EUROCONTROL Air Transport Innovation Network
Developing agile digital solutions and services to fit our stakeholders' needs.
Process
The solutions are developed following an agile approach with three key steps:
Connect
Identify and capture operational challenges and needs from our stakeholders trough an open call, and select the most pressing issues to address.
Build
Develop solutions for the selected operation challenges within a 6 to 12-month timeframe.
Promote
Share the results of our prototyping efforts to highlight the solutions and initiate the next call for operational challenges.
Project Lifecycle
Scoping
The project begins with a scoping phase, during which the EUROCONTROL team collaborates with project participants to understand their specific situations and consolidate them into a common use case. At this stage, we define the project scope, problem statement, and the proposed solution.
Build
Next, we develop the solution defined during scoping. While many of our solutions are AI‑driven, some rely purely on data analytics. When necessary, NDAs can be signed to ensure data confidentiality. Once built, the solution is integrated into one of our prototypes or made available through an API.
Validation
In this phase, we assess both the usability and performance of the solution. The most effective validation occurs through live trials, during which prototypes are connected to real operational data. Project participants use the solution in their daily operations to evaluate its practicality and acceptability.
Deployment
After successful validation, the project moves to deployment. Depending on stakeholder preferences, solutions can be integrated into Network Manager tools, provided via APIs, or released as open source.
Operational
Once deployed, the solution becomes fully operational, allowing users to rely on it directly in their day‑to‑day operations.
EATIN promote events
Every year, we are organising a “Promote” event to showcase the latest developments within the EATIN initiative.
The upcoming Promote event will take place on 23-24 June 2026
At the EUROCONTROL Innovation Hub.
Our projects
Here is an overview of our project's portfolio. For more information on projects, details can be found below.
FADE - Forecast of the ATFM delay evolution
Context
In their current day-to-day operations, airlines have limited information on their regulated flights slot evolution, which makes operational planning not as predictable as they would like. This lack of information can have major impacts on operations, their economic performance, and the passenger experience.
Solution
- Predicted delay – estimation of the final ATFM delay
- Probability to decrease – probability that the current ATFM delay is going to decrease
Benefits
Improved situational awareness, better decision-making on delayed flights, better planning of operations through the day, improved communication between actors (flight dispatchers, pilots and crew)
Partners
- Airlines: Air France, SWISS, Transavia France, Vueling, Turkish Airlines, SAS, Austrian airlines, TUI, SunExpress, Lufthansa, KLM, AirEuropa, WIZZ
- Airports: Copenhagen airport, Aéroports de Paris, Manchester airport
- ANSPs: DFS
Status
- Operational for airlines: FADE is available in NMP Flight and via API since May 2023
- Live trial to be started for airports and ANSPs in 2026
Curfew - Curfew Collaborative Management
Context
The user interface developed within this project uses machine-learning techniques to predict potential curfew infringements for a given flight at a given airport. The ML algorithm has been integrated into EUROCONTROL’s MIRROR platform following a successful phase of internal testing and trial by airlines and airports in 2021. MIRROR creates a sequence of flights for each aircraft allowing the Curfew Collaborative Management to activate mitigation actions much earlier in the day.
The outcome of the trials showed significant operational and financial benefit for airlines and airports. The newly created NMOC Airport Function has integrated the Curfew Collaborative Management application as part of its daily activities.
Solution
- Predicted delay – estimation of the turnaround and knock-on delay especially when ATFM delay induced
- Probability to curfew infringements – probability that the current ATFM delay is going to cause aircraft to infringe an airport curfew and potentially cause a flight diversion or operational flight cancellation
Benefits
Improved situational awareness using predictive delay functionalities, better decision-making on delayed flights, mitigate the impact of airport curfews by pro-active NMOC measures starting early afternoon
Partners
- Airlines: easyJet, SWISS, Transavia France, Vueling, Ryanair
- Airports: Aeroports de Paris
Status
- Operational integration in NMOC Airport Function; the Curfew Collaborative Management is part of the daily tasks executed by the NMOC Airport Function
Knock-on - Knock-on Delay Visualization
Context
Knock-on delays are mostly caused by aircraft rotations, i.e. a late inbound arrival of an aircraft causing a late departure on the next flight operated by the same aircraft. A smaller share of knock-on delays is caused by connecting passengers, crew, or cargo.
Solution
The project objective is to visualise predicted knock-on delay and the impact on traffic demand for the next 12h from an FMP, Airport and Airline perspective.
The project targets knock-on delays caused by the aircraft rotations and uses the calculations done by iDEMAND (based on business rules) and the RNN (Night Curfew) models to predict the impact of knock-on delays on the demand picture for FMPs, airports and airlines.
Benefits
The project aims to create a real-time Decision-Making Support tool:
- For FMPs, the new interface allows to visualise the predicted demand vs. the ETFMS demand (Airspace or Traffic Volume Set) and to identify if/when knock-on delay will dissipate traffic demand peaks rendering the use of ATFM Regulations unnecessary. Or the opposite, when knock-on delays will cause traffic peaks which are not yet visible in ETFMS.
- For airports, the new interface allows to visualise the predicted arrival demand vs. the scheduled demand to identify if/when knock-on delays will shift scheduled peaks with an indication of the predicted arrival delay for each individual flight.
- For airlines, the new interface allows to visualise the predicted departure delay for each flight using a weighing indicator taking the varying degrees of importance (length of delay, aircraft type, seat capacity) into account. The interface also visualises systemic reactionary delays in the airline network caused by suboptimal flight schedules and planning, i.e. identify lack of schedule resilience.
Status
- The knock-on delay visualisation is available in Mirror tool
READI - Real-Time Delay Code Information
Context
Proposed by Istanbul Airport, the objective of this project is to establish a real-time communication of the delay reasons.
If delayed, flights are more likely to have been affected by non-ATFM delays (e.g. knock-on delays, baggage loading, security checks, late fuelling, de-icing, etc.) with little real-time visibility to airport operators, ANSPs and the EUROCONTROL Network Manager on the reasons for these delays. Having a real-time insight into the delay reasons improves the situational awareness across the different actors and allows them to react during the operational day.
There is a lag in the current delay data reporting mechanism with the information becoming available too late to the airport operator to deploy remedial actions to stabilise or improve the local departure punctuality (OTP).
The real-time sharing of the delay information in a centralised manner would allow for the tactical management of staff and resources adjusted to the actual operational situation. This allows early identification of disruptions at outstations potentially impacting the operation and at the same time centralises the collection of delay reasons to be used for post-ops performance analysis.
Solution
The READI project addresses a current need for airport operators and its ecosystem of operational stakeholders: real-time awareness of the delay drivers at the (own) airport and understanding of the delay drivers at outstations (departing airports for inbound flights).
The READI project grouped airports, airlines, ground handlers with IATA endorsing the project and EUROCONTROL acting as “honest data broker” providing the technical platform.
The lack of real-time delay information can be addressed by technology (in the near future) but more importantly it’s the legacy delay code schema that limits the live sharing of delay causes. The legacy delay code schema focus is on the delay reason and stakeholder rather than the turnaround process.
The legacy IATA AHM730 delay code schema was therefore deemed not fit for the live trials which were conducted between May 8-21 2023.The READI project opted to use the new IATA AHM732 delay code schema which has a 3-layer structure:
- Process – which aircraft turnaround process is affected?
- Reason – what is the actual reason causing a flight delay?
- Stakeholder – which stakeholder is responsible? ex. handler, passenger, etc.
The READI project only used the first Process layer to power a performance dashboard displaying live operational information helping local stakeholders to deploy tactical measures. The second and third layer of the AHM732 was not used in the READI project but in future will allow in-depth post-operational analysis and steer strategic changes.
Benefits
The results of the limited (both in time and scope) live trials showed great value in a process centric reporting of tactical issues. On the day it is (often) sufficient for operational partners to know which turnaround processes are under pressure with post-ops analysis allowing for more in-depth analysis into the delay reasons and assess which stakeholders caused the delays.
Partners
- Airlines: SAS, flydubai, Swiss, TAP, Turkish Airlines, IATA
- Airports: Istanbul, Frankfurt, Prague, Brussels, Aeroporti di Roma, Schiphol
Status
- Pending for deployment through further development in the READI 2.0 project
READI 2.0 - Real-Time Delay Code Information application
Context
READI 1.0 project from EATIN Cycle IV focused on gathering information on pre-departure delays in real time, using the new IATA delay code schema AHM732 (three information layers: Process- Reason-Stakeholder). This information was collected via an app where the user would manually input the required information including the delay code.
The READI project used the first Process layer of these delay codes to power a performance dashboard displaying live operational information helping local stakeholders to deploy tactical measures. The second and third layers of the AHM732 were not used in the READI project but in future will allow in-depth post-operational analysis and steer strategic changes.
The task of reporting these delays and especially identifying the correct delay code can be a burden for the ground handlers and other personnel responsible for this task.
Proposed by Austrian Airlines, the main goal of READI 2.0 is to allow delay reporting in IATA AHM732 with the help of AI to take the burden off the users.
Solution
With the help of an LLM (Large Language Model) trained for this purpose, READI 2.0 will allow the user to describe the delay either via text or voice input and obtain a suggested delay code (2 first layers: Process- Reason). This functionality is integrated in the previously developed READI app (under READI 1.0) that supported the real time reporting of delays.
Benefits
Move from blame game to process oriented reporting.
- Higher number of partners reporting delay due to less workload for frontline staff.
- Easier and more efficient way of coding, reducing room for human error.
- Opening the possibilities to digital turnaround solutions in terms of delay coding.
Partners
- Airlines: Austrian, flydubai, SunExpress, Vueling, Pegasus, SAS Scandinavian Airlines, Iberia, Swiss, Air Serbia, Norwegian, Aer Lingus, Pegasus, IATA
- Airports: Istanbul, Dusseldorf, Prague, Brussels, SEA Milan Airports, Schiphol
- Ground Handlers: Swissport
ENIP - EUROCONTROL Network Manager Interface for Pilots
Context
EUROCONTROL must be at the heart of all planning technology to drive Airline and Pilot behaviour.
Solution
The project will develop a mobile app for pilots to access real-time planning data (on ground), mission planning based on flight plan data, and NM insights, as well as the outputs of mature EATIN solutions (FADE - delay prediction - will be the first model to be deployed on the ENIP application).
Benefits
With enhancements from AI models, the app will bring operational data into the cockpit and enable proactive responses taking into account the Network context. Pilots will benefit from an overview of the entire operational day, including delay propagation, and will gain a deeper understanding of ATFM and Network dynamics directly in the cockpit. The system will ensure consistency of data and information between the Operations Control Centre and pilots, reinforcing the engagement to “fly the plan.” Safety will be enhanced by providing the latest filed Flight Plan, and strict adherence to B2B user agreements, access controls, and data segregation will be maintained at all times.
Partners
Aegean Airlines / Aer Lingus / Air Dolomiti / Air Europa / Airbus (ATI) / AirExplore / Austrian Airlines / Avion Express / British Airways / Brussels Airline / Condor / easyJet / Eurowings / Iberia / JetHouse / LOT Polish Airlines / NetJets / SunExpress Airlines / Swiss / TAP Portugal / Transavia France / Transavia NL / TUI Group / Vueling Airlines / Wizz Air
PETA - Prediction of Estimated Time of Arrival
Context
Estimated time of arrival (ETA) is important for all aviation stakeholders because it is an input into various air traffic management (ATM) processes, during several flight phases, starting from the moment the flight plan is submitted up to the moment the flight departs.
Solution
PETA is essentially an algorithm that links several existing machine learning models in series to make more accurate estimated time of arrival predictions prior to departure. Specifically, the first model (Knock-on) anticipates rotational reactionary delays arising from unrealistic available turn-around times; the second model (FADE) forecasts the evolution of air traffic flow management delays for regulated flights; and the third model, AirborneTime, makes improved predictions by learning the systematic discrepancies between reported and actual airborne times. The fourth and final input takes the taxi times used by NM's ETFMS.
The model/algorithm can make ETA predictions for intra-European flights for all rotations on the day of operations.
Benefits
The average ETA prediction from PETA is closer to the Actual Time of Arrival (ATA) than the current system's average. However, sometimes PETA's ETA prediction can be worse than the current system's. The most significant improvement in ETA accuracy compared to the current system is around 2-6 hours before EOBT.
Expected operational benefits:
- Improved operational efficiency for ground operations (turnaround processes, gate allocation, allocating resources);
- Additionally, improved operational efficiency for ATC (specifically for the FMP) because demand is predicted more accurately;
- Cost-effectiveness for airlines from reduced missed connections;
- Better passenger experience (identifying and finding solutions earlier to likely missed connections; allocating appropriate resources to very early/late flights).
Partners
- Airlines: Air Portugal, Ryanair, Austrian, Swiss, KLM
- Airports: Gatwick, Istanbul, Heathrow, Prague
- ANSPs: NATS, ANS CR
ECCP - En-route Capacity Constraint Predictor
Context
The ECCP project aims to improve Demand-Capacity Balancing (DCB) in Europe. Today, ATFM regulations are often applied late and inefficiently, especially during convective weather events. ECCP seeks to deliver accurate and explainable forecasts to anticipate these situations. The main issues are the lack of anticipation of en-route capacity drops, particularly at D-1, heavy reliance on human expertise without systematic post-event analysis, and limited coordination between Air Navigation Service Providers (ANSPs) and the Network Manager (NM), which leads to costly reactive measures.
Solution
The project’s objectives are to predict en-route capacity reduction per Traffic Volume (TV) caused by convective weather from D-1 to D0 and provide an intuitive human-machine interface and, if requested, an API for operational trials. The current scope excludes hotspot or regulation prediction.
The approach combines a backend AI using machine learning models that leverage weather data, traffic data, and historical ATFM data, with a frontend HMI for visualising predictions such as weather maps, imbalances, hotspots, and network impact. The workflow integrates into the DCB process in pre-tactical and tactical phases, with the goal of anticipating issues at least 30 minutes before human detection.
Key deliverables include an API and HMI for trials in the NMVP environment and visualisations such as weather maps and risk zones, capacity degradation per TV over time, hotspot tables and maps with severity indicators, and network impact metrics like delays and affected flights.
Benefits
The expected benefits are a reduction of late regulations and ATFM delays, improved network resilience against convective storms, and stronger collaboration between ANSPs and the NM through shared, explainable forecasts.
Partners
- Airlines: Aer Lingus, British Airways, Iberia
- Airports: Dusseldorf Airport, Gatwick Airport
- ANSPs: Croatia Control, DFS, DSNA, ENAV, MUAC, NATS, Skeyes
KORI - Knock-on regulation impact
Context
In airport operations, delays can cascade across flight schedules. A common scenario occurs when an aircraft arrives late from its previous leg, causing a knock-on delay for its subsequent departure. This disruption can be further compounded if the next flight is then subjected to air traffic flow management regulations, resulting in additional delay. Such situations not only affect punctuality but also complicate resource planning and passenger handling, highlighting the need for predictive tools that can anticipate and mitigate these operational impacts.
Solution
Leveraging real-time operational data, KORI calculates the total expected delay affecting a flight, including delays inherited from previous legs. It also provides information on possible regulations on alternative routes.
Benefits
- Airlines: better plan their fleet taking in consideration the network impact on their flights, more informed decision-making
Partners
- Airlines: Vueling, Swiss and Transavia
OPTT - Machine learning Prediction of Turnaround Times
Context
An analysis performed by EUROCONTROL assessed that the total airport ground delay represented a cost of €1.8 billion in 2024 over the ECAC area. It is nowadays commonly recognised that the ATFM world lacks visibility on what happens between in-block and off-block although ground delay might propagate over the network. Frequent disruptions and manual entry errors make accurate TOBT estimation both challenging and essential for reliable pre-tactical and tactical planning, as they can significantly affect TSAT and CTOT calculations.
Solution
OpTT provides dynamic probabilistic predictions of turnaround duration and last TOBT release using data provided by the airports participating in the project.
Benefits
- General: Improve the stability of TOBT releases indicating that fewer updates will be necessary using the model predictions compared to the currently published TOBT values.
- Airports: improve gate and stand management by estimating aircraft readiness for departure; improve resource allocation (e.g., ground handling, runway use); minimize human error in TOBT entry and reduce last-minute TOBT changes
- Airlines: set reliable TOBTs, especially when operating outside their hubs; reduce manual TOBT adjustments
- ANSPs: allow earlier and accurate TSAT (Target Start-up Approval Time) calculations
Partners
Swedavia, Aeroporti di Roma, Geneve and Prague airports
Status
Operational for Prague airport
OPTT2.0 - Optimised Turnaround Time Prediction
Context
At CDM airports, the Target Off-Block Time (TOBT), set by airlines or ground handlers, indicates when an aircraft is expected to depart and must be updated if it deviates by more than ±5 minutes. Frequent updates due to flight delays, technical issues during ground operations, or late arrivals of passengers or crew members make accurate TOBT estimations challenging but critical for effective pre-tactical and tactical planning. Beyond these considerations, manual TOBT entry is prone to error, requiring frequent last-minute changes possibly impacting TSAT and CTOT calculations.
Solution
OpTT2.0 provides dynamic predictions of turnaround duration and of last TOBT value across all European CDM airports and for any airline operating from these airports. Predictions are available already several hours before turnaround operations (as soon as a link between inbound and outbound flights can be established) and update as turnaround operations approach.
Benefits
- General: Improve the stability of TOBT releases indicating that fewer updates will be necessary using the model predictions compared to the currently published TOBT values.
- Airports: improve gate and stand management by estimating aircraft readiness for departure; improve resource allocation (e.g., ground handling, runway use); minimize human error in TOBT entry and reduce last-minute TOBT changes
- Airlines: set reliable TOBTs, especially when operating outside their hubs; reduce manual TOBT adjustments
- ANSPs: allow earlier and accurate TSAT (Target Start-up Approval Time) calculations
Partners
- Airports: Brussels, Aéroports de Paris, GESAC Naples, Prague, Dusseldorf, Aeroporti di Roma, SEA Milan, Schiphol, Frankfurt, Copenhagen airports.
- Airlines: Vueling, Austrian, TAP, Turkish, SunExpress, AirDolomiti, Swiss Airlines
Status
OpTT2.0 is available via API and through on-site deployment at airport facilities requesting it.
PRIU - Predicted Runway-In-Use
Context
Airlines and other airport stakeholders often receive runway usage notifications for departures and arrivals with limited advance notice. This lack of timely and dynamically updated information can lead to conservative planning assumptions, such as overestimated fuel requirements, ultimately impacting operational efficiency and cost-effectiveness.
Solution
PRIU delivers dynamic runway usage forecasts for both inbound and outbound flights, taking into account current and anticipated weather conditions as well as airport-specific operational factors. These predictions are available several hours prior to departure and include probabilistic assessments for each potential runway, enabling more informed decision-making and improved tactical planning.
Benefits
- Airports: Earlier visibility in potential RWY direction changes leading to proactive adjustments to ground operations, stand allocation and contingency planning
- Airlines: More accurate flight plan information, reduced and optimised fuel consumption
Partners
- Airlines: Austrian, Vueling, SunExpress, Swiss, Air France, Transavia France, Transavia Netherlands
- Airports: Brussels, Heathrow, Aéroports de Paris, IGA Istanbul, Prague, Vienna, Stockholm
Status
PRIU is available via API. It will be available via OpenSource in 2026.
TITOP - Taxi in/out Time Prediction
Context
Airport operations currently rely on fixed taxi times, without accounting for factors such as weather conditions or ongoing airport work.
This lack of real-time information and dynamic adjustments to taxi times results in overestimated fuel requirements for airlines and inefficient resource planning at airports, including stand allocations and ground handler availability.
Solution
TITOP provides dynamic estimates of taxi-time durations for both arriving and departing flights, factoring in weather conditions and airport-specific operational parameters. These predictions are available several hours prior to departure and offer precise estimations of the expected taxi-time between a designated stand and the assigned runway, supporting more accurate planning and resource allocation.
Benefits
- Airports: More efficient stand planning for airports, better awareness for ground-handlers planning with better use of resources
- Airlines: More accurate flight plan information, reduced and optimised fuel consumption
Partners
- Airlines: Austrian, Vueling, SunExpress, Swiss, Air France, Transavia France, Transavia Netherlands
- Airports: Brussels, Heathrow, Aéroports de Paris, IGA Istanbul, Prague, Vienna, Stockholm
Status
TITOP API and OpenSource code will be available in 2026.
De-icing - Probability for de-icing
Context
Airport operators and ground-handlers typically receive notification of de-icing requirements approximately one hour prior to departure. This limited lead time restricts their ability to proactively allocate resources and coordinate operational activities, potentially impacting overall efficiency during winter operations.
Solution
The De-icing model estimates the likelihood of de-icing operations several hours in advance for each individual flight. This forecast is generated using detailed weather data, enabling proactive planning and operational efficiency in adverse weather conditions.
Benefits
- Airports: Proactive planning, resources optimization and operational resilience
- ANSPs: Better TSAT allocation
Partners
- Airlines: Austrian, Vueling, SunExpress, Swiss, Air France, Transavia France, Transavia Netherlands
- Airports: Brussels, Heathrow, Aéroports de Paris, IGA Istanbul, Prague, Vienna, Stockholm
Status
De-icing API and OpenSource code will be available in 2026.
MASSDIV - Massive Diversions
The project aims to create a tool that automatically assesses the stand capacity available in an airport’s network to improve stakeholders’ common situational awareness to manage situations related to massive diversions. The main challenges this project addresses include:
Heavy capacity reduction can cause massive diversions:
- Flights already airborne to alternate airports.
- Possible over-deliveries to other airports, if more than two flights expected.
- Apron congestion and possible safety issues.
Coordination process is commonly performed manually and involves:
- Surrounding airports to provide apron capacity.
- Different ATC units, to coordinate with pilots and handle diverted traffic accordingly.
Pilots and airlines, to decide where to divert and FPL changes if needed.
Partners
- Airlines: SAS, Swiss, Vueling, Euro Jet
- ANSPs: DSNA, ENAIRE, ENAC, ENAV, UK CAA
- Airports: AENA, Aeroporti di Roma, SEA Milan
- Universities: CVUT Prague
iSWAT - Stand Waiting Time
Context
In day-to-day airport operations, stand waiting time often arises when inbound flights arrive ahead of schedule or outbound flights experience delays. Existing stand allocation systems typically lack automated, real-time insight into stand availability at the estimated arrival time. This issue is further exacerbated during peak periods with limited stand capacity.
Currently, there is no automated mechanism to measure stand waiting time, nor is this delay factored into taxi-in time calculations, limiting the accuracy of operational planning.
Solution
iSWAT aims at improving the accuracy of stand waiting time predictions for arriving aircraft, focusing on reliable estimates 20–30 minutes before ELDT. Key efforts include refining landing time and off-block time estimations, integrating these into a stand allocation tool, and displaying predicted stand waiting times.
iSWAT will also define metrics to measure both actual and estimated stand waiting times, supported by a confidence index to assess reliability.
Benefits
- Airports: more efficient stand management and operational planning, increased situational awareness and decision-making in the stand allocation, improved predictability and punctuality with fewer additional taxi-in times, better resources allocation
- Airlines: increased situational awareness, fuel efficiency
Partners
- Airlines: Transavia France, Vueling
- Airports: Brussels, Aéroports de Paris, Prague, Dusseldorf, Swedavia and Frankfurt
Status
The project will validate the solution in 2026.
Aviation Data Corridor
Context
The project is addressing challenges such as multiple point-to-point data interfaces between airlines, airports and ground handlers; airport to airport data exchange; security aspects regarding the exchanged data, etc. Currently, data sharing often relies on one-to-one interfaces, with some processes automated (e.g., A-CDM data) and others still manual (e.g., passenger forecasts). In many cases, multiple interfaces coexist, increasing complexity, cost, and implementation time.
Solution
Its objective is to improve the information exchanged between different operational partners – airports, airlines, ground handlers and the EUROCONTROL Network Manager. The project team has built a data exchange platform covering flight related data relevant to planning and operations with a standard SWIM data interface.
Benefits
Real-time data exchange improves operational efficiency among all stakeholders.
Partners
- Airport: Gatwick
- Airline: Easyjet
- ANSP: NATS
- Industry: Thales
Status
EUROCONTROL made the documentation publicly available to the market, so that commercial providers can use it to further develop the platform as a product.
RWY incursions - Preventing Runway Incursions
Context
Runway incursion analysis is hindered by limited data sharing across Europe due to confidentiality concerns and inconsistent definitions, detection methods, and severity classifications. This results in fragmented and heterogeneous datasets. Airports must rely solely on their own limited incident data, which restricts the ability to identify broader trends or emerging risks. Consequently, safety analysis often leads to isolated, case-specific mitigations rather than systemic improvements, limiting the potential for collaborative learning and proactive risk management.
Solution
This process innovation project delivered a comprehensive approach to quantify the runway collision barrier model. Using the Network Manager Safety Functions Map (SAFMAP) comprehensive barrier model, the project researched and validated an approach to characterise and quantify the individual barriers success and failure rates and to integrate these into an overall risk and resilience quantitate model. The approach allows users to investigate model sensitivities to changes in the traffic conditions and to analyse the effects of addressing specific risk.
Benefits
A great benefit from using the approach is the ability to explicitly predict safety benefits of new investments – new infrastructure, procedures, or training.
TMA dashboard - Prediction of arrival airborne delays before take-off
Context
This project, initiated by airlines, aims at supporting fuel planning by providing information regarding airborne delays at arrival (i.e. in the terminal area) resulting from holding and sequencing. The motivation is to better take into account the impact of weather at destination on fuel planning. Indeed, today a conservative approach for fuel planning is generally taken, leading to an over estimation of the contingency or extra fuel.
A first project was conducted to predict arrival airborne delays before take-off, relying on a machine learning approach using historical data of arrival delays, traffic demand and weather (*). However, this required an extensive airline validation that was uneasy to conduct at that time.
Solution
A “quick win” was identified in the form of statistical dashboard of historical arrival additional times. Even if it only provided a partial view (no current additional time, no traffic demand, no weather, no airport event, ..), the participating airlines expressed an interest for fuel planning as it is not available in their tool (**).
The dashboard was made available online with the historical additional time per 30 minutes period of the top 30 European airports complemented by a view of the historical and current traffic demand. For a given airport, it display typical statistical values (median, 90th, 98th, and 99th percentiles) for a given day of the week and month (or quarter).
Benefits
This dashboard is expected to assist flight crews in a better estimation of the contingency or extra fuel, in view of improving flight efficiency (general case of over estimation) or reduce the risk of diversion with significant operating costs (rare case of under estimation). However, confidence expectations and regulatory aspects remain to be clarified.
Partners
- Airlines: Air Europa, Swiss, Transavia, Turkish airlines
SOPEL - SID/STAR Operations Predictions
Context
The project aims to develop a model to improve predictions of planned Standard Instrument Departure (SID) and Standard Terminal Arrival (STAR) procedures at the 12 and 3-hour mark prior to the Estimated Off-Block Time (EOBT) for each flight.
Solution
A machine-learning model dynamically predicting the most likely SID/STAR to be flown by each flight and the distance metric associated with the likely SID/STAR to be flown, coupled with a post-processing algorithm to identify whether the most likely procedure is plannable, and if not also displaying the most likely plannable procedure for the flight.
Benefits
Airlines
- Cost savings: With increased predictability of SIDs/STARs assigned during the flight planning stage, translated also into higher acceptance rate and reliance on Operational Flight Plans (OFPs) by flight crews, airlines would be able to improve their fuel planning and better optimise fuel for a given flight.
- Reduced carbon footprint: By being able to minimize the extra fuel for a flight, the aircraft would carry less mass, which will translate into less fuel burned, and therefore reduced carbon footprint of the flight.
- Reduced flight crew workload: Flight crew will be able to rely more on the information provided in the OFP. They will also be less exposed to last-minute changes to the trajectories flown before the take-off and descent.
Network Manager
- Improved network management: More accurate SID/STAR predictions may lead to better informed decision-making process when applying flow measures and calculating the impact on the overall network.
- Improved quality of shared information: NM can propagate more reliable flight trajectory information in the shared FPLs to act upon by the concerned ANSPs.
ANSPs
- Improved ATCO workload and ATC sector loading: ANSPs may utilize more accurate predictions of SIDs/STARs to inform their estimates of upcoming workload and ATC sector capacity/complexity metrics.
- Reduced carbon footprint: ANSPs may use the SID/STAR predictions to propose more environmentally friendly gate-to-gate trajectories for selected flights of interest.
Partners
- Airlines: Aer Lingus, Air France, Iberia, Lufthansa, SunExpress, Swiss Airlines, Vueling, Air Europa, Amelia
- Airports: Gatwick Airport
- ANSPs: MUA
CURDEP - Curved Departure Procedures
Context
Advanced curved departure procedures allow the aircraft to make an early turn (as soon as the aircraft crosses the departure end of the runway), which enhances runway throughput and flight efficiency. Under the EATIN CURDEP project, EUROCONTROL and partners are performing validations and data collections to demonstrate that the procedures are safe and feasible from a flyability point of view.
Solution
Trajectory data and flight crew feedback are collected, operating the procedures in 6 different EASA Level D certified flight crew training simulators (A350, A320, A220, B737, B747-8 and E190) and in various environmental conditions. The curved departure route is coded in the aircraft’s database using the ARINC 424 “Radius-to-Fix (RF)” path terminator, which makes the aircraft behaviour very reliable with extremely predictable ground tracks. By design, the planned turn radius and groundspeed allow control over the aircraft bank angle (limiting the bank angle at low altitudes).
The collected data and flight crew feedback will be used to demonstrate that the procedures are safe, reliable and efficient. Live flight trails are currently being discussed as a next step. Results will be presented at the ICAO Instrument Flight Procedures Panel (IFPP) to encourage the design of new instrument flight procedures criteria allowing the implementation of these procedures.
Benefits
The procedures are expected to provide potential fuel savings in the order of at least 30 kg for a typical intra-European flight, when compared to similar procedures using the current ICAO PANS-OPS design criteria. Initial feedback from flight crews involved in the validations is extremely positive.
Partners
- Airlines: NetJets, SunExpress
- Airports: Swedavia, Aéroports de Paris, Fraport
PAX Demand - Passenger Demand prediction
Context
The objective of the project was to improve the operational performance through the development of a predictive tool providing pre-tactical and tactical predictions of in-block and off-block time deviations supported by predictions on passenger demand (load factors).
Solution
We used the Power BI dashboard to visualise the outputs of the model and overall predictions at flight level. The predictive model and the dashboard designed for Geneva Airport have been validated through an operational trial, whose objective was to evaluate the accuracy of the model and to check the dashboard usability from an operational perspective.
The project members concluded that the validation results form a very good basis to continue the development by reworking the algorithms focusing mainly on in-block time predictions as well as on revising the layout of the dashboard.
Benefits
Improved predictability of potential disruptions within the landside infrastructure brought about by increased demand linked to passenger volumes.
Partners
35 European airports currently subscribed.
Status
Today, the daily predictions are available also via Network Manager (NM) B2B Services for those airports already connected. At the same time, the model can be calibrated with historical load factors, which are to be provided by airport operators through the same NM B2B Services.
PaxCOIN - Predicted Passenger Connections Indicator
Context
Airlines, ground handlers and airport operators may have limited automated information available to predict the passenger connectivity on the day of operation for Customer Care and iOCC to make effective decisions.
At present these services often rely on public websites (e.g. Flightradar24) to check the arrival time of third-party (long-haul) flights arrivals to assess whether or not connecting passengers and their baggage will make their transfer during periods of high ATFM or knock-on delays in the network.
Solution
The objective of the PaxCOIN project is to create a multi-tiered indicator which can be easily shared amongst all operational stakeholders at an airport showing the probability (based on EATIN’s PETA) and operational impact of passengers and/or baggage not making their connecting flights. The goal of this indicator is to enhance the travel experience for passengers and to boost the operational effectiveness for airlines, airports and ground handlers.
The existing EUROCONTROL MIRROR tool is used to visualise the PaxCOIN indicator, the PETA “powered” predicted delay and a newly developed tail-swap predictor supporting the decision making process of operational staff.
Benefits
The aim of the project is to conduct live trials during 2025 and assess whether the PaxCOIN based prioritisation of flights provides an operational and financial benefit (especially related to EU261 Passenger Rights Regulation).
This can be achieved through a reduction of missed passenger connections, an earlier detection when a passenger connection will be missed so mitigation measures can be activated sooner, a reduction of EU261 related costs, an optimisation of the passenger connections within and outside the airline concerned or an optimisation of the park and gate procedures.
Partners
- Airlines: Vueling, SunExpress, Austrian Airlines, Swiss Airlines, Air Serbia, TAP Airlines, Finnair
- Airports: Vienna Airport, Brussels Airport, Dusseldorf Airport, Aeroports de Paris
- Ground Handler: Swissport
PRELUDE – Prediction of Luggage Delivery Time
Context
Baggage waiting time at claim areas significantly affects the experience for arrival passengers at airports. In most cases, passengers are not provided with an estimated time indicating when their luggage will arrive at the claim area.
Solution
PRELUDE provides predictions for the arrival times of the first piece of luggage at the claim area, as well as for the duration of luggage loading onto the belts.
Benefits
- Passengers: make use of the predicted waiting time to plan onward travel (bus, taxi, train, etc..), inform friends or relatives coming to pick them up at the airport, or take care of other personal tasks.
- Airports: enhance belt allocation process and passenger management in the luggage claim area
Partners
- Airports: Brussels, Munich, GESAC Naples, Prague, Dusseldorf, Aeroporti di Roma, SEA Milan, AENA, Bologna, Copenhagen and Swedavia airports.
- Airlines: Vueling, SunExpress, Pegasus Airlines
Status
Solution has been handed over to Munich, Aeroporti di Roma, SEA Milan and Dusseldorf airports. While initial testing and local validation activities have begun, information to deploy the solution is already available to the requesting airports.
NO.BAG – Prediction of Baggage Amount and Location
Context
Mishandled baggage has become a significant concern for airlines and airports. This problem is particularly severe in Europe, where the rate of mishandled bags is notably higher compared to North America and Asia. Mishandled bags not only create logistical challenges but also result in substantial financial losses for airlines, with each mishandled bag costing between 80-100 euros (source: SITA Baggage IT Insights 2024)
Solution
Two models are proposed:
- · A machine learning model that predicts, for a given flight, the number of bags in the hold;
- · A machine learning model that predicts, for a given flight, the number of transfer bags in the hold. The models will be useful at least up to seven days before a flight’s departure.
Benefits
The models will be useful for ground handlers, the airport and the airlines.
Expected operational benefits:
- Improved cost efficiency in terms of reduced number of mishandled transfer bags because of better allocation or ground handling resources and assets;
- Improved operational efficiency for ground handlers because the reclaim hall and belt are allocated according to the number of bags, not number of passengers (more space for handlers to work);
- Reduced environmental impact because of earlier and better accuracy of weight and balance calculations, resulting in less excess fuel being carried;
- Better passenger experience (because of reduced time to reclaim bags, and fewer mishandled transfer bags).
Partners
- Airlines: Austrian, Air Europa, Vueling
- Airports: Schiphol
- ANSPs: AENA
EDDY - Early Diversion Detection System
Context
In 2019, out of 9.9 million flights with a destination in the EUROCONTROL Network Manager (NM) area, 20,257 flights (0.2%) landed at an airport other than the one initially planned. If the likelihood of diversion could be predicted in advance, appropriate mitigation measures like changing the schedule or delaying the flight could be implemented to alleviate the negative consequences such as crew and fleet disruption, economic impact and bad passenger experience.
The project has then emerged due to the need of airspace users and airports to have a better predictability of the periods of diversion due to weather events. Nowadays, the information on the periods of diversion is very limited for the airspace users and airports, raising high uncertainty on the operations management throughout the day when the weather gets worse. Stakeholders do not have any tool to know which is the probability for a given flight to be diverted in terms of weather.
Solution
Proposed by Vueling, an AI model (machine learning) has been developed to predict the likelihood (probability) and trend of diversion of a flight, due to weather events.
The machine learning model has been trained on more than 35 millions of flights (from which 0.2% diverted) from January 2021 to May 2025 to identify relevant predictors.
The tool is able to provide a probability of diversion for each flight (between 0 and 100%), helping the airlines and other actors to better either take mitigation actions to avoid the diversion (change scheduled flight or delay the flight), or at least anticipate a diversion (selection of the most suitable airport, preparation of the logistics, support to pilot decision-making…).
Benefits
EDDY allows airlines to better anticipate the disruption of a flight diversion. First, flight dispatchers may act to avoid the flight diversion by taking proactive measures such as rescheduling or delaying flights. In case the diversion cannot be avoided, by offering greater situational awareness to flight dispatchers and pilots, the airline is able to prepare the diversion (e.g. select the most suitable airport) and minimize its consequences (operational disruptions, logistics, delays, costs…).
Thanks to the anticipation, the airlines does not only reduce the high operational and financial costs of diversions but also minimizes environmental impact from unnecessary fuel burn.
Furthermore, the system enhances safety, efficiency and overall resilience in daily operations by providing flight dispatchers and pilots with enhanced situational awareness.
Partners
- Airlines: Vueling, SunExpress, Transavia, Swiss, Air Europa
- Airports: Prague airport, Istanbul airport
- ANSPs: ENAIRE
- Others: EBAA
Status
Operational via API and on-going deployment in NMUI Flight
WIND - Historic Weather Data Dashboard
Context
Launched in July 2023 with the active participation of four airlines – Vueling, Turkish Airlines, Netjets, and SunExpress – the WIND project was created to address a key industry challenge: the lack of an official, robust, and centralised database of airport weather information to support scheduling and performance analysis.
Solution
WIND was proposed by Vueling during the 6th ideation cycle of EATIN. It has since grown into a collaborative initiative, bringing together multiple airlines to co-develop and validate a solution that meets real operational needs.
At the core of WIND is a dynamic Power BI dashboard that provides airlines with easy access to historical METAR data for each airport.
- Flexible filters: date, time, and specific weather parameters.
- User-friendly visualisation: interactive charts and insights.
- Export options: data can be downloaded in Excel format for further analysis.
An open link version is available here
To offer a broader operational picture, WIND also integrates in a private version with pre-accepted access:
- Historical regulations
- Diversion statistics
- Runway utilisation percentages
Partners
- Airlines: Vueling, SunExpress, NetJets, Turkish Airlines, Iberia, Qatar Airways
Status
The tool has been validated by the participating airlines and is already in operational use. Data is updated on a three-month cycle, ensuring reliability and relevance.
HAWAII - Hazardous and adverse weather impact at airports
Context
When the predicted traffic demand exceeds the expected airport capacity, air traffic flow management (ATFM) regulations are frequently implemented to prevent potential overloads. To manage the situation, flights affected by ATFM regulations are assigned ground delays, known as ATFM delays, with the purpose of smoothing the traffic demand and keeping it below the capacity. ATFM regulations are quite common at European airports.
Precisely, from the resurgence of air traffic following the lifting of COVID-19 pandemic restrictions on June 15th, 2021, until May 31st, 2023 (just prior to the creation of this document), a total of 1.3K ATFM regulations were implemented at airports within the European Civil Aviation Conference (ECAC) region. These regulations resulted in a total ATFM delay of 430K minutes across the network. Most of these ATFM regulations were caused by air traffic control (ATC) capacity (25%) but also by adverse weather conditions (13%).
Needless to say, accurate estimation of airport capacity is essential for ensuring the effective and efficient implementation of ATFM regulations. Overestimating capacity may force flights to wait in holding stacks, causing significant environmental impact, and increasing fuel-related expenses. Underestimating capacity, on the other hand, would result in an excessive number of unnecessary ATFM delays, raising operational costs. However, even with a flawless prediction of airport capacity, the efficient implementation of weather-induced regulations at the airport remains uncertain.
Furthermore, despite having increasingly accurate weather forecasting models, aviation stakeholders do not have any means to predict the occurrence of weather-related regulations.
Solution
The goal of the HAWAII project is to offer comprehensive support to various stakeholders amidst challenging weather conditions, with a primary focus on providing them with accurate estimates of the impact on airport capacity.
To achieve this objective, a set of machine learning models has been developed:
- The first two models are designed to forecast airport capacity (expressed in movements per hour) for arrivals and departures,
- Another model aims to predict the likelihood of air traffic flow management (ATFM) regulation (expressed in probabilistic terms, from 0 to 100%),
- The last two models are predicting the actual rate of the regulation for arrivals and departures.
Benefits
HAWAII provides operational benefits for various stakeholders depending on the model used.
The Airport capacity prediction (departures and arrivals) aims at supporting the ANSP in the definition of the regulation rate. Regulations are thus optimized, mitigating the risk of over-regulating or under-regulating.
The Regulation probability prediction helps ANSPs, airports or even airlines to anticipate the regulation and its related disruptions up to 30 hours in advances. They can therefore take mitigating actions to limit the impact of the regulation on their operations.
The Regulation rate prediction (departures and arrivals) provides an additional information to the airports and airlines to better anticipate the capacity reduction at the airport.
Partners
- Airlines: SWISS, Transavia, Vueling, Turkish Airlines
- Airports: Istanbul Airport IGA, Prague Airport, Aéroports de Paris, Heathrow, Schiphol, Brussels Airport, Flughafen Zurich
- ANSPs: Fintraffic, ANS, Skyguide
RETOP – Reduced Engine Taxi Out Prediction
Context
There is significant variability in how RETO is implemented across airlines and airports. Adoption levels range widely—from roughly 20% at carriers with newer or less formalized programmes to more than 50% at airlines with more mature RETO practices. In most cases, the final decision to apply RETO rests with the captain, who evaluates real‑time operational factors such as expected taxi time, airfield complexity, airport layout, and prevailing weather conditions.
Solution
The objective of the RETOP project is to develop a real-time predictive model that accurately forecasts the aircraft's take-off time (TOT) just before the start-up clearance. As a consequence, RETOP is expected to assist the crew in assessing whether the taxi time and environmental conditions are suitable for executing a reduced-engine taxi out (RETO) operation. Additionally, the model/solution can recommend the optimal time to initiate the engine start-up process based on each engine warm-up time.
Benefits
Stakeholders identified several key benefits that RETOP could deliver:
- Improved operational confidence: Providing flight crews with more accurate predictions of take-off sequencing and timing would increase confidence in RETO decisions, potentially converting non-RETO operations to RETO and extending single-engine taxi durations.
- Environmental and cost benefits: Reduced fuel consumption, carbon emissions, and noise.
- Enhanced reporting capabilities: RETOP could support environmental performance dashboards and emissions reporting required for regulatory compliance.
- Optimal decision support: The system could serve as a decision aid without replacing pilot judgment, helping to "de-risk" RETO procedures and overcome conservative tendencies that currently limit implementation
Partners
- Airlines: Air Dolomiti, Aer Lingus, TAP Portugal, Air France, SunExpress, SWISS, Iberia, Austrian Airlines, Airbus, Air Europa
- Airports: Brussels, Munich, Prague, Copenhagen
MiniCO₂ - Runway Allocation / Minimising CO₂
Context
Paris CDG is conducting simultaneous independent parallel approaches using two arrival runways. The management of arrivals, including runway allocation, is supported by an arrival manager (AMAN). The allocation of the arrival runways currently uses two strategies. The geographical strategy considers the entry point of each flight to provide a segregation of arrival flows and avoid crossings in the air. It is typically used under high traffic to avoid excessive workload for the approach air traffic controllers.
The preferred strategy selects the runway that will result in a minimum taxi time according to the gate/stand allocated to the flight. While this strategy may de facto reduce fuel consumption and CO2 emissions for the taxi phase, it does not account for fuel and emissions in the terminal area.
Solution
The solution lies in the introduction of a new strategy for runway allocation in the arrival manager, aiming at optimising fuel burn from terminal area entry (typically 50NM from the runway) to the gate.
Benefits
Reduction of fuel burn / CO2 emissions for the approach and taxi phase.
Partners
- Main contributors: DSNA (Paris CDG and DTI), Thales and EUROCONTROL
- Reviewers: Air France, NATS and Swiss
Status
AMAN prototype available (Maestro by Thales), pending decision for trials or deployment by Paris CDG.