Information Technology Reference
In-Depth Information
Like many real world problems the GAP has several objectives. The pre-
sented model includes four objectives, which mirror the interests of the two sites
involved in this problem: an airport and an airline. From the perspective of an
airport one of the most important objectives is to obtain a robust allocation,
i.e. an allocation with gaps between flights which are allocated to the same gate
are as large as possible. It allows the absorption of minor delays and maintains
smoother airport operation. Another airport related objective is to use the gate
space effectively, i.e. to match the size of the aircraft. The aim is to avoid using
unnecessarily large gates, keeping the larger gates free since these have more
flexibility for use by flights which have to be moved on the day of operation.
Airline related objectives involve specific gate preferences and avoiding remote
allocation where possible. The multi-objective character has been modelled here
using a weighted sum of several objectives. This is a commonly used procedure
which results in the optimal solution potentially being strongly dependent on the
chosen weights. This is sucient for the aims of this paper. However, different
multi-objective approaches [3], which allow the analysis of the trade-off and/or
tuning of the weights for the various objectives will be investigated in future.
3 Previous Literature
The multi-objective nature of the problem has been gradually acknowledged by
researchers working on this problem. Initially research focused mainly on various
passenger comfort objectives. For example, Haghani and Chen [10] and Xu [15]
minimised the distance a passenger walks inside a terminal to reach a departure
gate. Yan and Hou [16] introduced an integer programming model to minimise
the total passenger walking distance and the total passenger waiting time.
Ding et al. [6,7] shifted the research interest slightly from passenger comfort
to airport operations and solved a multi-objective IP formulation of the GAP
with an additional objective: minimization of the number of ungated flights.
Dorndorf et al. [8] went further and claimed that not enough attention has
been given to the airport side when solving the GAP. They focused on three
objectives: maximization of the total flight to gate preferences, minimization of
the number of towing moves and minimisation of the absolute deviation from
the original schedule. Perhaps surprisingly at the time, in the context of many
previous publications, they omitted the walking distance objective, arguing that
the airport managers do not consider it an important aspect of the GAP. Our
experience with airports indicates that this is probably correct, which is why the
objective function presented in Section 4 does not consider passenger walking
distance but aims instead to reduce the conflicts by the gates, to allocate gates
so that the airline and the size preferences are maximised and to ensure that the
time gaps between two adjacent allocations are large enough.
The constraints that we have modelled in Section 4 include other aspects of the
problem which have already been discussed by other researchers. Dorndorf et al.
discussed relative sizes of gates and aircraft, airline preferences and shadowing
restrictions [8,9]. Similarly the importance of the maximisation of time gaps
