Presentation on theme: "Research for Ground Delay Program Enhancements October 31, 2001 Metron Aviation, Inc. Robert Hoffman, Ph.D."— Presentation transcript:
Research for Ground Delay Program Enhancements October 31, 2001 Metron Aviation, Inc. Robert Hoffman, Ph.D.
10/17/20012 Main Thrusts of GDP-E Research Allocation of Arrival Resources among –single airport, using distance (distance-based GDP) –multiple arrival fixes (multi-fix GDP) –multiple airports (multi-airport GDPs)
10/17/20013 FAA Ground Delay Programs Single airport scenario with a demand-capacity imbalance Apply FAA-assigned ground delays to flights bound for a common destination airport Ground delays reduce rate of arrival flow –alleviates airborne holding –transfers airborne queues to the ground
10/17/20014 Central (deterministic) GDP Issues Demand predictions (carrier supplied data) Capacity predictions (runway config, AARs) Geographical scope of the program Temporal scope of the program –start/end time, model time, potential cancellation time Equity among –flights –air carriers
10/17/20015 Stochastic GDP Issues Demand uncertainty –pop-up flights (FA delays) –cancellations (compression) –departure times questionable –en route time Capacity uncertainty –capacity prediction usually based on weather forecast
10/17/20016 Ground Delay or Air Delay? Aggressive program –ground delay lots of flights over long time –risk of unrecoverable ground delays Hedge against weather forecast –ground delay fewer flights over less time –risk of excessive airborne arrival queues
10/17/20017 Current practices Geographic scope of GDP set via combinations of ARTCCs (centers) Tiers (first, second, …) commonly used Power Run enumerates options
10/17/20018 How distance-based GDPs can help Choose inclusion to program via proximity to GDP airport –e.g., flights within 1500 miles subject to ground delays More options than tiers/centers –can only improve quality of solution –fine tune geographic scope Leverage off distance to –mitigate uncertainties –improve optimization (Power Run)
10/17/20019 Sample Tradeoff: Avg Delay Vs. Unrecov Delay Unrec Trade-off Avg BOS 2 Feb 1999
10/17/ Prototype Graphic Tradeoff Curve Miles UnRecov and Avg Delay (wtd sum) Tier 1Tier 3Tier 2 Recommended Not Recommended Max Delay: 271 Min Avg Delay: 135 Min Total Delay: 1557 Min Flight Count: 139 Ooh ooh, pick me!
10/17/ Tradeoff Curve is a function of MinExpected Duration of GDP Miles Tier 1Tier 3Tier 2 Hours User Set Unnecessary Delay
10/17/ New Power Run Air Hold Ouch Minutes Unrecoverable Delay Average Delay Miles Air Hold Unrecov Delay Average Delay Max Delay
10/17/ Can we optimize GDP settings? Yes, but –single parameter searches stronger than multi- dimensional searches –must be tempered with traffic flow expertise Limitations –quantification of equity principles –cost/objective functions harder than constraints Successes/Benefits –more uniform GDP settings –first formal treatment of stochastic factors
10/17/ Multi-fix GDPs Current model of flow into airport (during GDP) is a single queue
10/17/ Multi-fix GDPs In reality, flow into an airport is multiple queues over fixes.
10/17/ Objective of Multi-fix GDP To develop algorithms/procedures for controlling multiple flows into GDP airport Note: in FSM, we have single fix control capabilities only
10/17/ Issues in Multi-fix GDP Effective arrival fix balancing Equity among fixes (distribute delay evenly) Modify current substitution practices –open arrival slot over one fix does not mean opan arrival slot over another
10/17/ Multi-Airport GDP Objective To develop algorithms/procedures to coordinate GDPs at several airports Issues: –efficient use of regional airspace –equitable access to regional airspace by all airports –modify current substitution practices
10/17/ Multi-Fix Vs. Multi-Airport GDP Abstractly, each problem balances multiple flows into a common resource Solution to one is solution to the other
10/17/ Most General Problem Flows into fixes, into airports Multi-fix solution must work with multi-airport solution OriginsFixesAirports N S E W EWR LGA JFK TEB A B C D
10/17/ Case Study: New York City area
10/17/ Multi-queue/Multi-resource Allocation Issues Added equity consideration –equity among airports (or fixes) Greedy algorithms (e.g., RBS) no longer optimal –how sub-optimal is sub-optimal? Equity vs. Efficiency greater issue than ever before
10/17/ Progress on Multi-airport GDP Space-time flow model has been developed with equity a primary consideration Complex model has simple RBS-like solution Integer program, multi- commodity?
10/17/ Deployment Path Finalize algorithms and analysis Dialogue with operational people Prototyping, war games Validation –need to communicate to others –intuitively explainable? –how realistic? Plug into FSM ultimately
10/17/ Welcome By-product New ideas for rationing –apply to the single airport case Avoid RBS side effect –carriers with flights in rear of GDP get disproportionate delays Revamp rationing algorithm for single airport?
10/17/ Future work (multi-airport/fix) Finalize flow model and alternative approaches Select best balance of mathematical modeling, flow representation, and practicality Open questions: –how inefficient is the equity model? –how inequitable is the efficiency model? Merger of multi-airport/fix (long range)