Information Technology Reference
In-Depth Information
Tabl e 2.
Numerical results
Flights On-Pier Remote Resolved
in total
[%]
[%]
conflicts
Model
680
95
5
34
Real
626
92
8
0
grouped, with each of the groups introduced in previous sections being assigned
a different marker, so that it is easier to observe how the conflict constraint
changes the allocation and resolves possible conflicts. All of the flights that have
been assigned to the group of gates called GR1 are displayed as chains of squares
(
, effectively rectangular bars), GR2 as chains of x-shaped markers (
×
), GR3
as chains of star-shaped markers (
∗
), GR4 as chains of triangles (
)andGR5as
chains of dots(
). The remote allocations are also shown on the charts as chains
of crosses (+). The negative gate numbers symbolise remote gates. Each remote
allocation has been allocated to a different remote gate, so that it is easier to see
how many remote allocations there are. Note that in practice the towed flights
can be excluded from the conflict constraint thanks to the flexible implementa-
tion of towing. Since no strict towing time is assumed for each tow, the potential
conflicts can be resolved by controllers by postponing or speeding up some tows
if necessary. Four important observations can be made based upon the graph: (1)
The number of conflicting allocations is reduced. (2) The real solution has fewer
towed flights, since the long flights are kept on the gates in most cases, resulting
in lack of gates available for flights more often, and therefore: (3) the remote
stands having to be used more often. (4) The size of the time gaps between
allocations is much bigger in the model solutions.
An important objective of the presented model is maximisation of time gaps
between allocations (increasing the robustness of the solution). Figure 5 is in-
troduced in order to show the distribution of the gap sizes in the allocation
obtained for the full model (a) and for the real situation (b). The smallest gap
in the model solution is 90 minutes while in the real situation there were many
gaps which were smaller than that. The numerical results obtained for five days
are presented in Table 2. The first column of Table 2 shows the total number of
flights in the five days, the second one shows the percentage of on-pier alloca-
tions, the third the percentage of the remote allocations and the last the number
of conflicts which have been resolved by the introduction of the new constraint.
There are fewer remote allocations in the generated solution and 34 conflict
constraints have been resolved. However, it can be observed based upon the
total number of flights that the model introduces more tows into the solution,
which may not be good, especially at very busy periods. Another drawback of the
presented solution method is the calculation time, since it took in total over three
days to calculate the five days of allocations and, therefore, heuristic methods
will be considered now that the effects of the model have been established.
•
