Airspace Operations Lab @ NASA Ames

An optimization model for a scheduling problem for closely spaced parallel approaches has been formulated. It takes temporal, pairing, sequencing, separation route and grouping constraints into account. Simulations investigated possible advantages of advanced scheduling methods over first-come-first-served scheduling. Also, this study evaluated the performance differences between the computation of optimal solutions using mixed integer linear programming and computing solutions using genetic algorithms. The influence of the scheduling method, as well as the influence of varying the sizes of the pairing and estimated arrival time windows, have been investigated. A set of 20 aircraft, distributed over 30 minutes, was used as traffic data. Inputs to the model were the earliest and latest estimated arrival times, aircraft wake category, aircraft pairing group and route information. A schedule at a specific coupling point where aircraft are coupled for parallel approach was then computed. It is generally expected that closely spaced parallel approaches greatly enhance arrival throughput. The results of this study underpin this assumption. Further findings were: (1) for a sufficiently varied traffic mix, advanced scheduling methods can improve arrival throughput by approximately one parallel approach pair per half an hour compared to first-come-first-serve scheduling in visual meteorological conditions. Average delay can be reduced over first-come-first-serve scheduling in visual meteorological conditions by up to 36%. (2) schedules computed by an improved genetic algorithm are of similar quality as optimal solutions and can be made available after short computation times; (3) when minimizing makespan i.e., the arrival time of the last aircraft in a sequence, the size of the estimated arrival time window does not influence the characteristics of the computed schedules.