Authors:
Joana Leite
1
;
2
;
Rafael Guedes
3
and
Diogo Queirós
3
Affiliations:
1
Polytechnic Institute of Coimbra, Coimbra Business School, Quinta Agrícola - Bencanta, 3045-231 Coimbra, Portugal
;
2
CMUC - Center for Mathematics, University of Coimbra, 3001-501 Coimbra, Portugal
;
3
Faculty of Engineering, University of Porto, 4200-465 Porto, Portugal
Keyword(s):
Aircraft Sequencing Problem, Aircraft Landing Problem, Mixed Integer Programming, Constraint Programming, Decomposition Heuristic, Parallel Processing.
Abstract:
In this paper, we revisit the Aircraft Sequencing Problem (ASP), which consists of scheduling aircraft landings respecting a pre-determined time window and separation criteria. ASP has several versions, with the static single runway being the one with the longer solving times for the benchmark instances, for both mixed integer programming (MIP) and constraint programming (CP) implementations. We considered this version of the problem and addressed the possibility of using parallel processing to solve it. For this purpose, we developed a heuristic for splitting the instances, which always guarantees a feasible solution that is the optimal solution if a set of conditions is satisfied. The splitting allows for parallel processing, and opens the possibility of using the best method to solve each subset of the partition obtained. To explore this feature, we also analysed the performances of MIP and CP implementations and constructed a measure to point to the fastest one. For the benchmark
instances, the results show a time reduction over 50%, in the cases the optimal solution is known, and an improvement of over 12% on the value of the best-known feasible solution, in the cases the optimal solution is not known and running time has to be limited.
(More)