Flying around Slot Regulations
Improving Airport Utilization
by Declared Capacity
Master’s Thesis: Final Version
By Marnix L. Reijgersberg
08 December 2016
Abstract
Air traffic demand grew steeply over the last decade and is expected to keep on growing. On the other hand airport physical capacity congestion increases which will result in unmet flight demand. Airports in the EU are under slot coordination regulations which are designed to coordinate flight movements to serve physical capacity. It is shown that the current regulations are ineffectively doing this and therefore current literature has unsuccessfully focused on changing the regulations. This research has the objective to explore how airports themselves can influence slot coordination by their own capacity declaration in order to better utilize physical capacity, without changing the EU regulations. The capacity declaration has never been a field of study. This research has used a single case at Amsterdam Airport Schiphol, one of the busiest and most congested airports in the world which recently experiences problems in their pier gate utilization. In this study the effects of the capacity declaration are analysed on the gate utilization and different opportunities are created and evaluated and effective opportunities are found. This study also introduces a first classification in capacity declaration characteristics, which can be the starting point of a new line of studies.
Acknowledgement
I would like this chance to thank the people without whom this study would not have come together. First of all I would like to thank my academic supervisors Dr. Martin Land and Dr. Gu Pang. Especially the thorough feedback of Dr. Martin Land was appreciated much. Then I would like to thank Mark van Gaalen without whom the case study on Schiphol would not have been possible and for his feedback and opportunities. Thirdly I would like to thank every professional and academic expert who has taken the time to answer questions, both during the formal interviews and during coffee machine conversations. I also want to thank Marcel for doing his research in the same setting and his time giving feedback and helping me in the aviation jargon. Finally I would like to thank my parents, my flat mate and my girlfriend for their patience during the process and proof-reading and feedback.
Word count: 11.954
Master’s Thesis: Final Version
Flying Around Slot Regulations: Improving Airport Utilization by Declared Capacity 08 December 2016
M.L. (Marnix) Reijgersberg BSc. s2031477/ b5063053
m.l.reijgersberg@student.rug.nl
(+31) 6 159 039 36
MSc. Dual Degree in Operations Management 2015-2017 Course codes: EBM028A30/ NBS8399
Supervisor & Assessor: Dr. M.J. Land (m.j.land@rug.nl)
Second supervisor & Co-assessor: Dr. G. Pang (gu.pang@ncl.ac.uk)
Universities
University of Groningen, Faculty of Economics and Business Nettelbosje 2, 9747 AE Groningen, the Netherlands
Newcastle University Business School
5 Barrack Road, Newcastle upon Tyne, NE1 4SE, United Kingdom
Case company
Amsterdam Airport Schiphol
Table of contents
Abstract ... 3 Acknowledgement ... 4 Table of contents ... 5 1. Introduction ... 6 2. Theoretical Background ... 72.1. Airport Capacity and Utilization ... 7
2.1.1. Capacity Problems ... 7
2.2. Slot Coordination Process ... 8
2.2.1. Capacity Declaration ... 8
2.2.2. Final Coordination ... 9
2.2.3. Slot Coordination Rules and their Limitations ... 9
2.3. Improving Slot Coordination and its Implications ... 10
2.4. Research Question ... 11 3. Research Method ... 12 3.1. Case Selection ... 12 3.2. Data Collection ... 12 3.3. Data Analysis ... 13 3.4. Quality Assurance ... 14 3.5. Terminology ... 15 4. Case Analysis... 16 4.1. Descriptive Analysis ... 16
4.1.1. Yearly and Seasonal Movements... 16
4.1.2. Aircrafts in Position ... 17
4.1.3. Summary ... 18
4.2. In-Depth Analysis: Arrivals and Departures ... 19
4.2.1. Aircrafts in Position ... 19
4.2.2. Actual Movements versus Scheduled Movements ... 20
4.2.3. Scheduled Movements versus Coordinated Movements ... 21
4.2.4. Coordinated Movements versus Declared Capacity ... 22
4.2.5. Consequences of Declared Capacity... 23
4.2.6. Summary ... 25
5. Opportunity Exploration ... 26
5.1. Opportunity Selection ... 26
5.1.1. Change Capacity Criteria ... 26
5.1.2. Increase Punctuality ... 27
5.2. Opportunity Validation ... 27
5.2.1. Design of an Evaluation Model ... 27
5.2.2. Scenario 1: Decrease Peak Capacity ... 29
5.2.3. Scenario 2: Adding Departure Peaks ... 30
5.2.4. Scenario 3: Increasing Punctuality ... 31
5.2.5. Validating the Model ... 32
5.3. Summary ... 32
6. Conclusions and Discussion ... 33
6.1. Limitations and Future Research ... 33
6.2. Implications for Literature and Management ... 33
7. References ... 35
8. Appendix ... 37
8.1. Appendix 1: Capacity Declaration parameters ... 37
8.1.1. Amsterdam Airport Schiphol ... 37
8.1.2. Dusseldorf and Frankfurt Airport ... 38
8.1.3. Rotterdam - The Hague Airport ... 38
8.2. Appendix 2: Comparison between Third Quartile and Input Data ... 39
8.3. Appendix 3: Interview Guide... 39
8.4. Appendix 4: Interview Elements ... 40
8.5. Appendix 5: Pier Level Analysis ... 49
8.6. Appendix 6: Specific Gate Type Capacity Analysis ... 50
1. Introduction
Over the last decades both European and global flight movements and passenger numbers have grown enormously (World Bank 2015), and a stable growth is expected to hold on (Eurocontrol 2015). By increasing flight movements and passenger numbers capacity limits of airports are reached, especially during demand peaks. The major limiting factors are runway capacity, taxiway capacity, gate capacity and environmental constraints such as noise and emission limits (Hasan et al. 2009). One way of coping with reached capacity limits is expanding capacity by building new physical facilities (for example terminals and runways), which is – besides capital intensive (Sheng et al. 2015) – not always possible or constrained by existing metropolitan structures (Burghouwt 2013). Higher utilization of existing capacity might be a suitable alternative (Abeyratne 2000; Sheng et al. 2015).
Worldwide, airports have a capacity level (1-3) designated by the International Air Transport Association (IATA). If demand for airport usage exceeds capacity an airport is defined as Level 3 and slot coordination (or slot allocation) is needed. Globally there are 179 Level 3 airports whereof 107 are in Europe (IATA 2016). In Europe these rules are officially defined in the 95/93 EU regulation (European Commission 2016). In order to land on or take off from such an airport, carriers need slots allocated. These slots are time periods which allow the carrier to commence their traffic movement. Officially a slot is defined as follows: “An airport slot (or ‘slot’) is a permission given by a coordinator for a planned operation to use the full range of airport infrastructure necessary to arrive or depart at a Level 3 airport on a specific date and time”(IATA 2015, p.16). Its process all starts with the declared capacity of an airport itself in specific parameters, the so-called capacity declaration. On the basis of those capacity parameters slots can be coordinated. If a carrier has a single slot allocated over a continuous period of time it may claim interests over this “historic/ grandfather” slot (Abeyratne 2000). Historic rights are one of the biggest reasons of inefficient use of airport facilities.
Alternatives to the current EU slot regulations have been developed and described in literature extensively (Madas & Zografos 2006; Madas & Zografos 2008; Czerny et al. 2008; Czerny & Zhang 2014). Changing the 95/93 regulation is a political journey but no concrete changes have been implemented yet and they are not expected to happen soon. Therefore this study looks at what airports themselves can do to influence the slot coordination, which, as said, all starts with the capacity declaration. The capacity declaration has never been a subject of study and has been taken for granted before. Therefore this study has the objective to analyse this phenomenon in-depth to find how it can help airports to better utilize physical capacity.
This study took place at Amsterdam Airport Schiphol (AAS) which is a major hub airport in Europe and worldwide fifth in terms of international passenger number (2015: 58 million) (ACI 2016). Recently AAS has started to experience gate capacity problems. AAS’ slot coordination is done by Stichting Airport Coordination Netherlands (SACN), which is responsible for the slot coordination of AAS, Rotterdam - The Hague Airport and Eindhoven Airport (SACN 2016a). By the use of data analyses and semi-structured interviews the effects of the capacity declaration on the physical capacity are analysed. These analyses are also used to select opportunities and to develop a method to assess opportunities in a simulation study.
2. Theoretical Background
Before we come to the research questions, we have to take a few introducing steps to explore current literature for a better understanding of the broader picture. Firstly airport capacity is outlined to introduce this subject and to demonstrate the problem. Thereafter the current slot coordination process is described. Then, existing literature on this problem is discussed and lead to the relevance of this research which, lastly, introduces the research question.
2.1. Airport Capacity and Utilization
In Operations Management capacity is simply defined as the upper limit of process output. There is always a difference in theoretical capacity and practically used capacity which is measured in utilization. Only in perfect systems with no variability 100% utilization is possible (Hopp & Spearman 2008). With a constant inter-arrival time and a constant processing time a system can be utilized absolutely without waiting times. In practice, however, there is variability in inter-arrival time and processing time which creates the need for buffers, either by making queues or by under-utilizing capacity. Increasing utilization is most often a strategic business objective but has its costs in flexibility and/or service level (in aviation called Level of Service (LoS) (Bubalo 2011); examples of service parameters are delays, type of handling (on pier or remote) and transfer connection times).
An airport’s capacity limit is determined by its infrastructure, e.g. terminals, taxiways, runways and gates, or by governmental regulations, e.g. night departures, noise restrictions, pollution restrictions (Sheng et al. 2015). An airport’s utilization level usually is around the 60-70% of theoretical capacity and is determined by its smallest component (Wilken et al. 2011), which in operations management we know as ‘the bottleneck’. Airports choose to utilize their theoretical capacity less because of time delays that aircrafts encounter during peak hours, which could be seen as inter-arrival time variability for the airport. If an airport utilizes its capacity more there are not enough possibilities to cope with these delays which, as result, will increase intolerably (Wilken et al. 2011). Bubalo (2011) describes a method of finding airport capacity by so-called design peak days. An absolute demand peak will happen once in 10 to 30 days, this day then is chosen to characterize airport capacity in terms of airside capacity (e.g. runways, aprons, gates) and landside capacity (passenger facilities and airport access) (Bubalo 2011). The maximum number of aircrafts the airport could facilitate within the chosen service level is then considered as the airport’s capacity.
Lots of landside and airside factors influence overall airport capacity. For some airports one of the most important limiting factors in airport capacity is the gate capacity. Horonjeff et al. ( 2010, p. 538) have defined it in terms of how many aircrafts a number of gates can accommodate during a specific time period. Affecting utilization factors are the number of available gates, the mix of demanding aircrafts, the gate occupancy time, moving times into and out of gate, delay times and restrictions in using any or all gates.
2.1.1.
Capacity Problems
Figure 1: Worldwide air passenger growth (data from World Bank (World Bank 2015)).
Peaks are a major problem in capacity management. AAS, for example, has an average of 50 movements an hour over a given day, but has maximum peaks of 110 movements an hour (Amsterdam Airport Schiphol 2016). Peaks are a result of bans on night movements at the arriving or departing airport and bigger structures as transfer flights and destination / origin airport schedules. A figurative example of peak movements is given in Figure 2. It is defined in seats (passengers). It is not hard for us to imagine that this will create serious challenges for airport capacity management since it clearly demonstrates that this is not a perfect system at all. Compared to make-to-order companies airports face challenges in punctuality and service level, and therefore time buffers and backorders are not possible to deal with demand peaks (Bubalo 2011). So over-capacity is used to cope with variability.
Figure 2: Seat movements at AAS at a design peak day (Bubalo 2011).
2.2. Slot Coordination Process
Airports with higher flight demand than capacity are assigned a Level-3 status by IATA, the worldwide air carrier lobby, and therefore require slot coordination for air movement planning purposes (IATA 2015). Under European regulations, monetary slot trading is prohibited and therefore independent coordination committees or state administrators facilitate this coordination, as described by Sheng et al. (2015) who studied the effects of allowing monetary slot trading.
2.2.1.
Capacity Declaration
Slots are coordinated twice per year, divided in summer and winter season. A year before slots are coordinated, airports examine their critical capacity components and determine maximal capacity
-5% 0% 5% 10% 15% 0 5 10 15 20 25 1975 1980 1985 1990 1995 2000 2005 2010 A n n u al gr o wt h Year ly p assen ge r am o u n t (b ill io n s)
Worldwide air passengers
based on a required service level and traffic forecasts. Six months before slot coordination, a capacity declaration statement is made by airports for the slot coordinator. This statement usually consists of arrival and departure capacity for different time slot durations, or blocks. In Europe the length of time slots varies between 10 to 60 minutes and they might be defined strictly or more flexible (rolling). The capacity declaration might also include other parameters such as gate, airway and terminal capacity (SACN 2016b; Airport Coordination Germany 2016). After the capacity declaration finalization, carriers submit requests for certain slots (Kösters 2007).
2.2.2.
Final Coordination
About four months before start of season, a worldwide scheduling conference takes place with delegates from slot coordinators of the 179 airports and the carriers demanding to operate on these airports. This is where the slots are definitely allocated to carriers, but exchanges are still possible. At the semi-annual conference the carriers can either accept or return initially allocated slots. Bilateral discussions are held between carriers or between carrier and slot coordinator to adjust schedules. Because of the returns and rescheduling this is a very dynamic conference.
About two months before the season starts there is a slot return date. Carriers possessing a certain slot, but not planning to use them, must return those slots before that date. If they keep this slot it causes disadvantages in the next slot coordinating season. Minor adjustments are possible up to two days before operations, as shown by Kösters (2007) who studied the planning horizons for airport capacity (figure 3). Carriers may not operate out of given slot times deliberately (IATA 2015).
Figure 3: The airport scheduling process, it stages and its different actors (Kösters 2007)
2.2.3.
Slot Coordination Rules and their Limitations
90%) of slots are inherited because of grandfather rights (Madas & Zografos 2008; Sheng et al. 2015). The slot coordinator cannot withdraw historic slots from an carrier (IATA 2015), which decreases possibilities for change schedules and decreases flexibility during coordination. In their study, which offered possibilities to change the current set of European slot regulations, Madas & Zografos (2008) have proposed to prevent babysitting by losing the grandfather rule.
Then, slots that are not inherited and all newly generated slots are put in a “slot pool” (IATA 2015). IATA guidelines (2015) describe that at least 50% of the slots in the slot pool must be made available for new entrants to prevent the monopoly of existing carriers. This will give the new entrants priority in their slot requests. The second half of the slot pool will be allocated over the remaining requests by the slot coordinator. Because of the small number of remaining slots at hub airports this decreases the flexibility of the slot coordinator.
2.3. Improving Slot Coordination and its Implications
Several authors acknowledge the insufficiency of current slot coordination systems (Madas & Zografos 2006; Madas & Zografos 2008; Czerny et al. 2008; Czerny & Zhang 2014; Avenali et al. 2015). All authors agree that grandfather rights and use-it-or-lose-it may be within the core of the problem and some say that economy-based trade models might be good alternatives. Directions for solutions are introduced to change certain rules within the slot coordination process. Madas & Zografos (2006) produced a framework of every rule used in slot coordination and provided several alternatives to change some of these rules in the 95/93 EU regulation. Those changes are in removing the grandfather rights and/or use-it-or-lose-it, controlled trading of slots, congestion fees on popular slots or policy-designated slots. The same authors have developed a method to select a right policy based on airport characteristics in terms of size (Madas & Zografos 2008). The European Parliament has studied the effects of the proposed changes and concluded that implementing them will increase passenger numbers up to 4% without adding physical capacity (European Commission 2011).
Since 1993 the rules have been given small amendments for four times: in 2002, in 2003, in 2004 and in 2009 (EUR-lex 2010). Amendments and the running proposal of 2011 are summarized in table 1:
Amendment (year) Need for change Implications
Regulation No 894 (2002)
Terrorist attacks of 9-11-2001 reduced demand of air traffic. Chance of carriers losing grandfather rights.
Air carriers were entitled to same slots as before the attacks.
Regulation No 1554 (2003)
Iraq war and SARS outbreak reduced demand of air traffic. Chance of carriers losing grandfather rights.
Air carriers were entitled to same slots as before the events.
Regulation No 793 (2004)
New entrants and market access issues and congested airports.
Definition changes and clarification. Measures against carrier misuse.
Regulation No 545 (2009)
The financial crisis of 2008 reduced demand of air traffic. Chance of carriers losing grandfather rights.
Air carriers were entitled to same slots as before the crisis.
Proposal COM 827 (2011)
Inefficient utilization of airport facilities. Proposing multiple changes, among
which developed by Madas & Zografos (2006).
Table 1: Summary of amendments on the 95/93 EU slot regulation (European Parliament 2002; European Parliament 2003; European Parliament 2004; European Parliament 2009).
legal work from the airports themselves and probably carriers will attempt to prevent it from happening (as found in the European Commission study: “Air carriers are broadly satisfied with the functioning of the current Slot Regulation, so most respondents within this group do not support any changes” (European Commission 2011, p.5).
Also since the basis of the rules is in the IATA framework, which is the air carrier lobby organisation, the balance in this discussion tips towards the air carriers. Therefore changing the EU rules is seen as a hard process for airports and this stream of research, although interesting, as not pragmatic.
2.4. Research Question
The European airport capacity problem has been shown as a real problem and is expected to lead to unmet flight demand. Because many congested airports are bound by location and locked into metropolitan structures adding physical capacity is almost impossible and very capital intensive. Therefore given capacity must be utilized better. In Europe flights are allocated by an independent party according to the 95/93 EU slot regulation. This regulation bounds airports to use their capacity effectively.
Current research (Madas & Zografos 2006; Madas & Zografos 2008; Czerny et al. 2008; Czerny & Zhang 2014; Avenali et al. 2015) was aiming at changing the EU slot regulation. We have seen that changing these rules is out of control of the airport and requires lots of legal and lobby work and will be counteracted by the air carriers. This does not say it will never happen but chances seem low for now. The capacity declaration is the starting point of the slot coordination and provided by the airport itself, and therefore may give easier opportunities to control physical capacity utilization by defining allowed movements. In existing literature the capacity declaration has not been researched as a phenomenon on itself and therefore this study could fill that research gap. Madas & Zografos (2006) only report on it stealthily: “(…) whilst others [airports] have applied different capacity declaration schemes by pursuing capacity segmentations to safeguard regional services(…)” (p. 54), and Kösters (2007) takes its process and outcomes for granted in his framework (see figure 3).
Therefore this new point of view results in the following research question:
3. Research Method
Since the capacity declaration and its possibilities to optimize physical capacity utilization are phenomena which have not been studied before, a case study is commenced. The choice has been made to use a single case to study the system and the possibilities more in depth. According to Yin (2014) this kind of case study serves this goal perfectly. Single case studies have been used more often in airport capacity research to look at specific problems, like in Diepen et al.'s (2012) study in solving a scheduling problem and using data of AAS and Katsaros & Psaraki's (2012) study in slot misuse.
3.1. Case Selection
Airport capacity is a niche wherein Madas & Zografos (2008) made a categorization, to specify its clusters in terms of the percentage of granted slots and congestion: Small Spokes serve as satellites in hub structures and have more slots than requested and a high number of grandfathered slots (80%), e.g. Venice; Large Spokes and Small Hubs are larger in channelling through to hub airports, have less slots than requested but grandfathered slots are low (20%), e.g. Athens; Large International Hubs have a big national hub function, have about as much slots as requested but have a high percentage of grandfathered slots (71%), e.g. Copenhagen; Super Hubs are the busiest and most congested international hubs, with a substantial gap between requested and available slots and the highest percentage of grandfathered slots (90%), e.g. Heathrow. By using a case within one of the clusters generalizability within that cluster probably will be higher. This study also validates findings with academic experts.
The selected case had the criterion of an airport under slot coordination with many capacity stakeholders and a capacity problem. The unit of analysis must be a complete capacity component and its influence on other components at an airport.
Amsterdam Airport Schiphol (AAS), which is a major hub airport in Europe and worldwide fifth in terms of international passenger number (2015: 58 million) (ACI 2016), is chosen as case company. According to Madas & Zografos (2008) it fits within their Super Hub category of highly congested international hubs with a high number of grandfathered slots.
Slot coordination at AAS now done according to two parameters: noise and runway capacity but they start to experience problems in gate capacity utilization. AAS’s service level mission is to handle all aircrafts at a connected position, which is impossible right now (Visscher 2015). Because of the high passenger numbers and its complicated network structure the total set of gates at AAS is a good unit of analysis. AAS’s slot coordination is done by Stichting Airport Coordination Netherlands (SACN), which is responsible for the slot coordination of AAS, Rotterdam - The Hague Airport and Eindhoven Airport (SACN 2016a).
3.2. Data Collection
The data collection in this case study was twofold: quantitative and qualitative data. The quantitative data on aircraft movements (arrivals and departures) and positions have been gathered by exporting it from the ERP systems of AAS and SACN. It consists of in- and outbound (air) movement details, ground positions and movements and coordinated slots. The data needed restructuring in order to give the required information. The raw data only included movements at specific times from and to a location (either on ground or another airport). It was organized to show the amount of movements on per time interval and the amount of aircrafts in a ground position.
3.3. Data Analysis
In order to get to the core of the problem data analyses have been done. The behaviour of the gate utilization is traced back from the actual movements to the capacity declaration. This is used to understand the system in order to design a model to test opportunities.
According to Montgomery et al. (2015) data analysis starts with graphical displays of the given data to spot data characteristics by eye. By doing this on a macro-scale, problem areas were spotted and then analysed more in-depth. In this study we have done the following:
Descriptive analysis
o Yearly and Seasonally Movements o Gate Utilization
First seasonal patterns which might affect capacity were examined. Then the gate utilization, currently one of the biggest capacity constraints at AAS, was looked into to get a feeling for the problem and its possible causes. This is then followed by:
In-Depth Analysis and Design
o Movement planning and realisation comparisons Actual versus carrier schedule
Carrier schedule versus slot coordination Slot coordination versus declared capacity Actual versus declared capacity
The movements, which result into gate utilization, were explored more in-depth. The actual movements were compared to the different planning horizons which Kösters (2007) also described in his study: the carrier schedules, the coordinated slots and the airport’s capacity declaration to get to the root of the problem’s cause. This in-depth analysis also is used to understand the system and to use for design purposes when evaluating opportunities. Qualitative interviews have been transcribed and summarized (see Appendix 4), to increase understanding of structures and methods and to clarify the in-depth analysis. Based on the findings from the in-depth analysis and the interviews opportunities have been found. Finally:
Evaluative analysis
o Input data selection o Model design
o Simulation execution and analysis
3.4. Quality Assurance
The quality of this study and its data has to be assured. The quality criteria are discussed in Table 2 in short:
Criterion Coping method
Construct Validity - Multiple information sources (different interviews within the airport and with other stakeholders, multiple datasets and literature sources);
- Validation of the proposed framework with academic experts to verify interpretations;
- Peer debriefing with an academic audience at university.
Internal Validity - Data triangulation by asking the same questions to multiple people within the airport organization;
- Multiple internal sources (multiple interviews, multiple datasets); - Comparison of found results in data analyses by confirmation during
interviews.
External Validity - Generalizability within airport coordination niche;
- Airport category generalization of Madas & Zografos (2008); - In-depth analysis is performed to study a new phenomenon; - Findings are compared with other airports;
- Verification with academic experts.
Reliability - An interview protocol is used (Appendix 3 and 4);
- Transcription is used before interpretation (summaries in Appendix 4);
- Interviewees had opportunities to revise the interpretations, the summaries and interpretations;
- Data analysis revision by experts.
3.5. Terminology
In the case company and in the aviation industry different terms are used to define phenomena or systems than literature does. To be clear to every reader a list of variables is given in table 3:
Variable Definition
Carriers Companies managing the aircrafts/ airplanes (e.g. British Airways, EasyJet). Carriers have different strategies in transfer passengers, gate requirements and handling/ turnaround times. The carriers are the direct customers of the airport. Also: Airline
Pier gate A stand for aircrafts, connected to the passenger terminal by an air bridge, or where a passenger may walk from the aircraft to the terminal (walk-in/ walk-out) using both aircraft doors and stairs on the platform. Also: Connected gate
Buffer gate A remote stand for aircrafts, used for temporary storage of an aircraft or remote handling of passengers by bus. Also: Remote position or “concrete”
Bus gate A door in the terminal used by buses to handle passenger to or from an aircraft at a remote stand.
Movements Twofold definition: 1) flight movement either arrivals or departures to and from an airport; 2) ground movements from one position (runway, pier gate, buffer gate) to another.
Coordinated movements
Coordinated movements are the flight movements coordinated by the slot coordinator. Also: Allocated slot
Scheduled movements
Scheduled movements are the flight movements scheduled by the carriers.
Operation movements
Operation movements are the flight movements realized by the carriers.
AIP Aircrafts In Position: The amount of aircrafts standing at a position, either pier or buffer gate.
In/out block The actual moment an aircraft moves into/leaves at a position and a block is put/removed behind its wheels.
Schengen/ Non-Schengen/ North American screened
Refers to the Schengen agreement within Europe wherein international borders are abolished. Passengers may travel through certain terminals without going through border control. North American security is tighter than Non-Schengen because of a screening process of passengers. Therefore flights have different security and border control requirements and gates have different characteristics.
Cat. 1-9 Categorization on AAS to refer to aircraft size, 1 is very small and 9 is very big. Different category aircrafts need different gate sizes. A small aircraft may stand in a big gate but vice versa is impossible. An easier categorization is twofold: narrow-body and wide-body.
4. Case Analysis
First we do a descriptive analysis to better understand the data and the gate utilization problem at the case company. Then, in the in-depth-analysis, we attempt to get to the root of the problem’s cause by analysing the planning horizons and its influence on another. Both quantitative data analyses and results from interviews are used to integrally explain the system’s behaviour. These analyses are then used to find opportunities and to design an opportunity evaluation tool.
4.1. Descriptive Analysis
The descriptive analysis is divided in the yearly and seasonal movements and gate utilization figures, which are called aircrafts in position (AIP).
4.1.1.
Yearly and Seasonal Movements
Four quantitative data sets are combined: The capacity declaration parameters (for examples see Appendix 1), flight data, slot data and on-ground data. Over 1.5-year data analysis both a seasonal pattern as a growing trend in arrivals and departures can be identified.
Then we take the most recent and busiest (the most arrivals and departures per day) period of twenty full weeks: the summer of 2016. Because it has been the busiest period of AAS ever, it is expected to be representative for the problem and, besides, because all of the different data sets from this period are available it has been chosen as point of analysis. It consists of 20 full weeks and runs from Monday April 4th until Sunday August 21st. Because of the timing of this study later data was not available yet. Over this 20-week period a day-by-day movement pattern is found. There is both a cyclical trend in weekdays (Mondays and Fridays are peak-days, Saturdays are off-days), as there is a growing trend during the summer season, as seen in figures 4 and 5.
Figure 4: Actual realized daily movements (sum of arrivals and departures) from January 1st 2015 until August 21st 2016
Figure 5: Actual realized daily movements (sum of arrivals and departures) from April 4th until August 21st 2016 (point of analysis)
Min. (Saturday 9th April) Q1 Median Q2 Max. (Monday 25th July)
1160 1377,5 1422 1468 1518
Table 4: Daily movement distribution from April 4th until August 21st 2016 (point of analysis)
We analyse the amount movements and aircrafts in position (AIP) per time slot (20 min. interval), since this is the smallest interval for declared capacity (20 min.), slot coordination (20 min.), scheduling (5 min.) and operations (1 sec.). Individual timing and delays are not considered in this study. From here on (figure 6) thick lines are used to represent median values within a data set. Thin lines of the same
colour represent the first and third quartiles to demonstrate variability in the daily movements and gate utilization.
On a typical day arrivals and departures have a repeating arrival and departure pattern with subsequent arrival and departure peaks, as seen in figure 6. This peak is caused by the hub structure of AAS and the wave pattern the biggest carrier likes to operate transfer passengers in. As exemplified by the following quote from the interviews: “There still is capacity available in the quiet periods, but basically at the most attractive times there is hardly any capacity left” (see Appendix 4).
Figure 6: Actual arrival and departure movements: Median and first and third quartile per 20-minute slot.
4.1.2.
Aircrafts in Position
After the descriptive analysis of the movements we focus on the problem area itself: the gate utilization. Patterns in gate utilization are shown in figure 7. Distinctions are made between pier gates, which are connected to the terminal directly by an air bridge, and buffer gates. Buffer gates are either used as a spare gate for handling (then passengers will be transferred to the platform by bus) or to store a plane which is not handling passengers, the so-called towing (movements on land from pier gate to buffer gate or vice versa) movements.
Figure 7: The median, first and third quartiles of AIP (total) and at the different positions (pier gate, buffer gate and cargo ramps) of the data of the point of analysis.
As we can see, and as highlighted, during the morning both outflow of pier gates as inflow of buffer gates are relatively big, gates are made available for arriving planes. Then, both pier gates and buffer gates are utilized to their all-day-maximum. Thereafter utilization peaks follow the arrival peak patterns we have seen in figure 6. A downward trend in utilization is visible during the day until 18.40. Then the gates are utilized more. Cargo ramps are utilized rather evenly during the day.
0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Mo ve m en ts p er slot (20 m in ) Arrivals Departures 0 20 40 60 80 100 120 140 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 AIP
Of the 96 pier gates the median utilization is 57% (55 gates) during the analysed period. In the morning peak (9:20) the median utilization of pier gates is 82% (79 gates). In Appendix 6 we see that during peaks specific capacity is congested totally. This means that aircrafts with specific requirements coming in then, will have to be handled remotely.
The absolute maximum was on Saturday July 9th at 9:20 when 90% (87) of the pier gates were utilized. The median pier gate utilization is 63% of the practical maximum (55/ 87). If this number is increased, more aircrafts on pier gates, and therefore more passengers, could be handled. During the interviews (Appendix 4) we found out that it is the goal of Amsterdam Aircraft Schiphol to cover the morning demand during peak days the best as possible. After the morning peak four subsequent peaks follow. During the 16.40 peak the pier gate utilization is decreased but the buffer gate utilization is increased (as highlighted in Figure 7). This primarily has to do with carrier preferences for small aircrafts being handled remotely. However, also this preference is about to change to more connected handling (Appendix 4).
For a more elaborated analysis of the different piers and their utilization see Appendix 5. For the specific gate capacity in terms of aircraft size and security areas see Appendix 6.
4.1.3.
Summary
4.2. In-Depth Analysis: Arrivals and Departures
Gates have input, aircraft arrivals, and output, aircraft departures. The gate utilization is affected by those input and output numbers and the movement scheduling. When looking at Kösters (2007)’s framework (figure 3) we see different planning horizons of aircraft movements, from the airport’s capacity declaration to the actual movements. In this study the horizons are analysed as follows: first we look at the actual operations, which are compared to the carrier schedule. Then we look at the carrier schedules, which are compared to the coordinated slots. Thereafter we look at the coordinated slots, which are compared to the capacity declaration; and finally we compare the capacity declaration to the actual operations. Each step has its deviations, which in the end add up to more variability and deviation in operations.
4.2.1.
Aircrafts in Position
We seek for a way of combining the arrivals and departures with the gate utilizations. Combining the throughput with the gate utilization we see the following (figure 8). A throughput diagram (Nyhuis & Wiendahl 2009, p.25) shows the cumulative input and output of a system over time, the aircrafts in position (AIP) and the average turnaround time (Tt). In the airport environment we can use this to conceptualize the gate utilization. Below the figure we study it more in-depth.
Figure 8: Combined figure of median (median of the data of the point of analysis) AIP and median throughput by cumulative arrivals (input) and departures (output)
Because of the returning pattern of movements, the medians of the arrivals, departures and AIP levels per 20-minute interval in the whole data set have been used in the throughput diagram. On the left vertical axis the AIP is shown. It is divided between pier gates and buffer gates (black and purple). On the right vertical axis the cumulative movements during the day are shown. They are split between arrivals and departures (blue and orange). The vertical distance between arrivals and departures represents the amount of aircrafts in position (both pier gate and buffer gate) at that time, seen at the left vertical axis. The horizontal distance between arrivals and departures displays the average turnaround time for all aircrafts in position at that time, seen on the horizontal axis. Both average turnaround time and AIP, according to the throughput diagram, vary during the day as shown by measuring turnaround time and AIP for 7.20 and 17.20. The AIP variation is also seen in the AIP differentiation between buffer gates and pier gates. The cumulative movements result in the AIP and the turnaround time levels, which do not consider towing movements and long-term stays (more than
07:20, AIP: 65 Tt: 2.00 17:20, AIP: 47 Tt: 1.40 0 100 200 300 400 500 600 700 800 900 0 10 20 30 40 50 60 70 80 90 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Cumu lat iv e m o ve m en ts AIP
to analyse these in more detail. In the following sub-sections we study the occurrence of the throughput, by analysing the arrivals and departures and their planning horizons, more in detail.
4.2.2.
Actual Movements versus Scheduled Movements
First we compare operation movements, as seen before in figure 6. For the patterns and deviations from scheduled movements see figures 9, 10 and 11. Scheduled movements come from the carrier’s schedule (see Appendix 4), as explained by Kösters (2007) (see ‘airline schedule’ in figure 3).
Figure 9: Median and first and third quartiles of actual arrivals compared to the median of scheduled arrivals
Figure 10: Median and first and third quartiles of actual departures compared to the median of scheduled departures
Figure 11: Median, first and third quartile differences in scheduled and actual movements
In figure 11 we see the deviation from schedule. Every movement (both arrivals and departures) above the green line is one movement more than scheduled during that twenty-minute slot, every movement below is one movement less than scheduled. This is seen on the vertical axis.
As we can see the operation movements follow the scheduled pattern roughly, but also are significantly off. During the night the movements are following schedules better. This probably has to do with night regime monitoring and control (see interviews, Appendix 4). Departure peaks seem to happen after the scheduled times whereas arrival peaks happen before schedule. This is especially
0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Arriv als p er slot (20 m in ) Scheduled Actual 0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 De p ar tu re s p er slot (20 m in ) Scheduled Actual -20 -15 -10 -5 0 5 10 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 De viat ion p er slot (20 m in )
visible in the arrival peak at 15.20 and the departure peak at 21.00 (as highlighted in figure 11). Also the morning peak seems to be significantly off. The standard deviation of the arrivals per 20-minute slot, as seen in figure 9, is 2.5, of departures, as seen in figure 10, it is 2.3. The deviation from schedule is not easy to change and comes forth of a number of different external factors such as weather conditions, aircraft status and passenger disturbances. But, as seen during the night, monitoring and control seems to help increase punctuality. Now we know how actual movements relate to airline schedules, that deviations occur but that the pattern is followed.
4.2.3.
Scheduled Movements versus Coordinated Movements
A day before operating a definite flight schedule has been made by the carriers themselves. Even though the schedules are based on the slot coordination, deviations might happen. See the patterns and deviations between schedule and coordination, in figures 12, 13 and 14.
Figure 12: Median and first and third quartiles of scheduled arrivals compared to the median of coordinated arrivals
Figure 13: Median and first and third quartiles of scheduled departures compared to the median of coordinated departures
Figure 14: Median, first and third quartile differences in coordinated and scheduled movements
Figure 14 uses the same way of showing deviation as figure 11. As we can see the carrier flight
0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Arriv als p er slot (20 m in ) Coordinated Scheduled 0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 De p ar tu re s p er slot (20 m in ) Coordinated Scheduled -20 -15 -10 -5 0 5 10 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 De viat ion s p er slot (20 m in )
scheduled times. At first sight the scheduled movements follow the coordinated patterns. Although at sometimes they are off. Especially the arrivals at 13.00 are scheduled differently than coordinated. The standard deviation of both arrivals and departures is 1.5. So variability is low. Though single scheduled flights might deviate from slot, in total the number of scheduled movements per slot deviates less significantly(although 50% of individual coordinated movements deviates, with an absolute average of “just” 1.4 movements per deviating slot, the aggregation of all scheduled flights shows almost no deviation). Besides the slot coordinator allocates more slots than the carriers use to create slack for emergency situations (2.5%) (Appendix 4). Now we know that the schedule follows coordination agreeably. If schedule is not deviating from coordination that much, the causes of peak congestion must lie in the capacity declaration. In the next sub-section we will analyse this.
4.2.4.
Coordinated Movements versus Declared Capacity
Slot coordination is done according to the capacity declaration. The starting point of this coordination is that slot movements may not exceed declared capacity at given times. The AAS parameter is visually shown in movements per slot, which is elaborated on in figure 18. Carriers request slots and the coordinator grants or declines this request. In the graphs, figure 15, 16 and 17, the coordinated movements are shown and compared to declared capacity.
Figure 15: Median and first and third quartiles of coordinated arrivals and declared capacity
Figure 16: Median and first and third quartiles of coordinated departures and declared capacity
0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Arriv als p er slot (20 m in )
Coordinated Declared Capacity
0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 De p ar tu re s p er slot (20 m in )
Figure 17: Median, first and third quartile differences between declared capacity and coordinated movements.
At first sight there are no median movements above declared capacity. The coordinated movements follow the declared capacity patterns but capacity is still available during off-peaks (as highlighted in figure 15 and 16). Another observation could be made in terms of peak timing: arrivals tend to be coordinated late during a peak and departures early. This also has to do with carriers’ preferences for certain slots (see Appendix 4). The standard deviation of coordinated arrivals per 20-minute slot, as seen in figure 15, is 1.3 (remember: standard deviation of actual: 2.5), of departures, as seen in figure 16, it is 1.4 (remember: standard deviation of actual 2.3) which is significantly lower than standard deviations of actual operations. Therefore actual operations are more capricious than schedule. Variability is partially caused by the different daily movements and repeating weekly patterns (standard deviation per weekday are below 0.9).
Now we have seen how coordination is done, how it follows the capacity declaration and there is room for spreading within the declared capacity, we come to the root of the problem: the capacity declaration, which we analyse more in-depth.
4.2.5.
Consequences of Declared Capacity
We have seen that the capacity declaration forms the basis of movements. In this final sub-section we examine the capacity declaration’s characteristics, what underlying considerations have been applied and how it finally relates to the actual movements and therefore the gate utilization. Generally speaking capacity has been declared in arrivals and departures per time slot. An example which is used at AAS is found in Appendix 1. This statement translates to the pattern shown in figure 18:
Figure 18: the arrival and departure capacity pattern of AAS in summer 2016.
This capacity is based on two capacity parameters: the runway capacity in twenty-minute intervals and the noise-bound capacity in yearly total and yearly night movements (see interviews, Appendix 4). Other capacity constraints are not considered. The capacity declaration displays one parameter: the height of the arrival and departure capacity during a specific twenty minute slot. One could say for daily coordination a single-parameter capacity declaration is used: the runway limits. A quote perfectly
-20 -15 -10 -5 0 5 10 00:00 04:00 08:00 12:00 16:00 20:00 De viat ion p er slot (20 m in )
Arrivals Departures Declared Capacity
0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Cap acity p er slot (20 m in )
that, say 10 a day, I think they [AAS] cannot handle that in operation. That is not in the capacity declaration right now” (Appendix 4). At Rotterdam-The Hague Airport another single-parameter is used which considers aircraft size to service specific gates. Frankfurt Airport uses a multi-parameter for day-by-day coordination which considers runway capacity and North American airway capacity. For all airports also a total seasonal (winter or summer) limit holds because of noise regulations. Table 5 summarizes the distinction between single- and multi-parameter.
Airport Amount of parameters Parameter description
Amsterdam Airport Schiphol 1 (single) Runway capacity per 20-minute slot
Rotterdam-The Hague Airport 1 (single) Gate size capacity per 24 hours
Frankfurt Airport 2 (multi) Runway capacity per 10-minute slot;
North American airway capacity per 10-minute slot
Table 5: Summary of single- and multi-parameter capacity declaration categorization (see Appendix 1)
At AAS a peak pattern in arrivals and departures is found. According to the interviewees this has to do with the carrier demand, because of the hub function and passenger demand, and the history of the capacity declaration. In the past AAS took the flight schedules of the carriers to plan their capacity, this has been used for the first capacity declarations and now is bound by historic rights on slots. Because only a single parameter, runway capacity, is coordinated other constraints are not considered by the slot coordinator. When aircrafts are on ground it has to be handled somehow. In the past this was no problem since enough gates were available (see Appendix 4). Now we see congestion of specific gate types at certain times (see Appendix 7). The bottleneck has moved from the coordinated runway to the uncoordinated gates.
As said, at AAS during summer 2016, there are five arrival peaks with a maximum number of 23 arrivals per slot. Off-peak 12 arrivals per slot may be coordinated. There are six departure peaks with a maximum number of 25 departures per slot. Off-peak 13 departures per minute may be coordinated. Since almost every aircraft that arrives will depart sooner or later, departures are under-utilized more than arrivals. This has to do with more departure capacity than arrival capacity on runways because its safety is better manageable (see Appendix 4).
In theory 2074 daily coordinated movements are possible. In practice we have seen a maximum of about 1500 movements per day (see figure 5), which is 72% of the daily slots. Because of the total seasonal limit on movements, because of noise regulations, flights are spread over the days and less daily capacity may be used. During daily capacity off-peaks there is still room to coordinate more since runway capacity, the single parameter of AAS, is available. We also have seen, in section 4.2, that gate capacity is available then as well. Finally we will once more analyse actual movements. Now we compare them to runway capacity limits. See figures 19, 20 and 21:
0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Arriv als p er slot (20 m in )
Figure 20: Median and first and third quartiles of actual departures compared to the departure capacity limit
Figure 21: Median, first and third quartile differences in capacity limits and actual movements
Looking at the numbers, especially off-peak arrivals are above capacity limits. In total there are 14 median movements above capacity, 9 arrivals and 5 departures. This is caused by delays (either too early or too late) from schedule. Monitoring on delays is only done during the night regime. Increasing monitoring during the day might increase punctuality and decrease getting over capacity limits. However it must be studied if increasing punctuality also will increase better utilization of gate capacity, which will be done in the next section.
We have analysed the effects of the capacity declaration and its basis. We have characterized the capacity declaration of AAS and compared it to other airports’.
4.2.6.
Summary
Going through aircraft position utilization to actual movements, schedules, coordination and the capacity declaration we see that the basis of the operations lie in the capacity declaration. Peaks are declared, coordinated, scheduled and in the end operated. From every step to another deviations happen but the capacity declaration pattern is still followed accurately towards gate utilization. Because of the peak pattern of the capacity declaration of the past, the single parameter that has been used on runway capacity and the grandfather rights that are held, gates are congested now. During the analyses we have found opportunities which need to be explored in order to better utilize airport capacity. These lie in spreading the load more evenly over the single parameter or by going to a multi parameter capacity declaration. Another opportunity lies in increasing punctuality, as done during the night. These opportunities will be further discussed and evaluated in the next section.
0 5 10 15 20 25 30 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 De p ar tu re s p er slot (20 m in )
Actual Declared Capacity
-20 -15 -10 -5 0 5 10 00:00 04:00 08:00 12:00 16:00 20:00 De viat ion p er slot (20 m in )
5. Opportunity Exploration
During the in-depth data analyses we have found opportunities for physical capacity utilization improvement. In this section we explore those opportunities and evaluate them qualitatively and quantitatively. Qualitative evaluation is done by showing the restrictions and the possibilities from expert knowledge by interviews and quantitatively by simulating effectivity. For simulation a testing tool will be needed.
5.1. Opportunity Selection
Qualitative restrictions for this specific case for the different opportunities are given and counteractions are described in table 6. In the sub-sections we discuss them more elaborately.
Category Opportunity Restrictions Counteractions
Ch
an
ge
C
ap
aci
ty
Cri
teri
a
Decrease peak
capacity
Historic slots withhold changes from being really implemented. Carriers might not be willing to return historic slots.
Gradually let carriers return historic slots to the coordinator.
Add another
departure peak
Attempt to reason with carriers to give historic slots back.
Coordination efforts increase.
Add aircraft type
parameter
Govern by local rule to let air carriers give up some slots. Airports need to convince the member state.
Increas
e
Pun
ctu
al
ity
Increase
punctuality
Carriers experience delays out of their own control.
Monitor and punish also during day slots might give air carriers more incentives to operate according to coordination.
Decrease slot bracket size to spread the load more evenly and to monitor more strictly.
Table 6: Opportunity validation by expert and stakeholder interviews (Appendix 4).
5.1.1.
Change Capacity Criteria
From the in-depth analysis we have seen that lots of off-peak capacity is not used. Herein lie opportunities to spread movements more evenly such that peaks are less capricious and possibly better manageable. Right now almost every peak is fully coordinated. When arrivals or departures happen late during such a peak the gates might overflow. By creating more slack (or decreasing peak capacity) during the peaks buffers are created to cope with the variability in actual movements. For current capacity declarations runway capacity and noise limits are kept in mind to determine slot limits. On the other hand gate capacity could be taken more in mind to utilize it better. More short, arrival peaks could spread out the gate utilization peaks. Also since the big morning and evening arrival peaks are not used completely herein are opportunities without messing up the operational system and carrier preferences much.
see interview with the carrier in Appendix 4), reason with carriers to give specific historic slots back (e.g. flights with minimal transfer passengers, also see interview with the carrier in Appendix 4), or let the European member state (in this case the Dutch) government rule locally to force carriers to give specific historic slots back. The latter has been done in Belgium in the year 2009. At Brussels National airport three “silent nights” were implemented and historic rights had been decreased in a phasing out period of five years by a local rule (Belgium Slot Coordination 2009). However, to do this the national government must be convinced to use this political tool which before has only been used for environmental purposes (see interview academic expert in Appendix 4).
Another restriction is the increased efforts of the slot coordinator since their schedule work has more constraints to consider. According to the interview with slot coordination (Appendix 4) it is technically possible, though.
5.1.2.
Increase Punctuality
By giving carriers higher intentions of following slot coordination, operations are expected to be spread more evenly and according to capacity declaration (as seen in in section 4, when looking at the differences between coordination and actual movements, sub-sections 4.2.1 and 4.2.2). Literature has shown different slot coordination policies to increase carrier intentions. Also increased monitoring and punishment (monetary or in terms of losing historic rights) might give carriers higher intentions of increasing punctuality. By decreasing the bracket size in combination with this loads will may be spread more (which already is done in Germany, see Appendix 1.2).
5.2. Opportunity Validation
To test the opportunities we both talk to experts in interviews and we simulate the utilization behaviour in an Excel model. By changing the capacity declaration criteria or by increasing punctuality new patterns and results occur. Going back to the original capacity declaration of summer 2016, see figure 18, we could do a number of things. As proposed in table 6, we evaluate a lower maximum capacity, create more departure peaks (both seen in figure 22), increase punctuality or add another parameter (which will not be evaluated quantitatively because it possibly will change the complete system of coordination).
Figure 22: An example of current and tweaked capacity declaration parameters.
5.2.1.
Design of an Evaluation Model
We now know what we want to evaluate but now need a tool to do this. During the in-depth data analysis and the interviews (Appendix 4) we discovered how gate allocation is a result of flight movements. We use these insights to develop a simple simulation tool.
Model Requirements
The model needs to simulate the effects of the capacity declaration on gate utilization. It needs to consider the capacity declaration parameters, coordinated movements and deviation from
Original Arrivals Original Departures
coordination. Because we have seen that aircrafts sometimes are towed to or from a gate, and this affects gate utilization, this system needs to be incorporated as well.
Mathematically the model must show AIP as follows:
𝐼𝑥(𝑡) = 𝐼𝑥(𝑡 − 1) + 𝑐𝑎(𝑡) ∗ (∆𝑎(𝑡) + 𝑎(𝑡)) − 𝑐𝑑(𝑡) ∗ (∆𝑑(𝑡) + 𝑑(𝑡))
Whereas Ix(t) are the amount of aircrafts at x (either the pier gate or buffer gate position) at time (t); c
is the percentage of aircrafts allocated to (a) or leaving from (d) a pier gate at time (t) to simulate the effects of towing; Δ is the arrival (a) or departure (d) deviation from coordination; a(t) the amount of coordinated arrivals and d(t) the amount of coordinated departures at time (t).
Decision Variables
The decision variables are the declared capacity numbers per slot (a(t) and d(t)). They are either decreased in the first scenario or split over more peaks in the second scenario. Thereafter slot coordination is done manually using a heuristic. Slot coordination is seen as a scheduling task and therefore doing it manually is in line with the hardness of scheduling (Van Wezel et al. 2015). The heuristic used is if coordination exceeds the new capacity because of the new regime, its movements are transferred to an earlier slot until all coordinated slots are within capacity.
Set Variables
The deviations (Δ) between coordinated and scheduled movements and scheduled and realized movements are set per time slot based on the real deviation of the input data. In the third simulation round these deviations will be decreased used as decision variables and the other variables will be set. Also the percentage of aircrafts arriving or departing (c) at a pier gate or buffer gate are set per time slot.
Constraints
The total number of departures and arrivals is constrained. Also the starting values (amount of planes on pier gates and buffer gates) are set. Night movements still are bound and therefore coordination during the night is not changed. Simulation is done in a simplified environment. Aircrafts are not coordinated individually but in aggregation. Specific requirements of aircrafts and carriers are not considered.
Input Data
After the aggregated analyses we have seen patterns in the total data. June 10th is chosen as input day since its AIG occupation during the day is about the same as the third quartile of the data set (the lowest amount of differences in occupation numbers found by optimization), see Appendix 2. The median of utilized pier gates is 59 with a standard deviation of 13. In total 1471 movements have taken place on June 10th. This method is comparable to Bubalo's (2011) method, which uses a random monthly peak date. By using the third quartile in the data a more realistic day is chosen. The coordinated movements of this day are used as input for the scenarios. When they exceed the new declared capacity the described heuristic is used for coordination. The deviations from coordination come from June 10th as well.
The scenarios’ objectives are to increase the average number of aircrafts on pier gates and to decrease standard deviation in pier gate utilization. Two scenarios are simulated: decreasing peak capacity and using more and shorter peaks. Both are tested with the current punctuality and improved punctuality numbers (lower average deviation between scheduled and actual).
5.2.2.
Scenario 1: Decrease Peak Capacity
By decreasing the peak numbers by 10% (sim 1) and 20% sim (2) (arrivals from 23 to 21 and 19; departures from 25 to 23 and 20 and then manually coordinate a(t) and d(t)) the following results are visible, using the same reporting style as been used and explained in figure 8.
Figure 23: Position utilization patterns for current situation, sim 1 (-10%) and sim 2 (-20%)
Pier gate utilization Buffer gate utilization
Average Max. St.Dev. Average Max. St.Dev.
Current 55,1 82 13,2 20,6 36 6,7
Sim 1: -10% 56,2 78 12,0 20,7 37 7,3
Sim 2: -20% 57,2 74 11,4 20,3 38 7,1
Table 7: Performance results for current situation and the two simulations
As we can see in figure 23 the pier gates are utilized less during the peaks and more during off-peaks (as highlighted) . Buffer gates more or less follow the old pattern. As seen in the results, in table 5, the lower the peak capacity the higher the average utilization of the pier gates and the lower the maximum utilization. Also the standard deviation in utilization decreases. So the gate utilization is spread more evenly. 0 100 200 300 400 500 600 700 800 900 0 10 20 30 40 50 60 70 80 90 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Cumu lat iv e m o ve m en ts AIP
5.2.3.
Scenario 2: Adding Departure Peaks
By adding departures peaks (sim 3: add one more departure peak during the morning arrivals; sim 4: also add one more departure peak before the evening peak and then manually coordinate a(t) and d(t)) the following results are visible:
Figure 24: Position utilization patterns for current situation, sim 3 (1 more peak) and sim 3 (2 more peaks)
Pier Gate utilization Buffer Gate utilization
Average Max. St.Dev. Average Max. St.Dev.
Current 55,1 82 13,2 20,6 36 6,7
Sim 3 56,7 76 11,9 18,6 36 6,6
Sim 4 56,5 76 11,0 16,2 36 7,6
Table 8: Performance results for current situation and the two simulations
As we can see in figure 24 the pier gates are utilized less during the peaks and more during off-peaks (as highlighted). Buffer gates are used less. As seen in the results, in table 6, the more departure peaks the higher the average utilization of the pier gates and the lower the maximum utilization. Also the standard deviation in utilization decreases. So the gate utilization is spread more evenly. Adding a second extra departure peak, in the afternoon, seems not to have extra effects on the pier gate utilization. 0 100 200 300 400 500 600 700 800 900 0 10 20 30 40 50 60 70 80 90 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Cumu lat iv e m o ve m en ts AIP
5.2.4.
Scenario 3: Increasing Punctuality
By decreasing deviation from coordination and scheduling and scheduling and realisation (both by multiplying Δ by 0,5 (sim 5) and by 0 (sim 6) realisation operates more punctual according to coordination) the following results are visible:
Figure 25: Position utilization patterns for current situation, sim 5 (50% less deviation) and sim 3 (no deviation)
Pier Gate utilization Buffer Gate utilization
Average Max. St.Dev. Average Max. St.Dev.
Current 55,1 82 13,2 20,6 36 6,7
Sim 5 53,8 80 11,5 20,4 36 7,1
Sim 6 49,1 79 14,0 16,9 32 6,6
Table 9: Performance results for current situation and the two simulations
As we can see in figure 25 the pier gates are not utilized significantly less during the morning peak than currently. So the same gate capacity is still needed. However the peak is shorter and empties out earlier, as we can see in the average utilization, in table 6, as well. The other peaks are spread out more evenly. Also buffer gates are utilized less. Standard deviation of utilization varies. Operating as coordinated does not spread out the peaks and does not decrease maximal utilization.
0 100 200 300 400 500 600 700 800 900 0 10 20 30 40 50 60 70 80 90 00:00 02:00 04:00 06:00 08:00 10:00 12:00 14:00 16:00 18:00 20:00 22:00 Cumu lat iv e m o ve m en ts AIP
