Runway Operations Optimization in the Presence of Uncertainties

Abstract

A Runway Planning optimization model has been developed as a part of a comprehensive suite of models for the optimization of airport surface traffic in the presence of uncertainty. The runway planning problem is formulated as a two-stage stochastic program where the first stage determines an aircraft weight class sequence and the second stage assigns individual aircraft to the sequence. Stochastic attributes include pushback delay, time spent on taxiway, and deviation from estimated arrival time. The computational study shows that, if the schedule is dense enough, there is a potential benefit of using the stochastic runway planner over first-come-first-serve planning policy or a deterministic runway planner.

Effect of Uncertainty on Deterministic Runway Scheduling

“…Although stochastic methods for addressing runway operations planning have been developed 13 , there is a gap in testing the benefits of the deterministic schedulers in the presence of uncertainty. This paper attempts to fill that gap by testing a deterministic scheduler under uncertainty, and a large spectrum of possibilities (including different traffic densities, aircraft mix, presence of miles-in-trail separation constraints, varying values of uncertainty) are considered.…”

Extended Aircraft Arrival Management Under Uncertainty: A Computational Study

“…Although a well-applied SAA method can guarantee good-quality solutions for the original problem, computing times are very often inappropriate for real-time implementation. Consequently, fast solution methods based on SAA were proposed in the literature related to aircraft scheduling using stochastic programming [11,13].…”

“…The case in which predicted operations times are known with certainty, called the deterministic case, has been thoroughly studied in the literature [4,7,8], while the case under uncertainty has less often been addressed. So far in the related literature, three main approaches to optimization under uncertainty were applied to ASSP: probabilistic [9,10], stochastic [11][12][13] and robust [14][15][16] approaches. Pioneer studies such as [9,10] mainly enriched deterministic models by probability constraints and/or by a probability objective-value function.…”

“…Pioneer studies such as [9,10] mainly enriched deterministic models by probability constraints and/or by a probability objective-value function. Stochastic programming models, including two-stage and multi-stage models, were proposed in [11][12][13].…”

“…Since solving a single SAA problem with a large scenario set may be time-consuming, [11,13] opt for solving several replications of an SAA problem with a limited number of scenarios. Upon optimization, they are left with a pool of near-optimal solutions to the original problem (as many solutions as replications).…”