Domino workshops – First results and model validation
Author: Luis Delgado
Obtaining feedback from stakeholders and users is critical to validate and improve the model and its capabilities. Domino has participated in a workshop and organised another one to share and discuss its modelling approach, the new network metrics and the initial results.
First, Domino was presented at the EUROCONTROL Experimental Centre in Brétigny to airspace users who help us to validate and improve some of the model assumptions on AOC behaviour. This workshop was crucial to gain better insight on airline and flight behaviour. The views by the airlines present covered a wide range of type of operations as the workshop was attended by full (hub operators) carriers, regional airlines and low cost carriers. We are pleased that most of the assumptions in our model were shared by airlines but we also learned a lot!
Then, a dedicated full day workshop was conducted at the SESAR SU offices in Brussels focused to ATM experts. Don’t worry if you missed the workshop, you can still access the material used during the day by downloading it from our website! The contributions across the table will help us to adapt the model and prioritise the next activities for Domino. It was a good day also to share some of Domino modelling and metrics capabilities and to discuss how this can be taken forward. In particular, we recognise the need to identify how network metrics can be used from operational perspective. Stay tuned for future details on this!
The model was able to generate and capture unexpected behaviours which emerged from the agents behaviours. For example, higher gate-to-gate time for flights are observed when the AOC considers the tactical optimisation of the trajectory (4DTA mechanism). In some cases (Level 2 implementation), flights are actively delayed by the AOC to save fuel and hence cost. See how the speed selected (0-MRC, 1-MMO) vary for different scenarios and level of implementation and how the delay that will be recovered can in some cases be negative (i.e., generating delay) in order to prioritise fuel savings.
Domino also presented in the workshop the importance of considering metrics beyond averages, as some of the mechanisms have an impact which is reflected on the tail distribution of values. Below, for example, the average delay of the use of the Flight Arrival Coordinator mechanism (FAC) (which has a small impact) compared with its performance on the tail of the delay distribution (where the number of flight with extreme delay are reduced).
Finally, network metrics (centrality and causality) are relevant to understand the non-direct effect of the mechanism in the system both for flights and for passengers and for delay and costs. As mentioned, we’ll keep working on the interpretation of these metrics considering the feedback from the workshops.
New blog entries will present some of these results with more detail in the future focusing on the different mechanisms.
The results of the workshops and how the feedback will be incorporated in the project will be made available in D6.3 Workshop results summary, the future evolution of Domino for its final results will be reported in D3.3 Adaptive case studies description.