| Resumo : |
The air transportation demand growth in the last decades has not been accompanied by the capacity growth of airport systems at the same rate. As a consequence, imbalances between demand and capacity have become increasingly frequent, leading to congestion and flight delays. Air traffic flow management seeks to mitigate the impacts of such imbalances and has become a fundamental function of air traffic management systems worldwide. At the tactical level, the development of aircraft sequencing decision support tools is key to optimize the use of the runway capacity and reduce delays. Towards this goal, this research presents an efficient solution approach for the aircraft sequencing problem to maximize runway utilization. The proposed method is based on a hybrid-simulated annealing metaheuristic with an Optimized Latin Hypercube Sampling approach for hyperparametrization. The hybrid-simulated annealing approach uses a heuristic called Closest Aircraft Sequence with Time Windows (CAS-TW) to produce the initial solution, which is inspired by the approaches to solve the traveling salesman problem and combines tour constructive and insertion modules. Through this initial solution, the simulated annealing metaheuristic with fixed penalization allows for the discovery of further efficient solutions in a small computational time. The proposed method is adapted with inviolable, hard constraints of time window and safety separations, and handles the cases where the triangle inequality does not automatically hold when both arrivals and departures are scheduled together. Although the fine-tuning of metaheuristic parameters can influence the quality of a solution, it is hardly performed due to the high computational effort for large experimental design. The present study proposes a simple and fast method for optimizing the metaheuristic parameter setting as an alternative for the traditional parameter tuning techniques. The results showed that the proposed method was quickly implemented, equitable, easy to use, and obtained good solutions. It was quickly implemented because the solutions were generated with a maximum computational time of 5 s; it was equitable because it respected time windows constructed to deviate little from the first come, first served original order; and the solutions generated a reduction of 386.59 s or 10.4% in total runway utilization times when compared to the FCFS method, allowing for increased runway capacity. The method could produce results that presented low dispersion levels from the best-found solution and that were stable through the possibility of constant hyper-parametrization. The method was developed to be used for the sequencing of arrival and departure aircraft of different types and sizes on a single runway or on a runway system that, due to the distance between runways, operates in a dependent manner. |