ABCD: Aircraft Based Concept Developments. Work Package n 2

Transcription

1 ABCD: Aircraft Based Concept Work Package n 2 This document presents a synthesis of information aiming to support discussions concerning ABCD concept and processes. It does not represent the position of Agency. 1

2 Page: 2 DOCUMENT IDENTIFICATION SHEET DOCUMENT DESCRIPTION Document Title ABCD: Aircraft Based Concept Work Package n 2 Abstract This document describes the benefits of Aircraft-Based traffic concept and processes. Keywords CFMU Airport ATC ATFM Capacity Airlines Coordination CTOT Delay EOBT FPL Messages Slot Assignment Taxi time ANSP Turn around CONTACT PERSON: TEL: DIVISION: DOCUMENT STATUS AND TYPE STATUS CATEGORY CLASSIFICATION Working Draft Executive Task General Public Draft Specialist Task EATMP Proposed Issue Lower Layer Task Restricted Released Issue INTERNAL REFERENCE NAME: ELECTRONIC BACKUP HOST SYSTEM MEDIA SOFTWARE Microsoft Windows Type: Media Identification: 2

3 Page: 3 DOCUMENT CHANGE RECORD The following table records the complete history of the successive editions of the present document. EDITION DATE DESCRIPTION OF EVOLUTION SECTIONS / PAGES AFFECTED /10/2007 Working Draft Creation All 3

4 Page: 4 TABLE OF CONTENTS 1 GENERAL FRAMEWORK INTRODUCTION Context and background ABCD purpose PURPOSE OF THE DOCUMENT STRUCTURE OF THE DOCUMENT WORK-PACKAGE 1 MAIN RESULTS ROLES AND RESPONSIBILITY AIRLINE S INTERVIEWS CONCLUSIONS ABCD IMPACT ON DELAY REDUCTION PROCEDURE NON - WEATHER REGULATION ANALYSIS Results Geographical impact Airlines type impact FPL Message anticipation impact on ATFM delay Conclusions WEATHER REGULATION ANALYSIS Results Conclusions SCHEDULE ADHERENCE AT DEPARTURE Results Conclusions DELAY FROM PREVIOUS FLIGHT Results Conclusions CONCLUSIONS AND NEXT STEPS CONCLUSIONS NEXT STEPS TABLE OF FIGURES DICTIONARY OF ABBREVIATIONS ANNEX 1 - ATFCM TECHNICAL OVERVIEW ANNEX 2 ANALYSIS SCOPE JUSTIFICATION AND DATABASE USED

5 Page: 5 1 GENERAL FRAMEWORK 1.1 Introduction Context and background Nowadays, air transport stakeholders (airlines, ANSPs, CFMU, airports) have established several processes, which aim at maximizing the use of available capacity while ensuring safety and a fair, transparent and non-discriminatory use of existing facilities. The two main processes are airport and ATFM slots allocations, since they deal with the main bottlenecks of the system: airports and airspace. Each of these two processes has its own logic and sense: airport slot allocation process ensures the balancing of the airline s strategic demand regarding the airport s capacity, while the ATFM slot allocation process introduces the operational flexibility required in order to react to more tactical perturbations. These processes are complementary and take place at different chronological phases: the airport slot allocation process is ensured during the strategic phase, several months before the day of operations. The ATFM slot allocation process is operated during the day of operations, a few hours before real execution of flights. The FPL (Flight Plan) management links the processes; through FPL aircraft operators transform allocated airport slots into EOBT (Estimated Off-Block Times) and collect all the relevant information about planned and actual flights. For flights within the European airspace, aircraft operators send to the CFMU a message (FPL) containing basic information about the flight in order to obtain clearance to over-flight, take off and/or landing at European airports. The aggregation of all FPLs, sent by airlines and handled by the CFMU, constitutes a FPL database. With this definition, the FPL database provides a picture of the overall aircraft traffic, which is flying and going to fly over Europe in the future. This database is regularly updated upon the reception of messages, sent essentially by airlines to CFMU through the use of IFPS and through CDM (Collaborative Decision Making) at some major European airports and of course through the real aircraft position provided by radar. The FPL database is composed of individual flight plans, which are usually not linked to each other. In the same time, it has been determined that when a delay appears on a given flight, part of this delay propagates for the flight using the same aircraft. In one of his previous study ADV systems has evaluated the impact of an ATFM delay on a daily itinerary on the Air Fance fleet by measuring the knock-on effect which could be defined as the cumulative delay due to the propagation of the ATFM delay throughout the itinerary. The overall results are presented in Figure 1 below in which the knock on effects curves are represented for the occurrence of an ATFM slot at a given station and for sundry magnitudes of the ATFM delay. 5

6 Page: m in. 20 m in. 30 m in. 4 0 m in. 50 m in. Cumulated knock-on effects (min.) Station 0 Sta tion 1 Statio n 2 Station 3 Station 4 Station 5 Figure 1: Cumulated knock-on effects due to a single slot In the figure above, the knock-on effects are shown for different legs of an aircraft s itinerary: station 0 corresponds to the first leg of the itinerary, i.e. to the first airport of departure; station 1 corresponds to the second airport of departure (first airport of arrival) and so on. The overall result shows that knock-on effects are not a constant proportion of the initial ATFM delay. For instance, a 20-minute ATFM delay allocated at station 0 generates approximately 15 minutes in knock-on effects whereas a 50-minute ATFM delay at the same station produces nearly 60 minutes in knock-on effects. In the same way, a 50-minute ATFM delay occurrence at station 2 will generate less than 20 minutes in knock-on effects. Conversely, a slot allocated at the first, fourth or fifth station 1 generates maximum knock-on effects. Nowadays, because the FPL database is composed of individual flight plan not linked to each other, the CFMU could not foresee precisely (more than on stage ahead) what could be the impact of a delay on a given flight on future flights operated with the same aircraft than the one at which the delay occurred ABCD purpose The aim of the ABCD (Aircraft-Based Concept Development) concept consists in using the aircraft registration, which is already a parameter of the FPL message, in order to link the individual flight plans of the CFMU FPL database. The WP1 of the ABCD concept study has consisted in describing the ABCD concept, in analysing the benefits it could provide to air transport stakeholders through interviews with airlines : Airport Operator will be able to improve service provision to their customers through better allocation resources as they could be informed long in advance of 1 Respectively number 0, 3 and 4 on the figure 6

7 Page: 7 the future potential disruption in term of flight plan schedule consistency. These apply to both the tactical level, through stability in gate and stand allocation, as well as strategically, as better use of infrastructures supports greater passenger throughput. Airports should also be able to provide better information to their customers. Overall, the airport quality of service should be enhanced; As ABCD should help to control reactionary delays, improving the predictability of aircraft operations, the over-riding factor in controlling the financial costs of delays, aircraft operator will be able to improve their punctuality, leading to reduced tactical and strategic costs of delays ; Ground Handler will be able to improve the use of their resources, saving costs and providing an improved level of service; ATC / CFMU will benefit at both local and network level. ABCD will provide a better picture of future traffic flows, thus improving its management. It could reduce reactionary delays through better slot compliance. In the long term Air Traffic Control Centre will minimize the waste of their effective capacity as a consequence of the increased confidence in the predictability of traffic flows. Low-cost and regional airlines consider that ABCD would facilitate their delay management and optimize their slot allocation process and thus they stated their interests in the ABCD concept implementation. Moreover, thanks to the linkage of individual flight plans, it has been inferred that it could be possible to provide more accurate predictions of the downstream legs of an aircraft itinerary, in particular when a flight suffers from disruptions (involving delays). It has been shown through the analyses conducted in the WP1, that thank to the linkage of the individual flight plans of the CFMU database, the ABCD concept could improve the predictability and the efficiency of the airlines, ATM and airports operations. In particular, in the case of one regional airline, it has been proved through some real and tangible examples, that the sooner the delay anticipation for a given flight, the smaller its ATFM delay. One the objectives of the ABCD concept WP2 will focus on substantiating this point over a large scale basis and on showing to low-cost and regional airlines the interest of using ABCD as a tool to minimize ATFM delay, establishing to major airlines that implementing ABCD concept would also benefit them as a result of global network optimization, proving to ATM stakeholders that the use of ABCD enhance the predictability of the ATM network and hence reduces its uncertainty. 1.2 Purpose of the document The present document constitutes the first part of the project and intends to: Summarise the main results presented at the end of work-package 1: Aircraft Based Concept Consolidation; 7

8 Page: 8 Consolidate opinions given alt the end of work-package 1 by airlines concerning ABCD concept; Introduce the concept of ABCD, as a way to link the FPLs within the existing FPLs databases; Present sets of analysis showing the interest of ABCD concept; Proposes conclusions and way forward to implement ABCD concept. As the development of the ABCD concept requires support among involved actors, including airlines, CFMU and airports, a number of interviews have been conducted. The interviews were aimed at getting the feedbacks about the ABCD concept. 1.3 Structure of the document The document is split in 6 sections and 2 annexes: Section 1 Introduction presents the context of the project and the purpose of the present document; Section 2 Work-Package 1 Main Results presents main results from Work- Package 1 including the description of the role and responsibility of each stakeholder and the conclusions from airlines interview held at the end of WP1; Section 3 ABCD impact on Delay Reduction presents analysis showing expected improvements from ABCD strategy implementation. The impact of different factors such as geographical zone, size of airlines and type of regulations has been assessed; Section 4 Conclusions and Next Steps presents the conclusions from the analysis and proposed next steps towards ABCD implementation; Annex 1 ATFCM technical overview overview of necessary slot allocation process and Computer Slot Allocation Process (CASA) concepts in order to understand analysis presented in the document; Annex 2 Analysis scope justification and files/database used presents analysis scope justification and describes input data needed for the analysis. It also describes the way regulation slot lists were worked in order to select relevant data needed for the analysis. 8

9 Page: 9 2 WORK-PACKAGE 1 MAIN RESULTS Aircraft-Based Concept (ABCD) explores the impact of reorganising the flight plan management into a re-named aircraft-based management. ABCD is primarily based on the use of aircraft parameters, in order to establish a link between the individual flight plans of the CFMU database. This aspect intends to constitute a true evolutionary step forward in the improvement of predictability, and efficiency of the ATM operations. Thanks to the linkage of individual flight plans, it should be possible to provide more accurate predictions of the downstream legs of an aircraft itinerary, in particular when a flight suffers from perturbations (e.g. delays). Such accurate predictions are expected to lead to an optimised ATFM slot allocation, by e.g. allocating more efficiently the ATFM slots on an early up-to-date knowledge of the flight progress. Punctuality is the end product of a complex interrelated chain of operational and strategic processes carried out by different stakeholders during different time phases and at different levels (local/network) up to the day of operations. Punctuality is affected by the lack of predictability of operations in the scheduling phases and by the variability of operational performance on the actual day. Part of the unpredictability of actual situations with respect to scheduled ones, derives from the lack of information about the status of the flight during the different phases of the day. Improving punctuality means identifying sources of variability that can be reduced, reducing the variability of flight phases below given values, and being able to predict given values during the scheduling phases. As the aim of the ABCD (Aircraft-Based Concept Development) consists in using the aircraft registration, which is already a parameter of the FPL message, the ABCD concept will allow to link separated flight segments using a same aircraft. Thank to this linkage, it will probably be possible to update in a more efficient way the predictions and thus to provide a better picture of the future traffic flow over the European sky. 9

10 Page: Roles and responsibility The following table resumes roles and responsibilities presented in work-package 1 for each actor concerned by ABCD implementation. CFMU The CFMU is the central unit that collects and broadcasts flight data information. Airlines The airlines are responsible for the provision of their flight intentions to CFMU, in the form of RPLs, strategically, and/or in the form of FPLs & flight plan updates, at a pre-tactical level. Airport ground handlers Airlines agents or ground handlers are the representatives of airlines at the airport of operations. Each agent (or handler to which the task is subcontracted) covers turn around activities. ANSPs The ANSP is responsible for providing actual (real-time) position reports to the CFMU. In its role of central processor of the information, it is well positioned to make the linkage between various elements describing the course of an aircraft s itinerary i.e. between different flight plans. If a modification to an initial flight plan occurs, the CFMU should be able, when it has the necessary information (aircraft registration and turn around duration), to update the downstream flight plans. Considering that every airline has to define an aircraft allocation schedule in advance (generally they finalise the schedule during the day before the day of operations), the schedule could be made available to CFMU pre-tactically. The aircraft allocation notification should remain flexible enough in case of ad-hoc events, where the airline can decide to modify the aircraft allocated to the flights. In such cases, they transmit the update to CFMU, via a CHG message. The airlines should transmit TTMs (minimum duration time required for turn around operations at an airport and for a given aircraft type) The airlines agent at the airports monitors the progress of the turn around process. When the aircraft has taken-off, airlines headquarters are informed on the actual take-off time of the aircraft, via dedicated messages. Some of these messages can be sent by airlines agent (e.g. case of a MVT message). Others are sent by ANSPs (e.g. case of a DEP message). If a deviation is observed, the CFMU should calculate the impact on the entire downstream aircraft profile (i.e. linked flight plans). Table 1: ABCD roles and responsibilities overview 10

11 Page: Airline s Interviews Conclusions The development of the ABCD concept requires support among involved actors, including airlines, CFMU and airports. The objective of this section is to present the results of the interviews with the airline s Operational Control Centers (CCOs). The purposes of the interviews were: to obtain information on the organizational and operational framework of each airline, to present the ABCD project to each airline in order to get their feedback on ABCD concept. The interviews were held at the end of WP1 of the ABCD project, which consisted in establishing a state of the art of the current operations regarding flight planning. This section resumes the main conclusions of the interviews which were used as starting point for defining WP2 content and expected results. Recognising that different airlines have different requirements and expectations, the following categories of airlines were defined: 1. flag-carriers or major airlines, 2. low-cost airlines, 3. regional airlines, general aviation or other non-scheduled airline types. For each of the categories, an airline has been selected for an interview, for an overall coverage of the requirements and expectations regarding ABCD project. Air France Airlinair Ryanair Major Airline Low-cost airline Regional airline X On air X X Iberia X Table 2: List of airlines The following table aims at resuming opinions given by interviewed airlines about key ABCD concept topics. It is acknowledged that not all airlines bellowing to each of the groups would subscribe all the opinions resumed in the table. Nevertheless, the table is aimed at resuming the common opinions and strategies in most of the airlines belonging to each group. X X 11

12 Page: 12 A/C assignation to each flight Turn-around times communication Turn-around Times (TTM) strategy A/C reassigned in case of operational delays? DLA communication to CFMU strategies Interest in ABCD concept Major Airline Low-cost airline & Regional Airline YES. All interviewed airlines assign individual aircrafts to the flights at least the day before operations. However, this schedule can change mainly due to maintenance issues. YES. All interviewed airlines would be capable to send turnaround duration to CFMU. Usually have margin in some of the flight legs of the day in order to be able to recover from delays occurring during the day. YES. Use of Hubs allows major airlines to change A/C during the day to recover delays. Major airlines have their own tools (ABCD-concept tools & decision-making tools) and policies helping them to decide the optimum moment to communicate a delay to the CFMU. Low interest in the concept. Major airlines consider they have their own tools to handle delays and fear loosing some control in case the concept was deployed. They want to keep full control to manage delays. Tight schedules that try to keep turn-around times to the minimum (30 min.). In case of delay in one of the legs it is very difficult to recover. NO. These airlines can rarely change aircraft assignations during the day in order to recover delays due to point-topoint (as opposed to hub) strategy and lack of recovery strategies. Different strategies depending of the airline: some communicate delays at the very last moment (EOBT-10 ), trying to recover the delay until the last minute in order to avoid passing at the queue of slot allocation list, others communicate delays at the moment they appear without any treatment in their side. Strong interest in the concept as it would facilitate delay management and optimize slot allocation process for them. Moreover, small airlines are interested in automatic CTOT recalculation once delay can not be recovered as it would simplify their task and optimize their resources. Table 3: Resume of interviewed airline s opinions on ABCD concept 12

13 Page: 13 Some valuable conclusions about ABCD concept can be summarized from interviews to airlines. In order to provide input to ABCD concept, airlines would be able to send schedules, assigning individual aircrafts to flights, as well as information about turn around times at airports, at least one day before the day of operations. However, changes in aircraft assignation are quite common for some airlines. Therefore, the technical implementation of the ABCD concept should take into account that due to operational issues (maintenance,..) airlines need to keep the flexibility of swapping aircraft without any difficulties. It should be also taken into account when defining implementation strategy of ABCD concept that some airlines (mainly major airlines) want to keep control on the moment they decide to communicate delays to CFMU. Major airlines consider that they have their own tools, similar to ABCD, to handle delays and they do not want to use two different tools in case of ABCD implementation. Low-cost and regional airlines state that ABCD would be a very useful tool to facilitate their delay management and optimize their slot allocation process. In this way, they are very interested in the implementation of this concept. In order to consolidate the benefits, which could be brought to airlines, by the implementation of ABCD, the content of the work-package 2 will focus on substantiating on a large scale the following issues: To low-cost and regional airlines: ABCD will be of interest as it allows a better anticipation of delay communication to CFMU, which minimizes ATFM delay. To major and big airlines: the implementation of ABCD concept would also benefit them as a result of global network optimization. 13

14 Page: 14 3 ABCD IMPACT ON DELAY REDUCTION ATFM delay measures the difference between an estimated time over a regulated area predicted by airlines (ETO) and calculated time over a regulated area assigned during slot allocation process (CTO). Thus, the difference between ETO and CTO can be considered as the difference between the time the airline wants the flight to go through a regulated area and the time assigned to the flight. Respecting scheduled flight times is a key factor for passenger satisfaction, and thus keeping ATFM delay as low as possible is very important for airlines willing to maximize customers satisfaction. Hence, the common objective for airlines and CFMU is keeping ATFM delay as close to zero as possible. However, CFMU objective is to optimize global ATFM delay assigned to all airlines whereas each individual airline has its own strategy to minimize its own ATFM delay. CFMU has procedures and systems maximizing global satisfaction (resumed in annex 1) whereas each airlines deal with different strategies to minimize its delays and their consequences. The following analysis aims at showing that the anticipation of delay communication (DLA message) to CFMU reduces the ATFM delay which in return, knowing that the implementation of ABCD concept increases anticipation in delay detection and communications, will prove that the implementation of ABCD concept will decrease ATFM delay. In order to be as comprehensive as possible and avoid biases in studying the relation between the delay anticipation and its associated ATFM delays, the analysis will take into account the following parameters: Non-weather and weather regulations. Link between ATFM delay and DLA Anticipation in considering the geographic zone and airline nature. Link between ATFM delays and delay anticipation with flights that have not send DLA message. Link between FPL Message anticipation and ATFM delay. Therefore, a possible relationship between ATFM delay imposed by CFMU to airlines and the respect of the take-off slot (CTOT) has also been analyzed. Finally, the last section in this chapter shows that ABCD could improve current delay anticipation communication (DLA messages). 14

15 Page: Procedure The shorter the anticipation in sending a DLA message, the higher the risk of being constrained with a high ATFM delay. This assumption relies on the fact that the earlier the CASA algorithm is aware of the need to find a new slot for a given flight, the more chances it has to find a slot that fits airline s request. As explained in Annex 1, a message DLA with information about the time of the new EOBT (Estimate Off Block Time) is sent to CFMU by airlines when they are aware of a delay greater of 15 minutes with the new EOBT 1. DLA messages containing EOBT are registered in CFMU system. The system also records the exact time the delay message was sent. The difference between the time the DLA message was sent and the EOBT is the delay anticipation 2. Delay anticipation concept illustrates how much in advance the airline communicated a delay to the CFMU. An example illustrated by figure 2 can be useful to illustrate the concept. Flight RYR464T from September 17 th, 2006 had a delay. The airline knew that it would not be possible to recover this delay for next flight RYR465T which had a scheduled EOBT at 10:45. The airline sent a DLA message for flight RYR465T at 09:58 submitting an EOBT at 11:40. Knowing that this was the last DLA message sent for RYR465T flight then the delay anticipation for this flight was 102 minutes. Airline DLA message CFMU RYR EOBT ETOT RYR465T 9:58 11:40 11:50 Time Delay anticipation: 102 minutes Figure 2: Delay Anticipation Concept 1 New EOBT Initial EOBT > 15 minutes 2 Delay anticipation = time of last DLA - EOBT 15

16 Page: 16 Once the delay anticipation had been calculated for each single flight, flights were grouped depending on this anticipation in 20 minutes windows as shown in figure 2: flights that sent DLA message between EOBT and EOBT-20 belong to group 1, flights that sent DLA message between EOBT-20 and EOBT-40 belong to group 2, up to flights that sent DLA message between 140 and 160 minutes before EOBT. RYR 465T Group 1 Group 2 Group 6 Delay anticipation Figure 3: Repartition of flights in groups according to anticipation Delay anticipation analysis is limited to 160 minutes before EOBT because CASA algorithm starts giving priorities to flights based on FIFO policy at EOBT-180 (See CASA algorithm quick description at Annex 1). Hence, a DLA message sent before EOBT-160 is considered almost like a non delayed flight. As explained in Annex 2, ATFM delay assigned to each flight is known from regulation reports. The last step in the analysis consists in computing the average ATFM delay for each flight s group from the individual ATFM delays. The objective of the analysis was to show that ABCD concept could help improving the most penalizing regulations: Smoothing disruptions through ABCD concept. Hence, the analysis focused on severe regulations that impose severe ATFM delay to flights affected. The global sample covers one week (7 days), during September 2006: from Monday 11th of September 2006 until Sunday 17th of September For each day, the top 10 disturbing regulations having caused the highest overall delay were selected; therefore the analysis was performed on 70 regulations. In order to achieve the most comprehensive knowledge of ABCD impacts, input data was classified considering several criteria: Regulation typology Airline typology Geographical zone FPL messages Thus, the analysis was repeated for each data group. The following sections present the most relevant results and conclusions from the analysis. 16

17 Page: Non - Weather Regulation analysis Due to its own nature, weather regulations are difficult to predict and variable as they can evolve when weather conditions change. Under these circumstances anticipation in flight s schedule variations are more difficult to predict. Moreover, expected benefits from delay anticipation are less important as flow rate of the regulation will evolve continuously leading to a lack of predictability even if all flights would be able to anticipate their previous flights delay. The analysis was initially split in 5 groups, one for each regulation type: C ATC capacity G Aerodrome Capacity S ATC staffing T Equipment (ATC) W Weather However, preliminary analysis showed that non-weather regulations had similar results whereas the weather regulations present erratic behaviour when trying to link the delay anticipation and its amount. Thus, the analysis was finally split in two depending on the type of regulation: Non Weather regulations Weather regulations Non-Weather From the 70 regulations chosen for the analysis 28 were weather regulations and 42 were non-weather regulations. The following section presents results from the analysis in all non-weather regulations Results Correlation between ATFM delay and DLA Anticipation The ATFM delay for all flights that sent DLA message with anticipation between 0 and 160 minutes was computed and is shown in the graph. The picture below presents the result of the analysis with: ATFM delay in minutes on Vertical axis. Anticipation in minutes (delay anticipation windows of 20 minutes) on Horizontal axis. 17

18 Page: 18 ATFM delay in minutes Average ATFM delay 18, Anticipation in minutes Figure 4: ATFM delay vs DLA anticipation Non-Weather regulations The following points can be highlighted from Figure 4: ATFM delay significantly decreases when delay anticipation increases. For small delay anticipations comprised between 0 and 40 minutes, ATFM delay decreases from 22 minutes to 16 minutes, meaning a 30% ATFM delay decrease. The trend is globally negative until the window [140; 160[ and the curve seems to stabilize at an ATFM value around 14 minutes. The average ATFM delay imposed to flights than sent a DLA message (delayed flights) was 18 minutes. It should be noted that the number of flights available to compute the average ATFM delays decreases when the anticipation increases. The following figure shows the sample size and distribution split in 20 minutes time windows. 18

19 Page: Number Anticipation in minutes Figure 5: Number of flights in each anticipation range The graph shows, for instance, that in the analysed non-weather regulations there were 170 flights that sent a DLA message between 20 and 40 minutes before EOBT. Because of this distribution, average ATFM delays are statistically less representative for high anticipations than for short anticipations. The concentration of DLA messages between 0 and 20 minutes before EOBT can be explained mainly for 2 reasons: Delays occurred during turn around process in the airport. Certain airlines wait until the last minute to communicate delays to CFMU (as confirmed in airlines interviews and presented in 2.2). The figure below shows the cumulative distribution of occurrences split in 20 minutes anticipation windows. % ,0 52,8 61,4 68,1 74,8 81,0 85,6 89,3 92,1 100, Anticipation in minutes Figure 6: Cumulative distribution of flights according to anticipation 1 1 Anticipation times beyond 180 minutes have been limited to 180 minutes since CASA algorithm accepts receiving FPLs until 3 hours before EOBT, see annex 1 19

20 Page: 20 40% of the flights sent the DLA message less than 20 minutes before EOBT. More than half of the analysed sample (53%) is concentrated in anticipation between 0 and 40 anticipation before EOBT. Nearly 70% of delayed flights declare their delay with an anticipation of less than 80 minutes. Considering all flights that sent a DLA message, the average DLA message anticipation is 58 minutes. Figure 4 demonstrates clearly that the sooner the delay anticipation the smaller the ATFM delay. Figures 5 and 6 show that nearly 70 % of delayed flights declares its delay with an anticipation of less than 80 minutes. This figure is interesting when putting into perspective with an Eurocontrol 1 study, which stated that the average flight length of flights in Europe was 82 minutes in 2005 and would reach 81 minutes in This means that nearly 70% of delayed flights have not anticipated with a time anticipation greater than the average flight time. It has been shown that thank to the linkage of flight plan and the information about the turn around time, it would be possible to anticipate earlier than today expected delay on flights impacted by a delayed initial flight using the same aircraft than the initial one. According to the Figures 4, 5 and 6, this means that ABCD concept will help to have more flights with sooner delay anticipation than today and thus, inferred from Figure 4, that there will be more flights with a fewer ATFM delay than today and thus less ATFM delays than today. Link with flights that have not send DLA message The average ATFM delay imposed in the same set of analysed regulations to flights that did not send a DLA message was computed in order to compare it with the curve of flights that sent a DLA message. 1 (2005) Performance Review Report - An assessment of Air Traffic Management in Europe during the calendar year 2004, PRR8, Final Draft) 20

21 Page: ATFM delay Flights w/out any delay message ATFM delay in minutes ,1 ATFM_Delay Flights with DLA Anticipation in minutes Figure 7: ATFM delay vs DLA anticipation Non-Weather regulations with ATFM delay from flights without DLA The average ATFM delay for flights having not sent any delay message (flights ontime) is 14,1 minutes. The curve lies for high anticipations around the value of ATFM delay for flights that have not send any delay message. As in the previous case, this means that the higher the delay anticipation, the closer their ATFM delay will be from the ATFM delay they would have had if they had not announced any delay. This picture would show that if an airlines is capable to anticipate a delay more than 120 minutes before flight starts, in general it will not be imposed any supplementary ATFM delay. The underlying idea is that anticipating delay long time in advance is like not having any delay from an ATFM delay perspective Geographical impact Regulations are defined in a particular zone, the Area Control Center (ACC) defined by a country and a city. An analysis separating regulations by country was performed for countries that gathered the highest number of regulations (from the same 48 nonweather regulations) in order to check if geography could have an impact in the way ATFM delay was correlated to DLA anticipation. As expected, results showed no relevant differences in the graphs representing the correlation between ATFM delay and delay anticipation for different countries. 21

22 Page: Airlines type impact Airlines have different strategies to minimize delays consequences. This is, in particular, reflected in the way major airlines and low-cost or regional airlines handle delays that cannot be absorbed at airport and are going to propagate into reactionary delays. Major or large airlines, that have spare resources, are capable to swap aircrafts when one flight incurs too much delay. Thus they can absorb their delay using an aircraft available instead of the aircraft from a delayed flight. Low cost or regional airlines can not usually swap aircrafts as their entire fleet is continuously operating leaving little possibility of swapping. Therefore, an analysis differentiating major or large airlines from low cost and regional airlines was performed (using the same non-weather regulations). In order to perform this analysis, airlines, with the highest rate of disrupted flights inside the analysed regulation, were selected and then classified between major and big airlines and lowcost and regional airlines. Major or large airlines considered in the analysis are the following: Air France, Alitalia, British Airways, Deutsche Lufthansa, Iberia, KLM, LOT Polish Airlines, Scandinavian Airlines and Turkish Airlines. Low-cost or regional airlines considered are Air Berlin, Air Madrid, Centralwings, Excel Airways, First Choice Airways, German Wings, LTU Germany, Norwegian Air Shuttle, Ryanair, and Wizzair. The same analysis, than the one presented previously, for all flights affected by nonweather regulations was performed for both groups of airlines separately. Major and large airlines or Flag Carriers The picture below presents the results concerning flag-carriers or major airlines. 22 ATFM Delay in minutes Anticipation in minutes Average ATFM delay 18,4 Figure 8: ATFM delay vs DLA anticipation Non-Weather regulations major and large airlines 22

23 Page: 23 The graph presents the same general trend observed for all data from non-weather regulations: The trend is globally negative until an anticipation of 140 minutes. ATFM delay decreases from 21,6 minutes for delay anticipations lower than 20 minutes to 14,2 minutes for anticipations between 60 and 80 minutes. In the range , the curve seems stable around 14 minutes. ATFM delay continues decreasing again to get a minimum value of 11 minutes for delay anticipation between 120 minutes and 140 minutes which corresponds at the time when airlines receive the slot proposed by CASA. After 140 minutes, the curve goes up to an ATFM value around 16 minutes for an anticipation of 150 minutes. This could be explained by slot allocation process that keeps a provisional slot in the system which can still be change to a better or a worse one between 2 hours and 3 hours before EOBT. However it would not explain the reason why this effect is only visible for major airlines. Another hypothesis would be that the low number of occurrences within minutes range makes the average ATFM delay not very significant. Hence, the most sensible hypothesis seems to be the second one. The average ATFM delay for all flights that sent DLA between 0 and 160 minutes is 18,4 minutes. The size of the sample that is available decreases with anticipation as it can be seen in the following graph presenting the distribution of DLA anticipation for analysed flights Number of occurences Anticipation in minutes Figure 9: Number of flights in each anticipation range major or large airlines This distribution is very similar to the joint distribution for all airlines together. 23

24 Page: 24 The following graph presents the cumulative distribution of DLA anticipation from 0 to 180 minutes. % ,6 58,5 68,4 76,0 83,9 90,6 94,4 98,0 98,8 100, Anticipation in minutes Figure 10: Cumulative distribution of flights according to anticipation major or large airlines 1 The distribution is very similar to the distribution presented considering all airlines which means that major or large airlines behaviour to manage delays stays very close to the behaviour described considering all airlines. Low-Cost and regional airlines The number of flights found for this particular analysis was quite low. Therefore, in this case, delay anticipation windows of 40 minutes were considered in order to gather enough data in each window to compute significant average ATFM delay values. Thus, for instance, the point at 20 minutes represents average ATFM delay value for flights that had anticipation between 0 and 40 minutes. The picture below presents the results for the analysis concerning Low-Cost airlines. 1 Anticipation times beyond 180 minutes have been limited to 180 minutes since CASA algorithm accepts receiving FPLs until 3 hours before, see annex 1 24

25 Page: ATFM delay in minutes Average ATFM value 17, Anticipation in minutes Figure 11: ATFM delay vs DLA anticipation Non-Weather regulations low-cost airlines It can be pointed out from the graph: From the observation of this graph there seem to be a negative tendency on the overall curve, thus ATFM goes from 19,9 minutes to 13,5 minutes. However, there is not a clear decreasing tendency as ATFM delay for 100 minutes window anticipation seems to be higher than ATFM delay for 60 minutes anticipation window. The average ATFM delay for all flights that sent DLA with an anticipation between 0 and 160 minutes is 17,1 minutes. The distribution of anticipation for low cost airlines seems different from the distribution for all airlines or major airlines. The following graphs show the number of occurrences for each anticipation window for Low Cost or regional airlines and the cumulative distribution. 25

26 Page: Number of occurences Anticipation in minutes Figure 12: Number of flights in each anticipation range low-cost airlines Low cost and regional airlines present a different distribution of delay anticipation compared to major airlines or for all airlines. In the studied database, 35% of flights anticipated between 0 and 40 minutes which is inferior to rates observed for all airlines or major airlines. This could mean that low-cost airlines communicate delays earlier than other airlines. This affirmation contradicts announced policy from several low-cost airlines that stated that they waited until the very last moment to communicate any delay. Given the relatively short number of occurrences for low-cost airlines, there is not enough confidence to conclude that this graph represents all low-cost or regional airlines behaviour. However, this demonstrates the other point concluded from interviews in 2.2 dedicated to airlines feedback to ABCD concept explaining that there is no common delay management policy for these airlines. % ,2 34,8 42,7 47,6 54,3 63,4 73,2 82,3 87,8 100, Anticipation in minutes Figure 13: Cumulative distribution of flights according to anticipation low cost airlines 1 1 Anticipation times beyond 180 minutes have been limited to 180 minutes since CASA algorithm accepts receiving FPLs until 3 hours before EOBT, see annex 1 26

27 Page: 27 Besides, there are nearly 36% of flights that have anticipated more than 120 minutes whereas for all airlines the percent rate was 18%. Low cost airlines anticipate more than average and this may be explained because they cannot swap aircrafts as major airlines may do. Therefore, some airlines will communicate delays as soon as they realise they can not recover the delay during turnd-around FPL Message anticipation impact on ATFM delay As explained in Annex 1, FPLs have to be sent by airlines to CFMU at least 3 hours prior to flight s scheduled EOBT. It can be stated from analysed database that sometimes FPLs are modified afterwards in the timeframe comprised between EOBT and EOBT-180. In the frame of the study, it was analysed whether the modification in FPL would have the same impact in ATFM delay than the ones announced with DLA messages. The reason for this analysis is that airlines could try to manage delays by changing parameters on the FPLs others than the EOBT time (which is managed through DLA message). The sample analyses all flights in the database whom last FPL was the final message taken into account for EOBT. The number of occurrences for FPL anticipation below 160 minutes was quite low and there were too few data in each 20 minutes anticipation range as it can be seen in Figure 15. Therefore delay anticipation windows of 40 minutes were considered. The figure below presents results obtained for the analysis. 26 ATFM delay in minutes Average ATFM delay 19, Anticipation in minutes Figure 14: ATFM delay vs FPL anticipation Non-Weather regulations 27

28 Page: 28 From the observation of the graph, the following points could be highlighted: The overall trend of the curve is negative Between 0 and 80 minutes, there is a strong decrease of ATFM delay that goes from 26 minutes to 19,9 minutes. After the curve stabilizes until 120 minutes at an ATFM value of slightly inferior to 20 minutes. Then the curve decreases again until an anticipation of 160 minutes with an ATFM delay for the window [140,160[ of 13,6 minutes. The average ATFM delay for flights whom last FPL was the final message taken into account for EOBT and with an anticipation between 0 and 160 minutes is 19,1 minutes. In resume, imposed ATFM delay decreases when FPL anticipation increases. Moreover, average ATFM delay imposed when airlines modify initial FPLs trough FPL messages (19 minutes) is higher than when they modify trough DLA messages (18 minutes). The sample size available for the study increases with anticipation. The distribution of anticipation is presented below in two graphs, one presenting the distribution of occurrences according to anticipation and the second depicting the cumulative distribution of anticipation. Number of occurences Anticipation in minutes Figure 15: Number of flights in each anticipation range FPL 28

29 Page: 29 The observation of FPL anticipation distribution shows that a small proportion of flights sends FPL message during the last three hours before EOBT ,0 80 % ,7 7,1 8,8 2,1 2,9 3,7 4,8 11,4 1, Anticipation in minutes Figure 16: Cumulative distribution of flights according to anticipation FPL More precisely, 88% of flights have sent their last FPL before the beginning of slot allocation process (180 minutes). Moreover, the cumulative distribution curve shows that flights are nearly equally distributed among anticipation values from 0 to 160 minutes (nearly constant slope for that range). It can be concluded that ATFM delay decreases with FPL anticipation, as it was the case for DLA anticipation Conclusions The aim of the previous analysis, presented in this section was to demonstrate how ABCD implementation could improve the ATM global situation in minimizing the ATFM delays. The analysis focused on severe regulations, which correspond to regulations imposing heavy ATFM delays to affected flights. Focusing on ATFM delays with high values is of interest as the corresponding delayed flights may have an impact on airlines operations as well as on CFMU since the higher the rate of unused slots, the higher the waste of capacity. The analysis also focused on non-weather regulation as weather regulations due to its lack of predictability may trigger different behaviour from airlines and probably present different results in term of linking the delay anticipation time and delays values. 29

30 Page: 30 The analysis, performed in the present section, showed clearly that, whatever the process used for the delay information (DLA, Flight plan, etc.) and whatever the type of airlines (major, regional, low-cots) the earlier a delay is anticipated and communicated to the CFMU, the lower the resulting delay values. As the implementation of ABCD, thanks to the linkage of flight plan through aircraft identifications, will contribute to sooner anticipate delay and especially delays of subsequent flights using the same aircraft as a delayed flight, it could be stated that the use of ABCD will allow to decrease ATFM delays. As a matter of fact, through ABCD concept implementation, the propagation of a delay detected on a given flight and aircraft, could be analysed and its impact on subsequent flights using the same aircraft could be anticipated and the corresponding delays could be decreased. For example, if a delay is detected on a given flight and if this delay could not be entirely discarded during the following stop time stages, then ABCD will allow to anticipate delays more than one or two flight times before the flights departures. These times correspond to at least an average from 80 to 160 minutes anticipation instead of 40 minutes as it is, nowadays, for more that one flight over two. Hence, ABCD concept would help improving the detection of the most penalizing delays and would smooth the impact of most penalizing regulations. 30

31 Page: Weather Regulation Analysis As explained in 3.2, the analysis was split between weather and non-weather regulations due to the lack of predictability of the first ones. As it was previously pointed out, weather regulations due to its own nature are difficult to predict and variable, as they can evolve when weather conditions change. Under these circumstances anticipation in flight s schedule variations are more difficult to predict. Moreover, expected benefits from delay anticipation are less important as flow rate of the regulation will evolve continuously leading to a lack of predictability even if all flights would be able to anticipate their previous flights delay. Moreover, weather regulations are especially harmful in terms of ATFM delay. Hence, the analysis was done for the top 10 disturbing regulations in terms of global ATFM delay for each day of week from September 11 th to 17 th, From the 70 regulations chosen for the analysis 28 were weather regulations. This chapter explains the impact on ATFM delay that has delay anticipation on this type of regulation Results The figure below depicts the impact of DLA anticipation on ATFM delay for weather regulations. 30 ATFM delay in minutes Average ATFM delay 22, Anticipation in minutes Figure 17: ATFM delay vs DLA anticipation Weather regulations The result obtained for Weather regulations is different from results found for other type of regulations: 31

32 Page: 32 There is not a significant general negative trend as it was observed in the other cases. However, the average ATFM delay for flights with an anticipation inferior to 80 minutes is 23,4 minutes whereas flights that anticipated more than 80 minutes have in average 20,5 minutes of ATFM delay. Thus, the overall trend is negative. The curve oscillates around its average value and seems to be stabilized after 80 minutes. The fact that the curve is not decreasing is the consequence of the uncertainty in weather regulation s evaluation. However, it can be observed that the amplitude of the oscillations becomes lower for bigger delay anticipations. Moreover, the average ATFM delay imposed in weather regulations is bigger for weather related regulations (22,8 minutes) than for non-weather related regulations (18,1 minutes), which again shows the lower predictability of weather regulation s evolution. The distribution of occurrences according to anticipation is presented in the following graphs. Number of occurences Anticipation in minutes Figure 18: Number of flights in each anticipation range weather regulations The distribution of DLA anticipation for Weather regulations follows the same trend as for non-weather regulations: nowadays, most of flights anticipate between 0 and 20 minutes then the number of occurrences keeps decreasing as anticipation increases. 32