We have studied how different disturbances (operational, weather, airport related..) affect aviation, analysing the system resilience
Resilience2050 is a European FP7 aviation collaborative research project. Resilience2050 is coordinated by Innaxis, and there are six participants from five different countries.
How to measure resilience
Resilience is the property of a system to recover from unexpected disturbances. In aviation, many disturbances can affect air traffic. If the system is not resilient, these "disturbances" may cause what we call "perturbations". For example: weather issues (disturbances) frequently generate delay (perturbations). In order to know how resilient a system we need to be able to measure this property: are disturbances (bad weather, operational issues) generating perturbations (delay, stress)? Is the system able to avoid them? How "much"?
This project defines a resilience metric that quantifies the difference between disturbed states (those affected by disturbances) and reference states (those not affected). Resilient systems easily recover from disturbances, hence showing no deviation if compared with reference-state KPIs. That deviation has been measured in terms of the slope difference between linear regression fits:
- Several graphs like the following are generated (one graph per disturbance kind, one per airports/route). Within each graph, each dot represents data of one single flight. The X axe represents the first delay figure (departure delay at airport A). The Y axe measures the delay after its propagation (arrival delay at airport B).
- Regression is calculated for the reference flights (those with no disturbances) and for the disturbed ones (those affected by disturbances).
- If linear regression is used, the difference between the slope in each group represents the system resilience behaviour, or in other words, how it reacts after a given disturbance. More details available in Project deliverables section.