Introduction Runways delay analysis Runways scheduling integration Results Conclusion. Raphaël Deau, Jean-Baptiste Gotteland, Nicolas Durand

Midival Airport surface management and runways scheduling ATM 2009 Raphaël Deau, Jean-Baptiste Gotteland, Nicolas Durand July 1 st, 2009 R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 1 / 26

Contents 1 Introduction 2 Runways delay analysis 3 Runways scheduling integration 4 Results 5 Conclusion. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 2 / 26

1 Introduction 2 Runways delay analysis 3 Runways scheduling integration 4 Results 5 Conclusion. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 3 / 26

Introduction A-SMGCS requirements for busy airports Arrival and Departure Management (AMAN, DMAN) Combined with an efficient Surface Management (SMAN) With aircraft ground conflicts anticipation and resolution A need for the future Air Traffic Management system Known difficulties Inaccuracy of the current DMAN systems Taxi times uncertainties Management of the CFMU take-off slots Complexity of the problem for mixed runways... R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 4 / 26

Previous work Simulations of the traffic of Roissy CDG airport R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 5 / 26

Previous work Simulations of the traffic of Roissy CDG airport For each traffic situation (every minutes), R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 5 / 26

Previous work Simulations of the traffic of Roissy CDG airport Predict aircraft trajectories (integrating speed uncertainties), R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 5 / 26

Previous work Simulations of the traffic of Roissy CDG airport Detect conflicts and identify clusters of conflicting aircraft, R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 5 / 26

Previous work Simulations of the traffic of Roissy CDG airport Solve each cluster, while optimizing a global criterion R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 5 / 26

Previous work (ICRAT08) ICRAT08 Compute some optimal runways scheduling in accordance with the traffic situation Include the scheduling in the global criterion to optimize for the resolution of each cluster of aircraft (SMAN) R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 6 / 26

Previous work (ICRAT08) ICRAT08 Compute some optimal runways scheduling in accordance with the traffic situation Include the scheduling in the global criterion to optimize for the resolution of each cluster of aircraft (SMAN) Improvement of the management of the CFMU take-off slots Significant decrease of other departures mean delay But the runways scheduling appears unstable: SMAN uncertainties on speeds, runways occupancy times... Disturbances due to extra delay in gates and taxiways area R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 6 / 26

Goals of this study R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 7 / 26

Goals of this study How to improve the SMAN and AMAN-DMAN coordination? In order to increase the accuracy of the DMAN process And/or to decrease aircraft delay R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 7 / 26

Goals of this study How to improve the SMAN and AMAN-DMAN coordination? In order to increase the accuracy of the DMAN process And/or to decrease aircraft delay Estimate a lower bound of the airport daily delay (part 1) Delay resulting from an optimal runways scheduling As a function of the AMAN-DMAN parameters Improve the runways scheduling accuracy (part 2) Compare the minimal delay to the one resulting from the SMAN Try to absorb some measured SMAN extra delay within some more appropriate runways scheduling R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 7 / 26

1 Introduction 2 Runways delay analysis 3 Runways scheduling integration 4 Results 5 Conclusion. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 8 / 26

Optimal runways sequences Goal: estimate a lower bound of the airport daily delay R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 9 / 26

Optimal runways sequences Goal: estimate a lower bound of the airport daily delay Principle Build an optimal aircraft daily sequence for each runway, Taking into account the minimal taxi time of each aircraft But ignoring aircraft circulation difficulties Method A classical scheduling problem That can be solved with deterministic methods (Branch&Bound) Applied to the successive time periods of the day R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 9 / 26

Optimal runways sequences Goal: estimate a lower bound of the airport daily delay Principle Build an optimal aircraft daily sequence for each runway, Taking into account the minimal taxi time of each aircraft But ignoring aircraft circulation difficulties Method A classical scheduling problem That can be solved with deterministic methods (Branch&Bound) Applied to the successive time periods of the day Studied parameters Anticipation time, landing times flexibility, scheduling margins R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 9 / 26

Influence of the anticipation time 85 sec. Departure mean delay 41 sec. 36 sec. 36 sec. Complete simulations 5 minutes 10 minutes 30 minutes Optimal AMAN DMAN scheduling (anticipation time) Optimal runways scheduling is twice less penalizing than the whole airport constraints The surface management can be improved R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 10 / 26

Influence of the anticipation time 85 sec. Departure mean delay 41 sec. 36 sec. 36 sec. Complete simulations 5 minutes 10 minutes 30 minutes Optimal AMAN DMAN scheduling (anticipation time) Optimal runways scheduling is twice less penalizing than the whole airport constraints The surface management can be improved An anticipation time of 10 minutes seams to be enough but not for the ATM systems (approach sectors, CDM, etc.) R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 10 / 26

Influence of the flexibility of landing times 300 Departures mean delay (seconds) 250 200 150 100 Departure delay on mixed runway 27L 50 0 ±10 ±20 ±30 ±40 ±50 ±60 Landing times flexibility (seconds) R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 11 / 26

Influence of the flexibility of landing times 300 Departures mean delay (seconds) 250 200 150 100 Departure delay on mixed runway 27L 50 0 ±10 ±20 ±30 ±40 ±50 ±60 Landing times flexibility (seconds) Necessity of this flexibility for runways operated in mixed mode ±30 sec. seams an interesting value R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 11 / 26

Influence of a margin in the scheduling 110 Departures mean delay (seconds) 100 90 80 70 60 50 40 30 0 2 4 6 8 10 12 14 16 Scheduling margin (seconds) R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 12 / 26

Influence of a margin in the scheduling 110 Departures mean delay (seconds) 100 90 80 70 60 50 40 30 0 2 4 6 8 10 12 14 16 Scheduling margin (seconds) Significant increase of the delay with the margin No benefit when the margin is over 10 seconds Difficult compromise: scheduling accuracy vs departures delay R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 12 / 26

First conclusions at Roissy CDG airport Aircraft circulation difficulties can double the departures delay The flexibility of the landing times (±30 sec.) is necessary for the mixed runway Adding margins in the scheduling is not a good solution The sequencing system must keep some flexibility The SMAN-DMAN coordination is the key point R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 13 / 26

1 Introduction 2 Runways delay analysis 3 Runways scheduling integration 4 Results 5 Conclusion. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 14 / 26

The problem to solve Deal with some uncertainties concerning Aircraft taxiing speeds Off-block times R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 15 / 26

The problem to solve Deal with some uncertainties concerning Aircraft taxiing speeds Off-block times Fit as well as possible to the optimized runways sequences despite of the taxi delay Manage the low anticipation time of the SMAN system Aircraft must be put in the right order soon enough What is the right order for aircraft of different runways? R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 15 / 26

Operational concepts Assign priority levels to the aircraft Favour departures constrained by a CFMU take-off slot Favour departures on arrivals Consider the aircraft that are ahead of their scheduled take-off slot Control departures off-block times, to ensure A strategic number of aircraft in front of the runway The application of the take-off slots required by the CFMU R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 16 / 26

Strategic off-block times Goal: dispatch correctly the departure traffic. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 17 / 26

Strategic off-block times Goal: dispatch correctly the departure traffic Control the off-bloc time of each aircraft If a departure i is planned before j in the runway sequence, Ensure that the aircraft i won t be in front of j when leaving the gate By delaying the off-block time of j if needed. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 17 / 26

Strategic off-block times Goal: dispatch correctly the departure traffic Control the off-bloc time of each aircraft If a departure i is planned before j in the runway sequence, Ensure that the aircraft i won t be in front of j when leaving the gate By delaying the off-block time of j if needed R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 17 / 26

The simulation loop Situation studied at time t Prédiction Traffic prediction Runways sequences optimisation Conflicts resolution Manoeuvers to apply during the next resolution Situation studied at time t + Traffic prediction (...) R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 18 / 26

The simulation loop Situation studied at time t Prédiction Traffic prediction Runways sequences optimisation Conflicts resolution Manoeuvers to apply during the next resolution Situation studied at time t + Traffic prediction (...) Which method is the best to solve the ground conflicts? R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 18 / 26

Sequential surface management Method description Aircraft are classified by priority levels Consider each aircraft one after the other Find a trajectory without conflict with the previously considered aircraft Change the priorities when there is no solution for an aircraft Initial classification Given by the runways scheduling Except for the already landed arrivals (lower priority). Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 19 / 26

Sequential surface management Method description Aircraft are classified by priority levels Consider each aircraft one after the other Find a trajectory without conflict with the previously considered aircraft Change the priorities when there is no solution for an aircraft Initial classification Given by the runways scheduling Except for the already landed arrivals (lower priority) Advantage: simple, intuitive Disadvantage Not optimal Hardly depending on the chosen classification. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 19 / 26

Surface management with an hybrid GA Main idea Find the best aircraft classification associated with the best paths. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 20 / 26

Surface management with an hybrid GA Main idea Find the best aircraft classification associated with the best paths Chromosomes (variables) [(prio 0, path 0 );...; (prio n, path n )] Fitness (criterion) Apply the sequential method with the given priorities and paths Get the number of aircraft in conflict or the global criterion if none Criterion: balance between delay and CFMU slots deviations. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 20 / 26

Surface management with an hybrid GA Main idea Find the best aircraft classification associated with the best paths Chromosomes (variables) [(prio 0, path 0 );...; (prio n, path n )] Fitness (criterion) Apply the sequential method with the given priorities and paths Get the number of aircraft in conflict or the global criterion if none Criterion: balance between delay and CFMU slots deviations GA convergence acceleration Using the partial separability of the fitness A local fitness per aircraft. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 20 / 26

1 Introduction 2 Runways delay analysis 3 Runways scheduling integration 4 Results 5 Conclusion. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 21 / 26

Departures delay 85 sec. Departure mean delay 65 sec. 60 sec. 36 sec. Older simulations Sequential Hybrid GA Optimal scheduling R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 22 / 26

Departures delay Roissy airport 200 Sequential method without runways sequences optimisation Sequential method with runways sequences optimisation Hybrid GA with runways sequences optimisation Departures mean delay (seconds) 150 100 50 0 0 5 10 15 20 25 30 35 40 Number of active aircraft by time periods of ten minutes R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 22 / 26

Arrivals delay 160 140 Roissy airport Sequential method withtout runways sequences optimisation Sequential method with runways sequences optimisation Hybrid GA with runways sequences optimisation Arrivals mean delay (seconds) 120 100 80 60 40 20 0 0 5 10 15 20 25 30 Number of active aircraft by time periods of ten minutes R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 23 / 26

Deviations to the targeted take-off slots 400 350 Take off slots deviation Sequential method GA method 300 Number of aircraft 250 200 150 100 50 0 4 3 2 1 0 1 2 3 4 5 6 7 8 9 Deviation (min) R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 24 / 26

1 Introduction 2 Runways delay analysis 3 Runways scheduling integration 4 Results 5 Conclusion. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 25 / 26

Conclusions SMAN and AMAN-DMAN synchronization Is necessary to increase the accuracy of the DMAN system Can lead to absorb a significant part of the SMAN delay Allows to reduce the global airport delay during traffic peaks Optimization methods can help to perform this synchronization Assigning some strategic off-block times to the departures Managing taxiing aircraft in accordance with the runways scheduling Without penalizing too much the arrivals (GA method) R. Deau, J-B. Gotteland, N. Durand ()Airport SMAN and runways scheduling July 1 st, 2009 26 / 26