Periodic Reporting for period 2 - BEACON (Behavioural Economics for ATM Concepts)
Periodo di rendicontazione: 2021-07-01 al 2022-12-31
The project built on the concept of User-Driven Prioritisation Process (UDPP) developed by EUROCONTROL. UDPP is a simple way for airlines to avoid impact of massive delays on their fleet, by reordering their own flights when an Air Traffic Flow Management regulation hits them. Each delayed flight has a cost for the airline, but this cost is highly contextual, depending on the type of flight, number of passengers, even time of the day. Airlines are nowadays able to protect their most important flights (i.e. the most costly) by swapping their slots with slots assigned to less important flights.
BEACON studied the feasibility of extending UDPP to allow more efficient prioritisation processes, in particular allowing slots to be exchanged between airlines. For example, low-volume airlines (with a small number of flights) cannot exchange slots among themselves, but may hold slots that are of real value to other companies. Exchanging these slots against a potential compensation should thus allow for a more efficient system from the total cost point of view.
Different types of mechanisms and credit systems were tested in BEACON. On top of their overall efficiency, a special attention was paid to issues of fairness and equity, i.e. how different airlines are impacted. The mechanisms were tested using two models that can capture network effects in the airline cost structure, an important issue when it comes to their cost-minimisation decisions.
To properly capture the actors' behaviours, BEACON made use of behavioural economics. Non-rational behaviours like loss-aversion and hyperbolic discounting were embedded in the models in order to take them into account in the design of the new prioritisation mechanisms right from the start. Indeed, decisions featuring deviations from rationality may change the way the mechanisms behave with respect to a perfect rational baseline. Behavioural experiments were carried out in order to calibrate these effects, in particular using human-in-the-loop simulations.
BEACON showed that behavioural and gaming aspects may have a very important role in this context, severely decreasing the economic efficiency of inter-airlines mechanisms. The project proved that approximations and possible errors in the calculation of flight cost functions are crucial, and their effects are underestimated in the current research. Moreover, the project showed that designing an efficient inter-airline mechanism that can do better than an intra-airline one is difficult, due to the combined effects of gaming and behavioural effects, issues of approximations and errors, and relatively high level of cost efficiency of intra-airlines optimisations, less prone to these various issues. Finally, the project showed that equity issues are far from trivial, and that a consensus is needed on what the actual goal of UDPP mechanisms is in terms of economic efficiency and/or equity. BEACON increased the understanding on what Behavioural Economics can add to ATM concepts elaboration and validation methodologies and deepened and broadened the concepts of prioritisation in ATM beyond UDPP and their potential impacts on Network performance. It formed a guideline and numerous recommendations for future research in this field, both from a theoretical and a practical point of view.
BEACON studied the feasibility of extending UDPP to allow more efficient prioritisation processes, in particular allowing slots to be exchanged between airlines. For example, low-volume airlines (with a small number of flights) cannot exchange slots among themselves, but may hold slots that are of real value to other companies. Exchanging these slots against a potential compensation should thus allow for a more efficient system from the total cost point of view.
Different types of mechanisms and credit systems were tested in BEACON. On top of their overall efficiency, a special attention was paid to issues of fairness and equity, i.e. how different airlines are impacted. The mechanisms were tested using two models that can capture network effects in the airline cost structure, an important issue when it comes to their cost-minimisation decisions.
To properly capture the actors' behaviours, BEACON made use of behavioural economics. Non-rational behaviours like loss-aversion and hyperbolic discounting were embedded in the models in order to take them into account in the design of the new prioritisation mechanisms right from the start. Indeed, decisions featuring deviations from rationality may change the way the mechanisms behave with respect to a perfect rational baseline. Behavioural experiments were carried out in order to calibrate these effects, in particular using human-in-the-loop simulations.
BEACON showed that behavioural and gaming aspects may have a very important role in this context, severely decreasing the economic efficiency of inter-airlines mechanisms. The project proved that approximations and possible errors in the calculation of flight cost functions are crucial, and their effects are underestimated in the current research. Moreover, the project showed that designing an efficient inter-airline mechanism that can do better than an intra-airline one is difficult, due to the combined effects of gaming and behavioural effects, issues of approximations and errors, and relatively high level of cost efficiency of intra-airlines optimisations, less prone to these various issues. Finally, the project showed that equity issues are far from trivial, and that a consensus is needed on what the actual goal of UDPP mechanisms is in terms of economic efficiency and/or equity. BEACON increased the understanding on what Behavioural Economics can add to ATM concepts elaboration and validation methodologies and deepened and broadened the concepts of prioritisation in ATM beyond UDPP and their potential impacts on Network performance. It formed a guideline and numerous recommendations for future research in this field, both from a theoretical and a practical point of view.
First, the consortium discussed potential new mechanisms, as well as their scope. With the help of a workshop involving the advisory board, the consortium defined three types of mechanisms of interest: a centralised optimiser, using input similar to the current UDPP, a centralised credit mechanism, and a primary auction mechanism. These three mechanisms were selected based on criteria of expected efficiency, ease of use, and potential drop of efficiency due to behavioural effects.
Second, the project published an update on the previous state-of-the-art document on the cost of delay for airlines. Indeed, the cost function drives all the decisions of the airlines in this system. An accurate, up to date, cost of delay function is thus needed to have reliable results. Besides the update, the new report included additional considerations not featured in the previous version, like curfew costs.
Third, a survey was disseminated in order to collect data on the behavioural effects that may be expected from flight dispatchers in the mechanisms. The survey had a very poor response rate, and thus a literature review was performed to estimate some parameters of the decision-making process.
Fourth, modelling activities took place in two threads: a small-scale model aimed at testing the new mechanisms quickly with synthetic data, and a large-scale model (Mercury) able to compute a wide variety of KPIs in more specific situations. Complex behavioural rules were implemented for the agents within the model.
Fifth, a Human Machine Interface was developed to allow participants to interact with Mercury in order to solve the flight ordering in a regulation when it appeared in the simulation. Human-in-the-loop simulations were organised, with participants playing the role of flight dispatchers tasked with prioritisation of their flights during an ATFM regulation.
Finally, results were produced with fast-time simulations. The project showed the disruptive potential of gaming and behavioural effects from agents, explored issues related to the approximation of the cost functions, and produced estimations of economic efficiency and fairness of several mechanisms. General conclusions about the economic efficiency of mechanisms and their fairness was produced, highlighting the difficulty to design inter-airline allocation mechanisms that are better than intra-airline ones.
Second, the project published an update on the previous state-of-the-art document on the cost of delay for airlines. Indeed, the cost function drives all the decisions of the airlines in this system. An accurate, up to date, cost of delay function is thus needed to have reliable results. Besides the update, the new report included additional considerations not featured in the previous version, like curfew costs.
Third, a survey was disseminated in order to collect data on the behavioural effects that may be expected from flight dispatchers in the mechanisms. The survey had a very poor response rate, and thus a literature review was performed to estimate some parameters of the decision-making process.
Fourth, modelling activities took place in two threads: a small-scale model aimed at testing the new mechanisms quickly with synthetic data, and a large-scale model (Mercury) able to compute a wide variety of KPIs in more specific situations. Complex behavioural rules were implemented for the agents within the model.
Fifth, a Human Machine Interface was developed to allow participants to interact with Mercury in order to solve the flight ordering in a regulation when it appeared in the simulation. Human-in-the-loop simulations were organised, with participants playing the role of flight dispatchers tasked with prioritisation of their flights during an ATFM regulation.
Finally, results were produced with fast-time simulations. The project showed the disruptive potential of gaming and behavioural effects from agents, explored issues related to the approximation of the cost functions, and produced estimations of economic efficiency and fairness of several mechanisms. General conclusions about the economic efficiency of mechanisms and their fairness was produced, highlighting the difficulty to design inter-airline allocation mechanisms that are better than intra-airline ones.
BEACON tested promising new mechanisms for UDPP. These mechanisms go beyond the current processes and represent more than an incremental improvement, paving the way for future concept like trajectory-based operations. This should translate into more efficient delay management procedures, decreasing uncertainty in the airspace and airline costs overall.
BEACON progressed beyond the state of the art by developing models which are, to the knowledge of the consortium, the first ones to include non-rational behaviours in air transportation. Thus, the project scope goes further than UDPP only, and paves the way for better modelling capabilities in general. BEACON delivered general conclusions on the deviations from rationality observed within the system, thus allowing further research projects to estimate more easily whether they need to include these behaviours in their modelling approach.
BEACON delivered two mobility models of the air transportation system. To the knowledge of the consortium, very few models are able to model passengers and flights to this level of detail, featuring a very detailed cost model helping airlines to take realistic decisions within the model. These models can be used as simulation-based validation platforms in the future, including human-in-the-loop simulations.
BEACON produced an update of the cost of delay of airlines document (which is a de facto industry standard in Europe). This updated document includes the most detailed computations of costs at an ECAC level, in particular costs related to passengers, curfews, and turnarounds. This updated document can be used by any entity aiming to put a cost on the delays incurred by flights in Europe.
BEACON progressed beyond the state of the art by developing models which are, to the knowledge of the consortium, the first ones to include non-rational behaviours in air transportation. Thus, the project scope goes further than UDPP only, and paves the way for better modelling capabilities in general. BEACON delivered general conclusions on the deviations from rationality observed within the system, thus allowing further research projects to estimate more easily whether they need to include these behaviours in their modelling approach.
BEACON delivered two mobility models of the air transportation system. To the knowledge of the consortium, very few models are able to model passengers and flights to this level of detail, featuring a very detailed cost model helping airlines to take realistic decisions within the model. These models can be used as simulation-based validation platforms in the future, including human-in-the-loop simulations.
BEACON produced an update of the cost of delay of airlines document (which is a de facto industry standard in Europe). This updated document includes the most detailed computations of costs at an ECAC level, in particular costs related to passengers, curfews, and turnarounds. This updated document can be used by any entity aiming to put a cost on the delays incurred by flights in Europe.