to a dynamic simulation problem.
The optimisation process is performed in zero simulation time. Thus, the flight that fired the optimisation process and all the other actions in the airport simulation are paused while the new flight-to-gate assignment schedule is being calculated. The optimisation process is then performed for an ideal situation, where no flights are delayed for any reason. This is done since without the simulation, where passengers and aircraft are actually moving on paths, there is no way to know if an aircraft will be delayed. When the optimisation process is finished the simulation is continued. As soon as there is a deviation from the ideal circumstances, the metaheuristic optimisation is repeated.
11.2 Frequency of performing the optimisation process
The optimisation process, using the CE method, is performed in each of the following cases:
• for the first flight
• for every 25th flight after the previous flight for which it was performed
11.2 Frequency of performing the optimisation process
• each time an aircraft is delayed
• each time the gate for which an arriving aircraft had to wait becomes avail-able
If the arriving aircraft has a priority equal to the value of the priority of the flight that started the execution of the previous optimisation process plus 25, it is time to repeat the process. Each time the optimisation process is executed, i.e.
the CE method is applied, it is executed for the flight currently arriving at the airport as well as the 50 flights scheduled to arrive after that.
If the scheduled arrival time of the aircraft is less than the actual arrival time, i.e. if the aircraft arrives late, then the optimisation process is also performed.
However, in this study, the assumption is made that the only delays at the airport are caused by the airport itself. For example, because of an aircraft not being able to leave a gate at the scheduled departure time of the flight and before the next flight has to occupy that gate. Then the second aircraft is delayed. No arriving aircraft is delayed because of factors outside the control of the airport.
This assumption can be made since the same circumstances are used in every model and layout, which means they can be compared.
When the aircraft arrives, it already has a gate to which it was assigned in the last execution of the optimisation process, since this process assigns the current arriving flight as well as the 50 flights arriving after that to gates. This is different from the models without the metaheuristic optimisation where the flights are only assigned to gates once they arrive at the airport (Rule 1 and Rule 2). If the gate to which the aircraft was assigned in the previous execution of the optimisation process is not available, i.e. it is still occupied by the previous aircraft assigned to it, the aircraft cannot go to that gate and must thus be assigned to a new gate.
This means that the previous aircraft did not leave the gate at the scheduled departure time, because of a delay.
Here, the optimisation process is executed again, and now the flight cannot park at the gate to which is was originally assigned, since it is known that even though this gate is supposed to be available, it is not. If this is the case, the model must keep track of all the gates to which this first flight cannot be assigned in this run of the optimisation process. This is done by using a variable WrongGate(x)
11.2 Frequency of performing the optimisation process
where x is the gate number. If the value of WrongGate(x) equals one, the current arriving aircraft cannot be assigned to it, and if the value is zero, the aircraft can be assigned to it. When the aircraft is finally assigned to a gate that is available and to which it can start taxiing right away, the variable WrongGate(x) for x equals one to the number of gates at the airport (122) is again set to zero. This must be done so that when the next flight arrives for which the optimisation process must be performed, it can be assigned to all the available gates. If this value is not set to zero, then the first arriving flight in the next execution of the optimisation process will not be allowed to park at this gate even though it may have become available in the mean time.
When the optimisation process is performed due to the case described above, i.e. a gate that is supposed to be available is not available, there is no way for the model to know how long the aircraft currently occupying the gate will be delayed.
Thus, the gate is left out of the equation for a selected duration (15 minutes).
When this duration has expired, the gate can be used again. However, the gate may become available before these 15 minutes are over. When the optimisation process was performed, it was assumed that the gate would be occupied for 15 more minutes, thus all the flights were assigned to gates while taking that into consideration. Now, after the optimisation process was performed, the gate be-comes available. The previous calculated flight-to-gate schedule is not as good as it could have been when the correct time for the gate to be used again, was used.
Thus, the optimisation process is executed again with the gate being available.