In a deterministic setting, the centralised concept was most effective in pre-departure deconfliction, due to its application of a regular hexagonal grid, and an effective implementation of a sliding window joint optimisation of flight intentions. This resulted in evenly-distributed traffic, mitigating bottlenecks, while maintaining good path efficiency.
As anticipated, the results indeed show that strategic, pre-departure deconfliction is not sufficient on its own. Especially unanticipated events (such as evaluated in the rogue drone scenarios) are left unresolved, but even in the deterministic scenarios, too many conflicts remain in flight. Compared to the decentralised concept, computational load is very high.
The addition of a tactical separation management component is expected to further improve the results of the centralised concept.
The ability to deconflict strategically before departure depends on the predictability (or achievability) of flights. Although wind and rogue aircraft scenarios were considered in the evaluation, it is acknowledged that these simulated uncertainties are not representative of the breadth and potential severity of factors impacting predictability in real operations. These uncertainties will have an impact on the efficacy of strategic deconfliction, and in the worst case may negate the observed benefits of the strategic planning method employed in the centralised concept.
Like the centralised concept, also in the hybrid concept, the application of strategic deconfliction provided an effective way of distributing traffic, and avoiding hotspots and bottlenecks. Compared to the centralised concept, the hybrid concept clearly shows the added benefit of tactical separation, achieving the lowest number of intrusions. These results are also partially attributed to the concept’s effective airspace design: The ring-road structure applied in free airspace resulted in a high degree of alignment, and helped increase safety by a significant amount.
The drawback of the ring-road strategy is seen in a large reduction in route efficiency, where route length was on average almost twice as long as the shortest route from origin to destination. Next to reduced efficiency this also caused a high number of inoperable trajectories (flight duration exceeding the endurance of the vehicle) for the hybrid concept. This penalty is reduced when equal weighting is applied to flights in free and constrained airspace, but only in low traffic densities: for higher densities the efficiency results of the unweighted hybrid concept converge with those of the weighted hybrid concept. In the ‘delay dispersion’ metric it is seen that variability of assigned delay goes up significantly at higher traffic densities. This is an indication that the First Come First Served strategy of the Hybrid planning algorithm is unable to distribute delay equitably between early and later planned flights. The rolling-window approach of the centralised concept shows, however, that alternative implementations (not strictly FCFS) can provide better results.
Compared to the centralised concept, the strategic algorithm of the hybrid concept was conservative, resulting in low capacity and reduced equitability at higher densities. A combination with a strategic algorithm similar to that of the centralised concept could bring improvements.
Similar to the centralised concept, the effectiveness of strategic deconfliction depends on predictability.
The decentralised concept only provided separation management on a tactical level, combined with a degree of flow control, and a centralised ‘traffic information service’. Departure delays were implemented by the individual agents, and were limited to a low value of five minutes. As a result, the decentralised concept provided the most equitable access to airspace, with the least amount of imposed delay, and the highest flight efficiency. The observed number of intrusions shows that this does not come at the cost of severely reduced safety: results are on par with the centralised concept, and while the hybrid concept resulted in fewer intrusions, this is partially due to an effective airspace structure that can also be applied to other concepts.
In the decentralised concept, horizontal (re-)distribution of traffic was implemented through individual avoidance of hotspots: traffic density was observed centrally, from which information was provided to the individual agents so that they could re-route around dense areas. This is a reactive approach, which inherently still produces denser areas of traffic.
Tactical separation algorithms are almost without exception exclusively reactive: they act only in response to a detected conflict. In an airspace where conflicts can be considered as isolated events this is not a problem. This assumption however does not hold in dense and constrained airspace. The effect thereof can be seen in the high number of secondary conflicts observed for the decentralised concept in constrained airspace. While many of these conflicts can be seen as implicit coordination of intent, lowering the number of conflicts is nevertheless desirable. Pro-active strategies, where agents and neighbouring vehicles act to improve manoeuvrability without first detecting a conflict, can be a way to reduce the high number of knock-on conflicts.
In any purely distributed system, the possibility exists of destabilising emerging patterns, causing unsafe situations. Avoiding these will require (centrally imposed) safeguards in terms of, for instance, capacity limits and flow restrictions (e.g., geovectoring).
Summarising the above, several key observations can be made from the results:
- The degree of centralisation has not shown to be a main contributing factor to safety. The results in terms of the number of intrusions for strategic deconfliction and tactical separation are similar, even when the simulation is solely deterministic.
- Airspace structuring emphasising alignment can bring significant reduction in conflicts and intrusions. The ring-road structure of free airspace in the hybrid concept has shown to be an effective means of imposing alignment in the airspace and reducing likelihood of conflict.
- To mitigate the likelihood of secondary conflicts, tactical resolution should consider pro-active (the ability to act without first detecting a conflict) actions, in addition to reactive (resolution actions in response to a detected conflict) actions, especially for decentralised implementations.
- Airspace structuring and strategic routing should consider efficiency impact. Here, increased path length will be limited by endurance and economic viability.
Access, equity and capacity:
- Strict pre-planning can diminish accessibility and capacity.