AN EW S LOT A LLOCATION FOR ATFM Vu Duong † Fr´ ed´ eric Ferchaud †‡ Cyril Gavoille ‡ Mohamed Mosbah ‡ I NTRODUCTION Once the aircraft is in the air, it is supervised by ATCo 1 to guarantee safe separation. The capacity is defined according to ATCo workload, responsible of air traffic in a sector of airspace, simplified in a number of aircraft per hour. In Europe, the Central Flow Management Unit (CFMU) is responsible for the flow management in order to guarantee the capacity of the sectors. Air Traffic Flow Management (ATFM) is a service es- tablished with the objective of contributing to safe, orderly, and expedious flow of air traffic. The ATFM is divided into three phases[1]: • The strategic phase takes place between 18 month and seven days prior to the day of operation and includes research, planning and coordination activities. The output of this phase are the capacity plan for the following year. • The pre-tactical phase is applied during 6 days prior to the days of operation and consists of planning and coordination activities. The output is the ATFM Daily Plan. • The tactical phase is applied on the day of the oper- ation. This phase updates the daily plan according to the actuel traffic and capacity. The management of the traffic is achieved through slot allocation. In Europe, to realize a flight between two airports, a company must submit a flight plan to the CFMU. The flight plan contains the following informations : • aircraft identificator or call-sign, • departure airport, • arrival airport, • desired take-off time, • waypoint list (air road taken), • flight level associated with each waypoint. From flight plan information, the decizion to activate traffic balancing demand is to be made for the sectors where necessary (number of aircraft greater than the capacity). In accordance with the principle of the ”First filed-First Served”, the sytem in charge of the regulation, namely Computer Assisted Slot Allocation (CASA), extracts all the flights entering the specified airspace and sequences them into the order that they would have arrived at the airspace in absence of any restriction. On this basis, the Take-Off Time (TOT)) is calculated. The Calculated TOT (CTOT) is then transmitted to the †EEC : EUROCONTROL Experimental Centre ‡LaBRI : Laboratoire Bordelais de Recherche en Informatique 1 Air Traffic Controllers airlines concerned and to the control tower at the departure airport. In EUROCONTROL ATFM, the CFMU is in charge of the slot allocation for all flights and elaborates the daily planning of slots. A slot corresponds to a time window (-5 min, +10 min) of the CTOT during which the aircraft must take-off. Unfortunately, as a matter of fact, uncertain operational events (weather conditions, technical failure, waiting pas- senger...) occurs daily and disturb the CFMU planning, leading to safety problems and sub-optimally used capacity . We call these events operationnal aleas. When an aircraft can not take its allocated slot, we say it has got an alea. To accomodate the aleas, we introduce the notion of absorption areas (AA). An AA is a number of slots, left unfilled during the slot allocation process, allowing the absorption of such disturbances with least modification of the planning. Finding the best configuration of the AAs corresponds to balancing two complementary objectives: maximizing their effect and minimizing ”load loss”. Load loss correspond to the lost slots. The lost slots correspond to unused AAs or slots not taken by aircraft with aleas. We implemented AAs in a prototype ATFM simulator, called SIVOR 2 [2]. Our experimental results confirm that AAs could increase the throughput. Then we developed a theoretical approach to formally prove these results. We show that, whatever the rate of aleas is, the flow with AAs is always higher than without them. Another interest of AAs is that if we find an algo- rithm giving the best distribution of unfilled slots, then its implementation will neither change sector topologies, nor controller’s work nor the flight plans submission procedure. I. SIVOR A. Obtained results With SIVOR, we showed experimentally the utility of absorption areas. If an aircraft lost its slot, we must reallocate it. This aircraft could perturb the initial planning and can cause excess of sector capacity. So some other aircraft could be delayed. With SIVOR, we observed that whatever the rate of aleas is, absorption areas decrease the delays due to aleas, and sometimes also increase the throughput (see Figure 1). The results were obtained for a capacity of 15 aircraft per sector and per hour. On the Y-axis of the upper graph of Figure 1, the 0 value corresponds to a capacity of 15 for the pretactical planning, 1 to 14, 2 to 13, and 3 to 12. We 2 SImulateur de VOl en Route