摘要: |
This paper evaluates, by means of fast-time simulation, performance of a candidate system
for autonomous air traffic management. Advancing towards autonomy in air traffic
management may be necessary in order for new air vehicle types such as electric Vertical Take
Off and Landing (eVTOL) to operate safely and efficiently in airspace shared with
conventional traffic. To account for uncertain prediction, autonomous air traffic management
was divided into two integrated and coordinated subsystems: strategic scheduling, performed
at predeparture, and tactical conflict detection and resolution, performed throughout the
flight. The conflict detection and resolution subsystem contained a second tactical scheduling
function that applied to flights operating in the airspace near the destination airport. This
paper compares and contrasts the two subsystems and uses fast-time simulation to
demonstrate the comparisons. A scenario of 54 flights inbound to Newark Liberty
International Airport was simulated multiple times with different parameters. The scenario
was created using flight plans recorded from the National Airspace System on a low weather,
average traffic day in April 2018. Whereas the routes were not changed, the departure times
of the flights were modified to increase arrival rates at the Newark runway and arrival meter
fixes. Results of the simulations showed that the autonomous air traffic management system
was able to safely manage the traffic, even with prediction uncertainty. In addition, they
showed the importance of including flight holding maneuvers, in addition to path stretching,
in conflict detection and resolution and of coordinating strategic and tactical scheduling.
Finally, a tradeoff between absorbing the delay calculated by strategic scheduling on the
ground versus in the air showed that taking most of the delay on the ground is cost effective
for a simple idealized cost function. However, taking a little of the delay in the air prevented
throughput on the runway from dropping for short periods due to trajectory prediction
uncertainty. |