simulation modeling

of processes of customers service in retail drugstores. Management

Information Systems, 4, 39-44

Dorton, S. L. (2011). Analysis of Airport Security Screening Checkpoints using

Queuing Networks and Discrete Event Simulation: A Theoretical and

Empirical Approach. (Dissertation/Thesis), ProQuest Dissertations Publishing.

Retrieved from https://search-proquest-com.ezp01.library.qut.edu.au/

docview/1317978422?pq-origsite=summon

Dorton., S., & Liu, D. (2015). Effects of Baggage Volume and Alarm Rate on Airport

Security Screening Checkpoint Efficiency using Queuing Networks and

Discrete Event Simulation. Human Factors and Ergonomics in Manufacturing

& Service Industries.

Dowling, D., Krishnamoorthy, M., Mackenzie, H., & Sier, D. (1997). Staff rostering

at a large international airport. Annals of Operations Research, 72(0), 125-147.

doi:10.1023/a:1018992120116

Du, Y., Yu, S., Jiang, S., & Ji, Y. (2015). Distribution optimization model for

passenger departure via multimodal transit. Mathematical Problems in

Engineering, ID 858641(1024-123X).

Eilon, S., & Mathewson, S. (1973). A Simulation Study for the Design of an Air

Terminal Building. IEEE Transactions on Systems, Man, and Cybernetics,

3(4), 308-317. doi:10.1109/TSMC.1973.4309241

Eilon, S., & Mathewson, S. C. (1971). A simulation study for the design of an air

terminal building. Paper presented at the Proceedings of the 5th Conference on

Winter Simulation, New York, NY.

Fayez, M. S., Kaylani, A., Cope, D., Rychlik, N., & Mollaghasemi, M. (2008).

Managing airport operations using simulation. Journal of Simulation, 2(1), 41-

52. doi:10.1057/palgrave.jos.4250030

http://qut.summon.serialssolutions.com/2.0.0/link

232 Bibliography

Ferguson, J., Kara, A. Q., Hoffman, K., & Sherry, L. (2013). Estimating domestic US

airline cost of delay based on European model. Transportation Research Part

C: Emerging Technologies, 33, 311-323.

Fernandes, E., & Pacheco, R. R. (2002). Efficient use of airport capacity.

Transportation Research Part A: Policy and Practice, 36(3), 225-238. doi:

https://doi.org/10.1016/S0965-8564(00)00046-X

Fonseca, P., Casanovas, J., & Ferran, X. (2014). Passenger flow simulation in a hub

airport: An application to the Barcelona International Airport. Simulation

Modelling Practice and Theory, 44(0), 78-94. doi:

http://dx.doi.org/10.1016/j.simpat.2014.03.008

Fonzone, A., Schmöcker, J.-D., & Liu, R. (2015). A Model of Bus Bunching under

Reliability-based Passenger Arrival Patterns. Transportation Research

Procedia, 7, 276-299. doi: https://doi.org/10.1016/j.trpro.2015.06.015

Furian, N., Neubacher, D., Vössner, S., O’Sullivan, M., & Walker, C. (2014). Towards

holistic modeling and simulation of discrete event and individual based

behavior. Paper presented at the European Simulation and Modelling

Conference.

Genç, H. M., Erol, O. K., Eksin, İ., Berber, M. F., & Güleryüz, B. O. (2012). A

stochastic neighborhood search approach for airport gate assignment problem.

Expert Systems with Applications, 39(1), 316-327. doi:

http://dx.doi.org/10.1016/j.eswa.2011.07.021

Gronfula, M. G. (2014). Intelligent optimisation system for airport operation: Hajj

Terminal in Saudi Arabia. Brunel University London.

Guizzi, G., Murino, T., & Romano, E. (2009). A discrete event simulation to model

passenger flow in the airport terminal. Paper presented at the Proceedings of

the 11th WSEAS International Conference on Mathematical Methods and

Computational Techniques in Electrical Engineering (MMACTEE'09).

Heizer, J. (1996). Production & operations management (4th revis ed.): Prentice Hall.

Hsu, C.-I., Chao, C.-C., & Hsu, N.-W. (2014). Control strategies for departure process

delays at airport passenger terminals. Transportation Planning and

Technology, 38(2), 214-237.

Hsu, C.-I., Chao, C.-C., & Shih, K.-Y. (2012). Dynamic allocation of check-in

facilities and dynamic assignment of passengers at air terminals. Computers &

Industrial Engineering, 63(2), 410-417.

Hu, Y., & Chen, J. (2017). A Hybrid Parallel Queuing System Based on Passenger

Classification. Paper presented at the Proceedings of the 2017 International

Conference on E-Society, E-Education and E-Technology, Chengdu, China.

IATA. (2017a). Aviation Benefits Beyond Borders. Retrieved from

https://www.iata.org/pressroom/facts_figures/fact_sheets/Documents/fact-

sheet-economic-and-social-benefits-of-air-transport.pdf

IATA. (2017b). Economic performance of the airline industry. Retrieved from

https://www.iata.org/whatwedo/Documents/economics/IATA-Economic-

Performance-of-the-Industry-mid-year-2017-report.pdf

ICAO. (2018). Continued passenger traffic growth and robust air cargo demand in

2017 Retrieved from https://www.icao.int/Newsroom/Pages/Continued-

passenger-traffic-growth-and-robust-air-cargo-demand-in-2017.aspx

Imagine That Inc. (2013). ExtendSim User Guide (Ninth Edition. ed.).

Imrich, W., Klavžar, S., & Rall, D. F. (2008). Topics in graph theory: Graphs and

their Cartesian product: AK Peters/CRC Press.

https://doi.org/10.1016/j.trpro.2015.06.015

https://www.iata.org/pressroom/facts_figures/fact_sheets/Documents/fact-sheet-economic-and-social-benefits-of-air-transport.pdf

https://www.iata.org/pressroom/facts_figures/fact_sheets/Documents/fact-sheet-economic-and-social-benefits-of-air-transport.pdf

https://www.iata.org/whatwedo/Documents/economics/IATA-Economic-Performance-of-the-Industry-mid-year-2017-report.pdf

https://www.iata.org/whatwedo/Documents/economics/IATA-Economic-Performance-of-the-Industry-mid-year-2017-report.pdf

https://www.icao.int/Newsroom/Pages/Continued-passenger-traffic-growth-and-robust-air-cargo-demand-in-2017.aspx

https://www.icao.int/Newsroom/Pages/Continued-passenger-traffic-growth-and-robust-air-cargo-demand-in-2017.aspx

Bibliography 233

Ip, W. H., Chung, N., & Ho, G. (2010). Using Integer Programming for Airport Service

Planning in Staff Scheduling. International Journal of Engineering Business

Management, 2(2), 85-92. doi:10.5772/9716

Jacquillat, A. (2012). A queuing model of airport congestion and policy implications

at JFK and EWR. Massachusetts Institute of Technology. Retrieved from

https://dspace.mit.edu/bitstream/handle/1721.1/72652/808441276-

MIT.pdf?sequence=2

Jacquillat, A., & Odoni, A. R. (2015). Endogenous control of service rates in stochastic

and dynamic queuing models of airport congestion. Transportation Research

Part E: Logistics and Transportation Review, 73, 133-151.

Jensen, M. B. (2015). Scheduling at Aarhus Airport. Supervisor: Sin C. Ho:30-03.

Joshi, A. J. (2008). Study on the effect of different arrival patterns on an emergency

department's capacity using discrete event simulation. Kansas State

University. Retrieved from http://krex.k-state.edu/dspace/handle/2097/1024

Ju, Y., Wang, A., & Che, H. (2007, 21-25 Sept. 2007). Simulation and Optimization

for the Airport Passenger Flow. Paper presented at the 2007 International

Conference on Wireless Communications, Networking and Mobile

Computing.

Kalasky, D., M. Coffman, M. De Grano, and K. Field. 2010. Simulation-based

manpower planning with optimized scheduling in a distributed multi-user

environment. Paper presented at the Proceedings of the 2010 Winter

Simulation Conference (WSC), 5-8 Dec. 2010. Retrieved from

https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=5679034

Kamyszek, K. (2014). Modeling of the departure process at Poznań-Ławica Airport.

Journal of Mechanical and Transport Engineering, 66, 15-23. Retrieved from

https://www.infona.pl/resource/bwmeta1.element.baztech-9c97b53d-fb84-

4d8e-9de9-c4b4aad1b8e0

Kazda, A., & Caves, R. E. (2015). Airport Design and Operation Retrieved from

http://ebookcentral.proquest.com/lib/qut/detail.action?docID=2122284

Kierzkowski, A., & Kisiel, T. (2015). An impact of the operators and passengers

behavior on the airport's security screening reliability. Paper presented at the

Safety and Reliability: Methodology and Applications - Proceedings of the

European Safety and Reliability Conference, ESREL 2014.

Kierzkowski, A., & Kisiel, T. (2016). Simulation model of security control system

functioning: A case study of the Wroclaw Airport terminal. Journal of Air

Transport Management. doi: http://dx.doi.org/10.1016/j.jairtraman.

2016.09.008

Kierzkowski, A., & Kisiel, T. (2017). Evaluation of a Security Control Lane with the

Application of Fuzzy Logic. Procedia Engineering, 187, 656-663.

doi:http://dx.doi.org/10.1016/j.proeng.2017.04.427

Kim, S. H., Feron, E., Clarke, J.-P., Marzuoli, A., & Delahaye, D. (2013). Airport gate

scheduling for passengers, aircraft, and operation. arXiv preprint

arXiv:1301.3535.

Kirk, P. J. (2013). Passenger experience at airports : An activity-centred approach.

(Dissertation/Thesis), Queensland University of Technology. Retrieved from

https://eprints.qut.edu.au/60803/1/Philip_Kirk_Thesis.pdf

Kloosterziel, K., Oostra, R., Roche, E., Ringersma, P., Kuil, P., Banez, D., . . . Nuesink,

J. (2009). King Khalid International Airport Master Plan and Passenger

Terminal Concept General Authority of Civil Aviation. Riyadh, Kingdom of

Saudi Arabia, 1-251.

https://dspace.mit.edu/bitstream/handle/1721.1/72652/808441276-MIT.pdf?sequence=2

https://dspace.mit.edu/bitstream/handle/1721.1/72652/808441276-MIT.pdf?sequence=2

http://krex.k-state.edu/dspace/handle/2097/1024


https://www.infona.pl/resource/bwmeta1.element.baztech-9c97b53d-fb84-4d8e-9de9-c4b4aad1b8e0

https://www.infona.pl/resource/bwmeta1.element.baztech-9c97b53d-fb84-4d8e-9de9-c4b4aad1b8e0

http://ebookcentral.proquest.com/lib/qut/detail.action?docID=2122284

http://dx.doi.org/10.1016/j.proeng.2017.04.427

https://eprints.qut.edu.au/60803/1/Philip_Kirk_Thesis.pdf

234 Bibliography

Lapin, L. L. (1994). Quantitative methods for business decisions: with cases:

Wadsworth Pub Co.

Law, A. M., & Kelton, W. D. (1991). Simulation modeling and analysis (Vol. 2nd).

New York: McGraw-Hill.

Leone, K., & Liu, R. R. (2011). Improving airport security screening checkpoint

operations in the US via paced system design. Journal of Air Transport

Management, 17(2), 62-67.

Li, Y., Gao, X., Xu, Z., & Zhou, X. (2018). Network-based queuing model for

simulating passenger throughput at an airport security checkpoint. Journal of

Air Transport Management, 66, 13-24. doi:

https://doi.org/10.1016/j.jairtraman.2017.09.013

Lin, D., Xin, Z., & Huang, Y. (2015, 6-9 Dec. 2015). Ground crew rostering for the

airport check-in counter. Paper presented at the 2015 IEEE International

Conference on Industrial Engineering and Engineering Management (IEEM).

Livingstone, A., Popovic, V., Kraal, B. J., & Kirk, P. J. (2012). Understanding the

airport passenger landside retail experience. DRS 2012 Bangkok–Research:

Uncertainty, Contradiction and Value, 1.

Ma, W. (2013). Agent-based model of passenger flows in airport terminals.

(Dissertation/Thesis), Queensland University of Technology. Retrieved from

https://eprints.qut.edu.au/63457/1/Wenbo_Ma_Thesis.pdf

Ma, W., Fookes, C., Kleinschmidt, T., & Yarlagadda, P. (2012). Modelling passenger

flow at airport terminals: individual agent decision model for stochastic

passenger behaviour. Paper presented at the 2nd International Conference on

Simulation and Modeling Methodologies, Technologies and Applications

(SIMULTECH 2012), Rome, Italy.

Ma, W., Kleinschmidt, T., Fookes, C., & Yarlagadda, P. K. (2011). Check-in

processing: simulation of passengers with advanced traits. Paper presented at

the Proceedings of the Winter Simulation Conference.

Maertens, S., & Grimme, W. (2015). How to assess the percentage of transfer

passengers at airports? Discussion Paper Institut für Flughafenwesen und

Luftverkehr.

Manataki, I. E., & Zografos, K. G. (2009a). Development and demonstration of a

modeling framework for airport terminal planning and performance evaluation.

Transportation Research Record: Journal of the Transportation Research

Board, 2106(1), 66-75.

Manataki, I. E., & Zografos, K. G. (2009b). A generic system dynamics based tool for

airport terminal performance analysis. Transportation Research Part C:

Emerging Technologies, 17(4), 428-443. doi:

https://doi.org/10.1016/j.trc.2009.02.001

Manataki, I. E., & Zografos, K. G. (2010). Assessing airport terminal performance

using a system dynamics model. Journal of Air Transport Management, 16(2),

86-93.

Mason, A. J., Ryan, D. M., & Panton, D. M. (1998). Integrated simulation, heuristic

and optimisation approaches to staff scheduling. Operations Research, 46(2),

161-175.

Mazhar, S. (2015). A configurable airport reference model for passenger facilitation.

Queensland University of Technology.

Miller, J. G. (1978). Living systems. New York: McGraw-Hill.

https://eprints.qut.edu.au/63457/1/Wenbo_Ma_Thesis.pdf

https://doi.org/10.1016/j.trc.2009.02.001

Bibliography 235

Mohammadi, S., & Safa, M. (2016). A simulated annealing based optimization

algorithm for automatic variogram model fitting. Archives of Mining Sciences,

61(3), 635-649.

Mu, Q., Cheng, S., Zhang, Y., & Zhang, H. (2014). A Fuzzy Decision Tree Model for

Airport Terminal Departure Passenger Traffic Forecasting. Paper presented

at the Proceedings of the 14th COTA International Conference of

Transportation Professionals, Changsha, China.

Mujica, M. (2015). Check-in allocation improvements through the use of a simulation–

optimization approach. Transportation Research Part A: Policy and Practice,

77, 320-335. doi:http://dx.doi.org/10.1016/j.tra.2015.04.016

Nikoue, H., Marzuoli, A., Clarke, J.-P., Feron, E., & Peters, J. (2015). Passenger flow

predictions at sydney international airport: a data-driven queuing approach.

arXiv preprint arXiv:1508.04839.

Novrisal, D., Wahyuni, N., Hamani, N., Elmhamedi, A., & Soemardi, T. (2013).

Simulation of Departure Terminal in Soekarno-Hatta International Airport.

Paper presented at the Industrial Engineering and Engineering Management

(IEEM), 2013 IEEE International Conference Industrial Engineering and

Engineering Management, 10-13 Dec. 2013. Retrieved from


Odoni, A. R., & de Neufville, R. (1992). Passenger terminal design. Transportation

Research Part A: Policy and Practice, 26(1), 27-35.

Olaru, D., & Emery, S. (2007). Simulation and GA-optimisation for modeling the

operation of airport passenger terminals. Paper presented at the Proceedings

of the 29th conference of Australian institutes of transport research (CAITR).

Adelaide, Australia.

Owen, C. (2013). Selection of simulation tools for improving supply chain

performance. Aston University. Retrieved from

http://publications.aston.ac.uk/20896/1/Studentthesis-2013.pdf

Park, Y., & Ahn, S. B. (2003). Optimal assignment for check-in counters based on

passenger arrival behaviour at an airport. Transportation Planning and

Technology, 26(5), 397-416.

Parlar, M., Rodrigues, B., & Sharafali, M. (2013). On the allocation of exclusive-use

counters for airport check-in queues: static vs. dynamic policies. OPSEARCH,

50(3), 433-453. doi:10.1007/s12597-012-0110-5

Parlar, M., & Sharafali, M. (2008). Dynamic Allocation of Airline Check-In Counters:

A Queueing Optimization Approach. Management Science, 54(8), 1410-1424.

doi:10.1287/mnsc.1070.0842

Chiang, P. N., and K. Taaffe. 2014. Analysis of Passenger Flow in Airport Terminal.

Paper presented at the 2014 Tenth International Conference on Intelligent

Information Hiding and Multimedia Signal Processing, 27-29 Aug. 2014.


Pitchforth, J., Wu, P., Fookes, C., & Mengersen, K. (2015). Processing passengers

efficiently: An analysis of airport processing times for international

passengers. Journal of Air Transport Management, 49, 35-45.

doi:http://dx.doi.org/10.1016/j.jairtraman.2015.06.016

Popovic, V., Kraal, B., & Kirk, P. (2010). Towards airport passenger experience

models. Paper presented at the Proceedings of the 7th International Conference

on Design and Emotion.

Rauch, R., & Kljajić, M. (2006). Discrete event passenger flow simulation model for

an airport terminal capacity analysis. Organizacija, 39(10).

http://dx.doi.org/10.1016/j.tra.2015.04.016



http://dx.doi.org/10.1016/j.jairtraman.2015.06.016

236 Bibliography

Roanes-Lozano, E., Laita, L. M., & Roanes-Macıas, E. (2004). An accelerated-time

simulation of departing passengers’ flow in airport terminals. Mathematics and

Computers in Simulation, 67(1), 163-172. doi:

https://doi.org/10.1016/j.matcom.2004.05.016

Robertson, C. V., S. Shrader, D. R. Pendergraft, L. M. Johnson, and K. S. Silbert.

2002. The role of modeling demand in process re-engineering. Paper presented

at the Proceedings of the Winter Simulation Conference, 8-11 Dec. 2002.


Rodič, B., & Baggia, A. (2017). Airport Ground Crew Scheduling Using Heuristics

and Simulation. In M. Mujica Mota & I. Flores De La Mota (Eds.), Applied

Simulation and Optimization 2: New Applications in Logistics, Industrial and

Aeronautical Practice (pp. 131-160). Cham: Springer International Publishing.

Sabar, M., Montreuil, B., & Frayret, J.-M. (2012). An agent-based algorithm for

personnel shift-scheduling and rescheduling in flexible assembly lines. Journal

of Intelligent Manufacturing, 23(6), 2623-2634. doi:10.1007/s10845-011-

0582-9

Sachidananda, M., Erkoyuncu, J., Steenstra, D., & Michalska, S. (2016). Discrete

Event Simulation Modelling for Dynamic Decision Making in

Biopharmaceutical Manufacturing. Procedia CIRP, 49, 39-44.

doi:https://doi.org/10.1016/j.procir.2015.07.026

Sargent, R. G. (2013). Verification and validation of simulation models. Journal of

Simulation, 7(1), 12-24. doi:10.1057/jos.2012.20

Savrasovs, M., Medvedev, A., & Sincova, E. (2009). Riga airoport baggage handling

system simulation. Paper presented at the ECMS.

Schmidt, J. W., & Taylor, R. E. (1970). Simulation and analysis of industrial systems:

Richard D. Irwin.

Schultz, M., & Fricke, H. (2011). Managing passenger handling at airport terminals.

Paper presented at the 9th Air Traffic Management Research and Development

Seminars.

Shuchi, S. N. (2016). A novel concept for airport terminal design integrating

flexibility. (PhD Thesis), Queensland University of Technology. Retrieved

from https://eprints.qut.edu.au/93011/1/Sarah_Shuchi_Thesis.pdf.

Siadat, J., Arain, F., & Ruwanpura, J. (2012). Modeling Airport Check-In and Security

Procedures Using SimFC. Paper presented at the Construction Research

Congress 2012: Construction Challenges in a Flat World.

Siebers, P. O., Macal, C. M., Garnett, J., Buxton, D., & Pidd, M. (2010). Discrete-

event simulation is dead, long live agent-based simulation! Journal of

Simulation, 4(3), 204. doi:10.1057/jos.2010.14

Sigurðardóttir, G. L. (2011). Near-optimal staff scheduling using a mixed integer

model. (Masters Thesis), Reykjavík University, Skemman. Retrieved from

http://hdl.handle.net/1946/12761

Skorupski, J., & Uchroński, P. (2015). A fuzzy reasoning system for evaluating the

efficiency of cabin baggage screening at airports. Transportation Research

Part C: Emerging Technologies, 54(0), 157-175. doi:

http://dx.doi.org/10.1016/j.trc.2015.03.017

Skorupski, J., & Uchroński, P. (2016). Managing the process of passenger security

control at an airport using the fuzzy inference system. Expert Systems with

Applications, 54, 284-293. doi: http://dx.doi.org/10.1016/j.eswa.2015.11.014


https://doi.org/10.1016/j.procir.2015.07.026

https://eprints.qut.edu.au/93011/1/Sarah_Shuchi_Thesis.pdf

http://hdl.handle.net/1946/12761

Bibliography 237

Skorupski, J., & Uchroński, P. (2018). Evaluation of the effectiveness of an airport

passenger and baggage security screening system. Journal of Air Transport

Management, 66, 53-64. doi: https://doi.org/10.1016/j.jairtraman.2017.10.006

Solak, S., Clarke, J.-P. B., & Johnson, E. L. (2009). Airport terminal capacity planning.

Transportation Research Part B: Methodological, 43(6), 659-676.

Soukour, A. A., Devendeville, L., Lucet, C., & Moukrim, A. (2012). Staff scheduling

in airport security service. IFAC Proceedings Volumes, 45(6), 1413-1418. doi:

https://doi.org/10.3182/20120523-3-RO-2023.00169

Spyropoulos, C. D. (2000). AI planning and scheduling in the medical hospital

environment. Artificial Intelligence in Medicine, 20(2), 101-111. doi:

https://doi.org/10.1016/S0933-3657(00)00059-2

Statista. (2018). Annual growth in global air traffic passenger demand from 2005 to

2018. Retrieved from https://www.statista.com/statistics/193533/growth-of-

global-air-traffic-passenger-demand/

Statista. (2019). Number of passenger movements at Riyadh King Khalid International

Airport from 2005 to 2016 (in millions). Retrieved from

https://www.statista.com/statistics/675277/riyadh-passenger-movement-at-

international-airport/

Stefanik, M., Kandera, B., & Badanik, B. (2012). Aspects influencing arrival

behavioural pattern of air travellers. Archives of Transport System Telematics,

5, 35-40.

Sun, Y., & Schonfeld, P. (2015). Stochastic capacity expansion models for airport

facilities. Transportation Research Part B: Methodological, 80, 1-18. doi:

http://dx.doi.org/10.1016/j.trb.2015.06.009

Sun, Y., & Schonfeld, P. M. (2016). Capacity investment model for airport facilities

under demand uncertainty. Journal of Advanced Transportation, 50(8), 1896-

1911. doi:10.1002/atr.1435

Takakuwa, S., & Oyama, T. (2003). Modeling people flow: simulation analysis of

international-departure passenger flows in an airport terminal. Paper

presented at the Proceedings of the 35th Conference on Winter Simulation:

Driving Innovation, New Orleans, Louisiana.

Tang, J., Alam, S., Abbass, H., & Lokan, C. (2009). Modelling and evolutionary multi-

objective evaluation of interdependencies and work processes in airport

operations. Paper presented at the Proceedings of the first ACM/SIGEVO

Summit on Genetic and Evolutionary Computation, Shanghai, China.

Tarshizi, E. K. (2014). Multiple discrete-event simulation and animation models to

assist modern mining operations: University of Nevada, Reno.

Tošić, V. (1992). A review of airport passenger terminal operations analysis and

modelling. Transportation Research Part A: Policy and Practice, 26(1), 3-26.

doi: http://dx.doi.org/10.1016/0965-8564(92)90041-5

TranSafety, I. (1997). Study compares older and younger pedestrian walking speeds.

Road Management & Engineering Journal.

Transportarion Security Administration. (2009). Checkpoint Design Guide (CDG).

Retrieved from http://www.aci-na.org/static/entransit/OPT%20%20

Checkpoint%20Design%20Guide%20(CDG)%202009.pdf

van Boekhold, J., Faghri, A., & Li, M. (2014). Evaluating security screening

checkpoints for domestic flights using a general microscopic simulation model.

Journal of Transportation Security, 7(1), 45-67. doi:10.1007/s12198-013-

0129-8

https://doi.org/10.1016/S0933-3657(00)00059-2

https://www.statista.com/statistics/193533/growth-of-global-air-traffic-passenger-demand/

https://www.statista.com/statistics/193533/growth-of-global-air-traffic-passenger-demand/

https://www.statista.com/statistics/675277/riyadh-passenger-movement-at-international-airport/

https://www.statista.com/statistics/675277/riyadh-passenger-movement-at-international-airport/

http://dx.doi.org/10.1016/0965-8564(92)90041-5

238 Bibliography

Van Landeghem, H., & Beuselinck, A. (2002). Reducing passenger boarding time in

airplanes: a simulation based approach. European Journal of Operational

Research, 142(2), 294-308. doi: http://dx.doi.org/10.1016/S0377-

2217(01)00294-6

Verbraeck, A., & Valentin, E. (2002). Transportation applications of simulation:

simulation building blocks for airport terminal modeling. Paper presented at

the Proceedings of the 34th Conference on Winter Simulation: Exploring New

Frontiers, San Diego, California.

Williams, K. A., Chambers, C. G., Dada, M., McLeod, J. C., & Ulatowski, J. A. (2014).

Patient punctuality and clinic performance: observations from an academic-

based private practice pain centre: a prospective quality improvement study.

BMJ Open, 4(5). doi:10.1136/bmjopen-2013-004679

Winston, W. L., & Goldberg, J. B. (2004). Operations research: applications and

algorithms (4th ed.). Belmont, CA: Thomson Brooks/Cole.

Wu, P. P.-Y., & Mengersen, K. (2013). A review of models and model usage scenarios

for an airport complex system. Transportation Research Part A: Policy and

Practice, 47, 124-140.

Wu, P. P.-Y., Pitchforth, J., & Mengersen, K. (2014). A Hybrid Queue-based Bayesian

Network framework for passenger facilitation modelling. Transportation

Research Part C: Emerging Technologies, 46, 247-260. doi:

http://dx.doi.org/10.1016/j.trc.2014.05.005

Xin, Z., Lin, D., Huang, Y., Cheng, W., & Chong Teo, C. (2014). Design of service

capacity for the ground crew at the airport check-in counters. International

Journal of Quality and Service Sciences, 6(1), 43-59.

Yamada, H., Ohori, K., Iwao, T., Kira, A., Kamiyama, N., Yoshida, H., & Anai, H.

(2017). Modeling and managing airport passenger flow under uncertainty: A

case of Fukuoka airport in Japan, Springer Verlag, 10540, 419-430.

Yan, S., Tang, C.-H., & Chen, J.-H. (2014). Common-use check-in counter

reassignments with a variable number of service lines and variable length of

time window. JOURNAL OF THE CHINESE INSTITUTE OF ENGINEERS,

37(5), 643-658. doi:10.1080/02533839.2013.815009

Zhou, X., Huang, H., Jia, X., & Jiang, N. (2014). Integrated passenger terminal

capacity analysis under the capacity constraints of surrounding road network.

Physica A: Statistical Mechanics and its Applications, 414, 194-203.

Zidarova, E. D., & Zografos, K. G. (2011). Measuring quality of service in airport

passenger terminals. Transportation Research Record: Journal of the

Transportation Research Board, 2214(1), 69-76.

Zografos, K. G., & Madas, M. A. (2006). Development and demonstration of an

integrated decision support system for airport performance analysis.

Transportation Research Part C: Emerging Technologies, 14(1), 1-17.

Zografos, K. G., Madas, M. A., & Salouras, Y. (2013). A decision support system for

total airport operations management and planning. Journal of Advanced

Transportation, 47(2), 170-189.

http://dx.doi.org/10.1016/S0377-2217(01)00294-6

http://dx.doi.org/10.1016/S0377-2217(01)00294-6

Appendices 239

Appendices

Appendix A: Simulation Parameters

The flight schedule data for outbound in international Australian airport.

Table A- 1: Timetable of outbound flight used in the model

Airline Departure

time

Number of

PAX

Airline

Group

Departure

Gate Emirates 3.3 222 1 1

Qantas 6.2 134 2 2

Emirates 7.25 222 3 3

Air Vanuatu 8.2 134 4 4

Aircalin 8.2 134 5 5

Cathay Pacific 8.3 134 1 6

China Airlines 9 134 2 7

China Southern

Airlines

9 128 3 8

Etihad 9.1 241 4 9

Jetstar 9.15 134 5 10

Air Canada 9.3 120 1 11

Air New Zealand 9.3 51 2 12

Air Niugini 9.35 241 3 13

Air Vanuatu 10.3 134 4 14

Aircalin 10.3 134 5 15

Cathay Pacific 10.4 180 1 16

China Airlines 11.15 222 2 17

China Southern Airlines

11.2 251 3 18

Emirates 12.3 258 4 19

Etihad 13.4 258 5 1

Fiji Airways 14.2 258 1 2

Hawaiian Airlines

14.4 265 2 3

Jetstar 16.45 134 3 4

Korean Air 17.4 99 4 5

Nauru Airlines 17.4 134 5 6

Philippine

Airlines

17.45 134 1 7

Qantas 18 134 2 8

Singapore

Airlines

18.3 134 3 9

Solomon Airlines 18.35 180 4 10

Thai International 20.45 222 5 11

Virgin Australia 21.2 134 1 12

Singapore

Airlines

22.4 134 2 13

Solomon Airlines 22.45 243 3 14

Thai International 23.45 265 4 15

Virgin Australia 23.5 265 5 16

240 Appendices

The flight schedule data for inbound international Australian airport.

Table A- 2: Timetable of inbound flight used in the model

Flight # Arrival time Flight capacity Arrival Gate VA46 5:45 176 1 QF16 6:10 353 2 QF52 6:15 297 3 QF52 6:15 239 4 EK434 6:25 489 5 QF62 6:45 297 6 KE123 6:50 276 7 QF98 6:50 297 8 VA8 6:50 361 9

SQ235 7:05 285 10 AC35 7:15 251 11

QF124 7:15 168 12 QF68 7:30 297 13

VA153 7:35 176 14 VA103 8:00 176 15 QF124 8:15 168 16 CZ381 8:25 284 17 VA153 8:35 176 18 VA127 8:45 176 19

PX3 9:25 188 1 CX103 9:50 251 2

VA119Z 10:05 176 3 NZ135 10:05 332 4 FJ921 10:30 118 5 SQ255 10:35 285 6 CI53 10:45 313 7

PR221 11:00 368 8 NZ135 11:05 332 9 TG473 11:50 264 10

JQ6 13:20 335 11 VA115 14:25 176 12 PE532 14:45 264 13 IE700 15:30 156 14 VA176 15:45 275 15 QF134 16:30 168 16 NZ805 16:30 168 17 NZ739 16:35 168 18

PX5 16:40 188 19 QF126 16:55 168 1 VA107 17:10 176 2 VA188 17:10 176 3 VA170 17:25 176 4 NZ805 17:30 168 5 NZ739 17:35 168 6 VA117 17:35 176 7 EY484 17:40 231 8 ON1 17:45 130 9

QF126 17:55 168 10 NF20 18:10 170 11 SB150 18:20 146 12 EK435 19:10 489 13 SQ245 19:30 285 14 HA443 19:45 294 15 NZ733 20:20 168 16 CI54 21:15 313 17

NZ733 21:20 168 18 CX157 22:15 251 19 EK432 22:30 354 1 VA161 22:50 176 2

Appendices 241

Appendix B:

Code of the development of Airport library and Advanced Resource

management (please see the link below):

https://www.dropbox.com/home/Airport%20code

https://www.dropbox.com/home/Airport%20code