Evaluating the Performance of NextGen Using NASA s Airspace Concepts Evaluation System (ACES) Frederick Wieland, Greg Carr, George Hunter, Alex Huang, Kris Ramamoorthy
Agenda Approach NextGen Conops (Summer 2006) Mapping Conops to Simulation Parameters NextGen Performance Results Implications
Approach Used to Compute NextGen Performance Improvements
Methods of Computing Benefits (Notional) 20 18 16 Jul 27 Average Delay/Flight 14 12 10 8 6 Number of flights is constant, delays decrease Feb 19 May 10 Delays constant, capacity increased 4 2 0 Number of Flights
Approach for This Study (Notional) Average Delay/Flight 60 50 40 30 20 10 : Current system + OEP improvements, # flights set at delay threshold for worst-case weather Increase flights until worst-case delays reach an intolerable level then... Compute the delay reduction with NextGen improvements 0 Number of Flights
A Word About the Metric Used Herein for Performance Analysis Delay Metric: Sched departure time ACES-computed minimum flight time Delay ACES-computed arrival time Airline scheduled departure time explicit; scheduled arrival time ignored Provides the effect of the system on flights, without regard to schedule padding Also...all flights are flown by the simulation, regardless of their delay Allows realistic, meaningful comparisons between system configurations Avoids the issue of flight cancellation policy as a function of air carrier business model
NextGen CONOPS
En route ADS-B surveillance Moving map RNAV routes Weather-savvy decision tools Wake diminishing airframe designs climb NextGen Technologies Data-linked exchanged trajectories 4D trajectory re-negotiation Auotmated separation assurance voice Controller Continuous descent arrivals RNP route from cruise to runway descent Arrival scheduling/sequencing tools Runway reconfiguration forecasts Terminal RNP routes to cruise takeoff Wake vortex prediction VMC departure rates in IMC Controller Controller Self-separation in some conditions Reduced arr/dep sep for closelyspaced parallel runways landing High speed taxi exits Variable touchdown points Wake vortex detection Surface Moving map for low-viz taxi Time-based surface mgt RTSP, CDTI, ADS-B Simultaneous single-rw ops VMC arrival rates in IMC taxi taxi gate SMS Controller Dispatch 4D gate-to-gate trajectories SWIM information enables better pre-departure planning Dispatch Controller gate
Modeling Tools
EAD Integrated Modeling and Analysis Process ETMS AvDemand TSAM Boeing Capacity model Segments 3, 7 demand sets Today s traffic Today s capacities FAA benchmark capacities AP caps for Segs 3, 7 LMINET Airport Weather Trim demand Enroute Weather Feasible throughput ACES Wx-impacted enroute capacities ProbTFM 3-D Viz Tool NextGen OI performance ADSIM, RDSIM Env Analysis To gate Sensitivity of performance to wx predictability Sensistivity of performance to ATC capability Animation showing enroute and terminal-area effect of wx, pre & post NGATS Capacity Benefit Calculation
Validation of ACES Validation date: February 19, 2004 Reference: Post, Joseph, James Bonn, Sherry Borener, Douglas Baart, Shahab Hasan, Alex Huang, A Validation of Three Fast Time Air Traffic Control Models, Proceedings of the 5 th ATIO Conference, September, 2005
Modeling Assumptions: Translating NextGen Improvements into Airport and Enroute Modeling Parameters
Estimating Airport Capacities for NextGen Improvements Models Factor Wx Segment 3 Segment 5 Segment 7 ROT (exit velocity) All Wx 30 kts 30 kts 45 kts 45kts Mean departure release time & standard deviation VMC MVMC IMC 8 sec, 6 sec 8 sec, 6 sec 8 sec, 6 sec 4 sec, 2 sec 4 sec, 2 sec 4 sec, 2 sec 2 sec, 1 sec 2 sec, 1 sec 2 sec, 1 sec 1 sec, 0 sec 1 sec, 0 sec 1 sec, 0 sec Single Runway Constraints Predictability at outer marker Final approach path length All Wx VMC MVMC IMC 18 sec 3 nm 5 nm 5 nm 12 sec 3 nm 3 nm 3 nm 9 sec 3 nm 3 nm 3 nm 6 sec 3 nm 3 nm 3 nm Arrival / arrival separation All Wx 2 touchdown points 2 touchdown points 4 touchdown points Arrival / departure separation All Wx Departure / arrival separation IMC Same as VMC Same as VMC Same as VMC Departure / departure separation All Wx Source: Monica Alcabin, Boeing Corporation
Calibrating Airport Capacities Current Operations FAA Benchmark Capacities FAA Benchmark Report Runway Configurations Constraints Identification Capacity Equations Modified Constraints Set Fleet Mix Runway Exits Aircraft Parameters Single Runway Model Arrival Rates Departure Rates Mixed Rates Constraint Values Constraints Calibration Airport Capacities NGATS Airport Capacity Benefits Assessment Operational Improvement Roadmap Seg 3 Inputs Seg 5 Inputs Seg 7 Inputs Airport Capacity Model Segment 3 Capacities Segment 5 Capacities Segment 7 Capacities Source: Monica Alcabin, Boeing Corporation
Modeling Assumptions, Broad Area Precision Navigation + Aircraft Trajectory-Based Operations Segment 3 Roadmap Enhancements RNP routes are available everywhere Meaning they have been designed and published; not mandatory Time-based trajectories available everywhere; arrival and departure sequencing and spacing tools available only at OEP airports All high-altitude (FL290+) flights managed by 4D trajectory, and exchanged via data-link No vectoring: whole trajectory is recomputed upon conflict Modeling Approach, Segment 3 It is clear that RNP routes are not mandatory, hence there is still controller workload in transition airspace In reviewing benefits literature, we agreed to a 10% decrease in workload for Segment 3 A similar workload decrease was observed during the introduction of URET: we hypothesize that the introduction of these procedures would have an impact at least as great as URET. Source: Kerns, Carol and Alvin McFarland, Conflict Probe Operational Evaluation and Benefits Assessment, MITRE/CAASD MPW0000239 Additionally, a partially-implemented datalink assumed to reduce overall controller workload by 15% sector capacity increased by 15% Source: Center for Naval Analysis CPDLC Benefits Story, 2003 (Powerpoint presentation
Modeling Assumptions, Broad Area Precision Navigation + Aircraft Trajectory-Based Operations Segment 7 Roadmap Enhancements Time-based and metered RNP routes flown to and from all runway ends at top 100 airports controller workload decreases dramatically 4D gate-to-gate trajectories are filed and flown by flights arriving or departing from OEP airports All commercial and enroute traffic managed by 4D trajectories Modeling Approach, Segment 7 RNP routes aircraft become invisible to controllers in transition airspace for the top 100 airports Based upon Eurocontrol experiments of pilot self-separation in the terminal area, the group decided upon a 50% reduction in controller workload when aircraft are in tubes Sources:» Zingale, Carolina M., Pilot-Based Separation and Spacing on Approach to Final: The Effect on Air Traffic Controller Workload and Performance, DOT/FAA/CT-05/14, 2005.» Grimaud, I., E. Hoffman, L. Rognin, and K. Zeghal, Towards the use of spacing instructions to sequencing arrival flows, Operational Datalink Panel Working Group presentation, 2003. Additionally, a fully-implemented datalink assumed to increase sector capacities by 30% Source: Center for Naval Analysis CPDLC Benefits Story, 2003 (Powerpoint presentation
Modeling Assumptions, Minimize Applied Separation Segment 3 Three mile separation standard applied to new airspace But not yet implemented Segment 7 Aircraft performance variability further reduced through tighter aircraft performance standards Modeling Approach, Segment 7 Three mile separation standard + tighter aircraft performance standards allows us to assume 3 mile enroute longitudinal separation
Not Modeled Flexible Airspace: Splitting/recombining sectors for workload management Mostly a cost issue anyway General Aviation corridors in Class B airspace Further reduces controller workload, in proportion to the amount of transiting GA traffic Effect of reduced performance variation on system performance Except for reduced enroute and terminal-area spacing, which is modeled
Convective Weather
Three Weather Days February 19, 2004 May 10, 2004 Mostly clear, some fog in the AM and some snow in the mountains July 27, 2004 Dominated by a low pressure system across the midwest into the northeast Major frontal system from the southeast to the northeast; heavy precipation and T- storms in northeast
The Weather Information Integration Approach Analyze traffic and weather data and forecasts Unify all relevant demand information Historical trends, flight plans, weather and winds, TFM initiatives, etc. Unify all relevant capacity information All types of weather phenomena, SUAs, security events, volcanic ash, etc. Create system capacity and loading forecasts with probability distribution Construct congestion forecast database 0100Z 2400Z 2300Z 2200Z 2100Z 2000Z Source: Ramamoorthy, K. and G. Hunter, Modeling the Performance of the NAS in Inclement Weather, Proceedings of the 6 th ATIO Conference, 2006.
Dynamic Sector Capacity Changes ZDC 18 Sector Capacity (MAP Value) 20 15 10 5 NextGen Improvement 0 600 800 1000 1200 1400 1600 1800 2000 GMT Minutes
Results
Determining the Performance 80 70 1X NAS ~1.5X NAS Average Delay/Flight 60 50 40 30 20 10 NextGen performance baseline 0 Number of Flights ASPM subset shown here for reference only
Effect of NextGen on NAS Performance Average Minutes of Delay 80 70 60 50 40 30 20 10 0 Seg 3 Seg 7 Seg 3 Seg 7 July 27 th May 10 th Feb 19 th Seg 3 Seg 7
Delay Distribution y = Flight Count (Thousands) 18 16 14 y = Flight Count (Thousands) 12 10 8 6 4 2 0 25 20 15 ACES Data Power Law Distribution 0.25 1.25 2.25 3.25 4.25 5.25 6.25 7.25 8.25 9.25 10 x = Minutes of delay (30 second bins) x = Minutes of delay (30 second bins) February 19, 2004 baseline 5 0 February 19, 2004 + y = 15,375 x Segment 7-0.9958 NextGen February 19, 2004 + Segment 3 NextGen 25 ACES Data 20 Power Law Distribution 15 10 y = 20,356 x -1.23 5 0 0.25 1.25 0.25 1.25 2.25 3.25 5.25 6.25 7.25 8.25 9.25 2.25 3.25x = Minutes 4.25 of delay (30 5.25 second bins) 6.25 7.25 8.25 9.25 y = Flight Count (Thousands) y = Flight Count (Thousands) February 19, 2004 + Segment 7 NextGen 35 30 25 20 15 10 5 0 ACES Data ACES Data Power Law Distribution y = 32,611 x -1.52 Power Law Distribution 0.25 1.25 2.25 3.25 4.25 5.25 6.25 7.25 8.25 9.25 y = 20,356 x -1.23 x = Minutes of delay (30 second bins)
Large Flight Delays (> 30 mins) 140 ACES Data--Tail Behavior of Delay 120 y = Flight Count 100 80 60 40 20 0 30 34 38 42 46 50 54 58 x = Minutes of delay (30 second bins) Overall NAS performance largely influenced by those flights that experience significant delays i.e., the tail of the distribution NAS performance improvements should address what happens to the abnormally delayed flights, not just normal operations
Delays Benefits ATA estimates cost of delays at approximately $50/minute With approximately 87,000 flights in the demand set, the delay savings in dollars is: $52.2 million/day for the February weather day $282 million/day for the May weather day $239 million/day for the July weather day
Implications/Conclusions Modeling the system in various weather conditions is useful in deriving performance information The leptokurtic ( fat-tailed ) power-law distribution implies that a minority of the flights skew the average delay Current system either lets them go (common for long-haul internationals) or cancels them (esp. short-haul domestic when other flights are available) NextGen end state performs well because the distribution tightens, i.e. there are fewer very-long delays Suggests that the highest payoffs involve ameliorating excessive delay, as opposed to reducing flights with average delay System policies/procedures/rules should concentrate on the highest-delayed flights Reducing excessive delays also improves system predictability and air carrier s ability to plan and respond to the delays