Deconfliction with Constraint Programming
Nicolas Barnier and Cyril Allignol
barnier@recherche.enac.fr,allignol@tls.cena.fr
ENAC – DTI/R&D
International Workshop on Constraint Technology for Air Traffic Control & Management
INO’08 12/02/2008
Outline
1 Introduction
2 Context
ATC and ATFM Ground-Holding
3 Deconfliction by Ground-Holding Model
Search and Optimization Results
4 Further Works
5 Conclusion
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 2 / 23
Introduction
Introduction
Congested European Sky
Traffic still growing by a yearly 5%
Increasing regulation delays due to en-route sector capacities Structural limits of the ATM system reached
Optimization of airspace structure and ATFM regulations EC Single European Sky (SESAR) / Episode 3 - WP3 Pre-tactical Deconfliction with Constraint Programming
Deconfliction by ground-holding
Highly combinatorial/disjunctive large scale problem Constraint Programming (CP) technology :
versatile modelling tool
side constraints incrementally added experiment with various search strategies
Feasibility stage : CP able to achieve optimality proof
Introduction
Introduction
Congested European Sky
Traffic still growing by a yearly 5%
Increasing regulation delays due to en-route sector capacities Structural limits of the ATM system reached
Optimization of airspace structure and ATFM regulations EC Single European Sky (SESAR) / Episode 3 - WP3 Pre-tactical Deconfliction with Constraint Programming
Deconfliction by ground-holding
Highly combinatorial/disjunctive large scale problem Constraint Programming (CP) technology :
versatile modelling tool
side constraints incrementally added experiment with various search strategies
Feasibility stage : CP able to achieve optimality proof
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 3 / 23
Context ATC and ATFM
ATC and ATFM
Objectives
1 Safety : maintaining aircraft separated
2 Efficiency : expedite the flow of traffic Layered Filters with Decreasing Time Horizon
1 Strategic (several months) : AirSpace Management (ASM), design of routes, sectors and procedures
2 Pre-tactical (a few days to a few hours) : Air Traffic Flow
Management (ATFM), sector openings and capacities, flow regulation by delaying and rerouting (Central Flow Management Unit)
3 Tactical (5-15 min) : Air Traffic Control (ATC), surveillance, coordination, conflict resolution
4 Emergency (< 5 min) : safety nets, ground-based (STCA, MSAW) and airborne (TCAS, GPWS)
Context ATC and ATFM
ATC and ATFM
Objectives
1 Safety : maintaining aircraft separated
2 Efficiency : expedite the flow of traffic Layered Filters with Decreasing Time Horizon
1 Strategic (several months) : AirSpace Management (ASM), design of routes, sectors and procedures
2 Pre-tactical (a few days to a few hours) : Air Traffic Flow
Management (ATFM), sector openings and capacities, flow regulation by delaying and rerouting (Central Flow Management Unit)
3 Tactical (5-15 min) : Air Traffic Control (ATC), surveillance, coordination, conflict resolution
4 Emergency (< 5 min) : safety nets, ground-based (STCA, MSAW) and airborne (TCAS, GPWS)
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 4 / 23
Context Ground-Holding
Ground-Holding
Pre-tactical Flow Regulation
Safest than handling the traffic while airborne Costly for airlines and passengers, snowball effect Sector Capacity and Regulation
Air Traffic Control Centres (ATCC) opening schedules : designed by experts (FMP)
Open sectors capacities : hourly entry rate
Regulation on flows crossing overloaded sectors : Computer Assisted Slot Allocation (CASA) at CFMU
CASA
Greedy algorithm : optimality, consistency
“First-come, first-served” questionable principle Operational setting, real-time updates
Context Ground-Holding
Ground-Holding
Pre-tactical Flow Regulation
Safest than handling the traffic while airborne Costly for airlines and passengers, snowball effect Sector Capacity and Regulation
Air Traffic Control Centres (ATCC) opening schedules : designed by experts (FMP)
Open sectors capacities : hourly entry rate
Regulation on flows crossing overloaded sectors : Computer Assisted Slot Allocation (CASA) at CFMU
CASA
Greedy algorithm : optimality, consistency
“First-come, first-served” questionable principle Operational setting, real-time updates
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 5 / 23
Context Ground-Holding
Ground-Holding
Pre-tactical Flow Regulation
Safest than handling the traffic while airborne Costly for airlines and passengers, snowball effect Sector Capacity and Regulation
Air Traffic Control Centres (ATCC) opening schedules : designed by experts (FMP)
Open sectors capacities : hourly entry rate
Regulation on flows crossing overloaded sectors : Computer Assisted Slot Allocation (CASA) at CFMU
CASA
Greedy algorithm : optimality, consistency
“First-come, first-served” questionable principle Operational setting, real-time updates
Context Ground-Holding
Slot Allocation with CP
Optimize upon CASA Solutions
SHAMAN : CP model over 30 min periods (CENA/RFM) ISA : CP and LP [Junker et al]
Marabout : sort constraint with FaCiLe to smooth the entry rate [ATM’01]
“Complexity” of Traffic
Relevance of sector capacity to model controller workload ? Discrepancies between planned schedule and actual openings More pertinent metrics w.r.t. real-time merge/split decision [Giannazza, Guittet 06]
Prior Opening Schedule Optimization Optimize upon FMP’s opening schedule
Multiple partitioning problem, possibly with side transition constraints [Barnier 02]
Lower cost for slot allocation
With other workload metrics [Giannazza 07]
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 6 / 23
Context Ground-Holding
Slot Allocation with CP
Optimize upon CASA Solutions
SHAMAN : CP model over 30 min periods (CENA/RFM) ISA : CP and LP [Junker et al]
Marabout : sort constraint with FaCiLe to smooth the entry rate [ATM’01]
“Complexity” of Traffic
Relevance of sector capacity to model controller workload ? Discrepancies between planned schedule and actual openings More pertinent metrics w.r.t. real-time merge/split decision [Giannazza, Guittet 06]
Prior Opening Schedule Optimization Optimize upon FMP’s opening schedule
Multiple partitioning problem, possibly with side transition constraints [Barnier 02]
Lower cost for slot allocation
With other workload metrics [Giannazza 07]
Context Ground-Holding
Slot Allocation with CP
Optimize upon CASA Solutions
SHAMAN : CP model over 30 min periods (CENA/RFM) ISA : CP and LP [Junker et al]
Marabout : sort constraint with FaCiLe to smooth the entry rate [ATM’01]
“Complexity” of Traffic
Relevance of sector capacity to model controller workload ? Discrepancies between planned schedule and actual openings More pertinent metrics w.r.t. real-time merge/split decision [Giannazza, Guittet 06]
Prior Opening Schedule Optimization Optimize upon FMP’s opening schedule
Multiple partitioning problem, possibly with side transition constraints [Barnier 02]
Lower cost for slot allocation
With other workload metrics [Giannazza 07]
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 6 / 23
Context Ground-Holding
Results
Standard Model
15 30 35
20 25
10 5 0
0 200 400 600 800 1000 1200 1400
Non régulé 50
45 40
Standard
Entries within the next δ min
Flights in the sector
Marabout
0 0 5 10 15 20 25 30 35 40 45 50
400 600 800 1000 1200
200 1400
Non régulé Tri
30
0 Standard
200 400 600 800 1000 1200 1400
25
20
15
10
5
0 Tri
Context Ground-Holding
Results
Standard Model
15 30 35
20 25
10 5 0
0 200 400 600 800 1000 1200 1400
Non régulé 50
45 40
Standard
Entries within the next δ min
Flights in the sector
Marabout
0 0 5 10 15 20 25 30 35 40 45 50
400 600 800 1000 1200
200 1400
Non régulé Tri
30
0 Standard
200 400 600 800 1000 1200 1400
25
20
15
10
5
0 Tri
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 7 / 23
Context Ground-Holding
Results
Standard Model
15 30 35
20 25
10 5 0
0 200 400 600 800 1000 1200 1400
Non régulé 50
45 40
Standard
Entries within the next δ min
30
0 25
20
15
10
5
0
200 400 600 800 1000 1200 1400
Flights in the sector
Marabout
0 0 5 10 15 20 25 30 35 40 45 50
400 600 800 1000 1200
200 1400
Non régulé Tri
30
0 Standard
200 400 600 800 1000 1200 1400
25
20
15
10
5
0 Tri
Context Ground-Holding
Results
Standard Model
15 30 35
20 25
10 5 0
0 200 400 600 800 1000 1200 1400
Non régulé 50
45 40
Standard
Entries within the next δ min
30
0 25
20
15
10
5
0
200 400 600 800 1000 1200 1400
Flights in the sector
Marabout
0 0 5 10 15 20 25 30 35 40 45 50
400 600 800 1000 1200
200 1400
Non régulé Tri
30
0 Standard
200 400 600 800 1000 1200 1400
25
20
15
10
5
0 Tri
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 7 / 23
Context Ground-Holding
Results
Standard Model
15 30 35
20 25
10 5 0
0 200 400 600 800 1000 1200 1400
Non régulé 50
45 40
Standard
Entries within the next δ min
30
0 25
20
15
10
5
0
200 400 600 800 1000 1200 1400
Flights in the sector
Marabout
0 0 5 10 15 20 25 30 35 40 45 50
400 600 800 1000 1200
200 1400
Non régulé Tri
30
0 Standard
200 400 600 800 1000 1200 1400
25
20
15
10
5
0 Tri
Context Ground-Holding
Results
Standard Model
15 30 35
20 25
10 5 0
0 200 400 600 800 1000 1200 1400
Non régulé 50
45 40
Standard
Entries within the next δ min
30
0 25
20
15
10
5
0
200 400 600 800 1000 1200 1400
Flights in the sector
Marabout
0 0 5 10 15 20 25 30 35 40 45 50
400 600 800 1000 1200
200 1400
Non régulé Tri
30
0 Standard
200 400 600 800 1000 1200 1400
25
20
15
10
5
0 Tri
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 7 / 23
Context Ground-Holding
Conflict-Free 4D Tubes
4D Trajectory Planning
European Commission Episode 3 project (WP3)
4D trajectory planning to reduce conflicts number and controller workload
Many opportunities : flight level, speed, rerouting...
Large scale combinatorial optimization problem
Deconfliction by Ground-Holding
Finest grain vs aggregated model (sector capacity) Same degree of freedom than slot allocation
Solve all conflicts above a given FL by delaying flights only Standard (flight plan) and direct routes considered
Assumption : aircraft able to follow their 4D trajectories precisely...
Context Ground-Holding
Conflict-Free 4D Tubes
4D Trajectory Planning
European Commission Episode 3 project (WP3)
4D trajectory planning to reduce conflicts number and controller workload
Many opportunities : flight level, speed, rerouting...
Large scale combinatorial optimization problem
Deconfliction by Ground-Holding
Finest grain vs aggregated model (sector capacity) Same degree of freedom than slot allocation
Solve all conflicts above a given FL by delaying flights only Standard (flight plan) and direct routes considered
Assumption : aircraft able to follow their 4D trajectories precisely...
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 8 / 23
Deconfliction by Ground-Holding Model
Model
Data
Flight plans and airspace data for one day of traffic Simulation with CATS [Alliot,Durand 97]
Trajectories sampled every 15s (shortest conflicts not missed) over French controlled airspace
Notation : flight i at point pik at time tik if not delayed
Variables and Constraints
Decision variables : delay δi for each flight i Auxilliary variables : θki = tik + δi dij = δj − δi
Constraints : two flights cannot be at two conflicting points of their trajectories at the same time
Deconfliction by Ground-Holding Model
Model
Data
Flight plans and airspace data for one day of traffic Simulation with CATS [Alliot,Durand 97]
Trajectories sampled every 15s (shortest conflicts not missed) over French controlled airspace
Notation : flight i at point pik at time tik if not delayed
Variables and Constraints
Decision variables : delay δi for each flight i Auxilliary variables : θki = tik + δi dij = δj − δi
Constraints : two flights cannot be at two conflicting points of their trajectories at the same time
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 9 / 23
Deconfliction by Ground-Holding Model
Constraints
Conflict Constraints
∀i 6= j, ∀k, l, such that dh(pik, pil) < 5 NM ∧ dv(pki, pil) < 1000 ft : θki 6= θjl
tik+ δi 6= tjl+ δj dij 6= tik− tjl Note : bandwidth coloring as a particular case
Non European Flight
Flights originating outside the ECAC zone cannot be delayed by Eurocontrol instances (≈ 10%)
Delay fixed to 0
Remaining conflicts discarded (a few dozens)
Deconfliction by Ground-Holding Model
Constraints
Conflict Constraints
∀i 6= j, ∀k, l, such that dh(pik, pil) < 5 NM ∧ dv(pki, pil) < 1000 ft : θki 6= θjl
tik+ δi 6= tjl+ δj dij 6= tik− tjl Note : bandwidth coloring as a particular case
Non European Flight
Flights originating outside the ECAC zone cannot be delayed by Eurocontrol instances (≈ 10%)
Delay fixed to 0
Remaining conflicts discarded (a few dozens)
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 10 / 23
Deconfliction by Ground-Holding Model
Conflict Detection
Conflicting Points Detection
0000 1111 00 00 0 11 11
1 000000000000 11111111 11115 NM 1000 ft
pik
pjl
i j
d < 5 Nm & d < 1000 fth v
Na¨ıve 3D Conflicting Segments
3D transitive closure of segments of conflicting points
Forbidden time interval corresponds to extremities of segments Same route : whole trajectory conflicting !
Deconfliction by Ground-Holding Model
Conflict Detection
Conflicting Points Detection
0000 1111 00 00 0 11 11
1 000000000000 11111111 11115 NM 1000 ft
pik
pjl
i j
d < 5 Nm & d < 1000 fth v
Na¨ıve 3D Conflicting Segments
3D transitive closure of segments of conflicting points
Forbidden time interval corresponds to extremities of segments Same route : whole trajectory conflicting !
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 11 / 23
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 2w
tik ∈ [1000, 1180], tjl ∈ [600, 750]
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 2w
ti1= 1000
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 12 / 23
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 2w
ti2= 1015
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 400 2w
ti3 = 1030, [tj3 = 630, tj5 = 660], dij 6∈ [370,400]
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 12 / 23
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 415 2w
ti4 = 1045, [tj3 = 630 − tj6= 675], dij 6∈ [370, 415] ⊆ [370,415]
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 430 2w
ti5 = 1060, [tj3 = 630 − tj7= 690], dij 6∈ [370, 430] ⊆ [370,430]
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 12 / 23
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 430 2w
ti6 = 1075, [tj4 = 645 − tj8= 705], dij 6∈ [370, 430] ⊆ [370, 430]
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 385 445 2w
ti7 = 1090, [tj4 = 645 − tj8= 705], dij 6∈ [385, 445] ⊆ [370,445]
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 12 / 23
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 385 445 2w
ti8 = 1105, [tj5 = 660 − tj9= 720], dij 6∈ [385, 445] ⊆ [370, 445]
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 385 460 2w
ti9 = 1120, [tj5= 660 − tj10= 735], dij 6∈ [385, 460] ⊆ [370,460]
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 12 / 23
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 400 460 2w
ti10= 1135, [tj6 = 675 − tj10= 735], dij 6∈ [400, 460] ⊆ [370, 460]
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 415 460 2w
t11i = 1150, [tj7 = 690 − tj10= 735], dij 6∈ [415, 460] ⊆ [370, 460]
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 12 / 23
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 415 430 460 2w
t12i = 1165, [tj9 = 720 − tj10= 735], dij 6∈ [415, 430] ⊆ [370, 460]
Deconfliction by Ground-Holding Model
4D-Conflict Constraints
dij
0 i
j
−2w 370 460 2w
dij = δj − δi 6∈ [370, 460]
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 12 / 23
Deconfliction by Ground-Holding Model
Mutliply-Conflicting Flight Pair
-10000 -5000
0 5000
10000 15000
20000 25000-20000 -15000
-10000 -5000
0 5000
10000 15000 0
50 100 150 200 250 300 350
4503 4589
900 910 920 930 940 950 960 970
dij = δj − δi 6∈ [lb1ij..ub1ij] ∪ · · · ∪ [lbkij..ubkij]
Deconfliction by Ground-Holding Model
Flight Conflicting with Many Other
Constraint Graph of High Degree
Maximally delayed flight potentially conflicting with 130 other Highest degree > 300
Large cliques > 60
One single large connected component
-40000 -30000
-20000 -10000
0
10000 20000
30000
-50000 -40000 -30000 -20000 -10000 0 10000 20000 0 100 50 150 200 250 300 350 400 450
500 550 600 650 700 750 800 850 900
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 14 / 23
Deconfliction by Ground-Holding Model
Further Instance Conditioning
Simulator Data
Date of the day of traffic Standard or direct routes Trajectories sampled every 15s
Instance Filtering
TMA : trajectories cut around airports (10 NM) for takeoff/landing Maximal delay : problem size grows as more conflicts may occur Minimal flight level (usually upper airspace ≥ FL290)
Minimal gap between two disjoint conflicting intervals of the same pair, otherwise merged
Time unit (1 min) : scaled with strict enclosure of conflicting intervals Flights without conflict are withdrawn
Deconfliction by Ground-Holding Model
Further Instance Conditioning
Simulator Data
Date of the day of traffic Standard or direct routes Trajectories sampled every 15s
Instance Filtering
TMA : trajectories cut around airports (10 NM) for takeoff/landing Maximal delay : problem size grows as more conflicts may occur Minimal flight level (usually upper airspace ≥ FL290)
Minimal gap between two disjoint conflicting intervals of the same pair, otherwise merged
Time unit (1 min) : scaled with strict enclosure of conflicting intervals Flights without conflict are withdrawn
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 15 / 23
Deconfliction by Ground-Holding Search and Optimization
Search and Optimization
Search Strategy
Directly labelling the delay decision variables is inefficient High-order decision scheme by analogy with disjunctive tasks in scheduling problems
Order conflicting flights by branching within the disjoint intervals of their dij domain
Dynamic variable ordering : choose dij with highest sparsity first Choose smaller interval first to maximize propagation
Then label the delays δi by increasing values Optimization
Cost = maximum delay : equity, easiest for optimality proof Sum/Mean : exponentially harder
Leximin ? might be too propagation-costly
Deconfliction by Ground-Holding Search and Optimization
Search and Optimization
Search Strategy
Directly labelling the delay decision variables is inefficient High-order decision scheme by analogy with disjunctive tasks in scheduling problems
Order conflicting flights by branching within the disjoint intervals of their dij domain
Dynamic variable ordering : choose dij with highest sparsity first Choose smaller interval first to maximize propagation
Then label the delays δi by increasing values Optimization
Cost = maximum delay : equity, easiest for optimality proof Sum/Mean : exponentially harder
Leximin ? might be too propagation-costly
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 16 / 23
Deconfliction by Ground-Holding Results
Results
Instance Size
Traffic from 2006/2007 Up to 8000 flights
Up to 300 000 conflicting pairs
Best solvable volume of airspace down to FL0 (except TMA) for the easiest day
Disjoint conflicts for the same pair : up to 26 with raw data, 4 after processing
Difference domains with up to 97% sparsity Limitations
Instance size limited by memory usage (4 GB) Running times < 30 min (Core 2 Duo @ 2.4 GHz) No optimization of the mean/sum
Deconfliction by Ground-Holding Results
Results
Instance Size
Traffic from 2006/2007 Up to 8000 flights
Up to 300 000 conflicting pairs
Best solvable volume of airspace down to FL0 (except TMA) for the easiest day
Disjoint conflicts for the same pair : up to 26 with raw data, 4 after processing
Difference domains with up to 97% sparsity Limitations
Instance size limited by memory usage (4 GB) Running times < 30 min (Core 2 Duo @ 2.4 GHz) No optimization of the mean/sum
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 17 / 23
Deconfliction by Ground-Holding Results
Minimal Flight Level vs Number of Flights
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
100 150 200 250 300 350 400
060709s 060709d 070123 070622 060425
Deconfliction by Ground-Holding Results
Number of Flights vs Number of Conflicts
0 50000 100000 150000 200000 250000 300000
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
060709s 060709d 070123 070622 060425
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 19 / 23
Deconfliction by Ground-Holding Results
Number of Flights vs Computation Time (Proof)
0 200 400 600 800 1000 1200 1400 1600 1800
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
060709s 060709d 070123 070622 060425
Deconfliction by Ground-Holding Results
Minimal flight level vs optimal cost
0 20 40 60 80 100 120 140 160 180
260 280 300 320 340 360 380 400
060709s 060709d 070123 070622 060425
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 21 / 23
Further Works
Further Works
Validation and Robustness
Validation of the solutions with the CATS simulator (undergoing) Robustness of solutions w.r.t. uncertainty : vertical and ground speed, takeoff time
Perspectives
Rotation constraints : easy to implement but not provided by airlines Prior flight level allocation : pre-deconfliction, lower delay costs [CP-AI-OR’02]
Larger (European) instances with soft constraints and other
optimization paradigms : local search (LS), meta-heuristics, combined with CP (LNS)
Post-optimization of the sum/mean with LS or CP once the maximum delay is bounded
Further Works
Further Works
Validation and Robustness
Validation of the solutions with the CATS simulator (undergoing) Robustness of solutions w.r.t. uncertainty : vertical and ground speed, takeoff time
Perspectives
Rotation constraints : easy to implement but not provided by airlines Prior flight level allocation : pre-deconfliction, lower delay costs [CP-AI-OR’02]
Larger (European) instances with soft constraints and other
optimization paradigms : local search (LS), meta-heuristics, combined with CP (LNS)
Post-optimization of the sum/mean with LS or CP once the maximum delay is bounded
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 22 / 23
Conclusion
Conclusion
ATM
Ground-Holding for deconfliction vs macroscopic regulation Large problem but optimality proof obtained (w.r.t. max) with CP Some instances with solution compatible with CFMU figures, too costly for some others
Better results with direct routes
Has to be combined with other strategies, like flight level allocation, to lower the delay cost
Uncertainties : have to be taken into account in the operational setting, until accurate 4D-FMS are designed
CP
CP technology scalable to such LSCOP
Still scalable to European instances 20 000-30 000 flights/day ? Combined with other search paradigms : LS to solve CSP, CP to speed up LS...
Conclusion
Conclusion
ATM
Ground-Holding for deconfliction vs macroscopic regulation Large problem but optimality proof obtained (w.r.t. max) with CP Some instances with solution compatible with CFMU figures, too costly for some others
Better results with direct routes
Has to be combined with other strategies, like flight level allocation, to lower the delay cost
Uncertainties : have to be taken into account in the operational setting, until accurate 4D-FMS are designed
CP
CP technology scalable to such LSCOP
Still scalable to European instances 20 000-30 000 flights/day ? Combined with other search paradigms : LS to solve CSP, CP to speed up LS...
N. Barnier & C. Allignol (ENAC – DTI) Deconfliction with Constraint Programming ATM-CT 12/02/08 23 / 23