《donohue286rpptHarvard University》由会员分享，可在线阅读，更多相关《donohue286rpptHarvard University（39页珍藏版）》请在金锄头文库上搜索。
1、Air Transportation Network Load Balancing using Auction-Based Slot Allocation for Demand ManagementGeorge L. DonohueGeorge L. DonohueLoan Le and C-H. ChenLoan Le and C-H. ChenAir Transportation Systems Engineering LaboratoryDept. of Systems Engineering & Operations ResearchGeorge Mason UniversityFai
2、rfax, VAHarvard UniversityHarvard University18 March, 200418 March, 2004OutlineqNecessity of Demand ManagementqHistory of US Demand ManagementqAuction model for airport arrival slots-auctioneer optimization model-airline optimization modelqAtlanta airport case study-simulated scenarios-results and i
3、nterpretationqObservationsWhy Demand Management?Atlanta Airport - FAA Airport Capacity Benchmark 2001qOver-scheduling causes delay and potentially compromises safetyData Indicates Loss of Separation Increases at High Capacity FractionStatistics at ATL, BWI, DCA and LGA airports (Haynie)qOver-schedul
4、ing causes accident pre-cursor events and potentially compromises safetyObserved WV Separation Violations vs. Capacity RatioHaynie, GMU 2002Atlanta Airport - FAA Airport Capacity Benchmark 2001Flight Banking at Fortress Hubs Creates Inefficient Runway UtilizationqOver-scheduling causes accident pre-
5、cursor events and potentially compromises safetyqUnder-scheduling wastes runway capacityEnplanement Capacity is More Important than Operational CapacityqSmall aircraft make inefficient use of runway capacityseats/aircraftCumulative Seat Share vs. Cumulative Flight Share and Aircraft SizeATL total op
6、erations (OAG Summer 2000) Cumulative flight shareCumulative seat shareExcess Market Concentration May Lead to Inefficient Use of Scare ResourcesHHI is a Metric used to Measure Market ConcentrationHirschman-Herfindahl Index (HHI) is standard measure of market concentrationDepartment of Justice uses
7、to measure the competition within a market placeHHI= (100*si)2 with si is market share of airline iRanging between 100 (perfect competitiveness) and 10000 (perfect monopoly)In a market place with an index over 1800, the market begins to demonstrate a lack of competitionHistory of US Demand Managemen
8、t- Limited #IFR slots during specific time periods- Negotiation-based allocation1968High-Density-RuleApr2000Exempted from HDR certain flights to address competition and small market access AIR-21Jan2001Cap of the #exemption slotsLottery1978DeregulationUse-it-or-lose-it rule based on 80% usage1985Slo
9、t ownership2007End of HDR. Whats next?-Congestion pricing?-Auction?LGA Airport Slot ControlDemand Management ApproachesqAdministrative-negotiation-based IATA biannual conferencesqEconomic-weight-based landing fee: no incentive for large aircraft inefficient Enplanement capacity-time-based congestion
10、 pricing: not reveal the true value of scarce resources-DoT supervised Market-based Auctions Market-based Auctions of Arrival Metering-Fix Time SlotsqHybrid Auction Model Design IssuesqFeasibility-package slot allocation for arrival and/or departure slots-politically acceptable pricesqOptimality-eff
11、iciency: throughput (enplanement opportunity) and delay-regulatory standards: safety, flight priorities-equity: stability in scheduleairlines need to leverage investmentsairlines competitiveness : new-entrants vs. incumbentsqFlexibility-primary market at strategic level -secondary market at tactical
12、 levelDesign ApproachqObjective:-Obtain Better Utilization of Nations Airport Network Infrastructure Network Load BalancingNetwork Load Balancing-Provide an Optimum FleetOptimum Fleet Mix at Safe Arrival CapacitySafe Arrival Capacity-Ensure Fair Market AccessFair Market Access Opportunity-Increase S
13、chedule PredictabilitySchedule Predictability - reduced queuing delaysqAssumptions-Airlines will make optimum use of slots they licenseqAuction rules: Bidders are ranked using a linear combination of: -monetary offer (combination of A/C equipage credit and cash)-flight OD pair (e.g. international ag
14、reements, etc.)-throughput (aircraft size) ?-airlines prior investment ?-on-time performance ?Strategic Auction Analytical ApproachNASAuction ModelSchedulesAnalysis & FeedbackBidsSlotsAirlinesAuctioneersNetwork Model-Auctions only at Capacitated Airports-Auction Licenses good for 5 to 10 yearsAuctio
15、n Model ProcessMore bids than capacityCall for bidsEnd auction processSubmit information and bidsYesNoAuctioneers actionAirlines actionSimultaneous bidding of 15-min intervalsDetermine factor weights, initial bids and incrementsSort the bids in decreasing ranksSequence flights for each intervalsLoca
16、l optimum fleet mix order: smalllarge 757heavyARR DEP X= (x1 xj xn)T xj=1 if Pj wins a round0 otherwisemoney#seats Bid vector Pj=Weight vector W = (w1 w2)TRank of a bid vector : WPjC = (WTP1 WTPj WTPn)TLP : max z = CTX s.t. (ARRX)i, (DEPX)i lies within the Pareto frontier i airlines combinatorial co
17、nstraintsAuctioneer ModelCapacity constraints for 15-min bins:(ARRX)i 25(DEPX)i 25max z = CTX s.t.A X bairlines combinatorial constraintsA = ARRDEP, b = 2525Let: Atlantas VMC Auction Modeldeparture per hourarrival per hour100,100ATLs VMC capacity (April 2000)Airline Bidding ModelqDetermine markets,
18、legs, frequencies and departure timesqFleet assignment : -(aircraft type,leg)-line-of-flying (LOF): sequence of legs to be flown by an aircraft in the course of its day Bidding is all about scheduling1,52,63,74,81,32,4BCDAEFqDetermine markets, legs, frequencies and departure timesqFleet assignment :
19、 -(aircraft type,leg)-line-of-flying (LOF): sequence of legs to be flown by an aircraft in the course of its day timeARR DEP 111111simple package biddingBidding is all about schedulingSimple Flight Schedule Example1,52,63,74,81,32,4BCDAEFDaily arrivals and departures at A of one LOF:qDetermine marke
20、ts, legs, frequencies and departure timesqFleet assignment : -(aircraft type,leg)-line-of-flying (LOF): sequence of legs to be flown by an aircraft in the course of its day timeARR DEP 111111complex package biddingBidding is all about schedulingSchedule Banking Constraints1,52,63,74,81,32,4BCDAEFDai
21、ly arrivals and departures at A of one LOF:Assume the Airlines have a Near Optimal Schedule and Try to Maintain in Auction Airlines elasticity for changing schedule original scheduled 15-min interval15min15minbids withdrawnbids withdrawn Airlines bid reasonably and homogeneously by setting an upper
22、bid threshold proportional to #seats (revenue) No fleet mix changeAirline Agent Tries to Maximize ProfitObjective function: Maximize revenue and ultimately maximize profitMaximise Subject to:Airlines package bidding constraintsTo bid or not to bidUpper bound for bidsLower bound for bidsMbig positive
23、 valueysbinary valueBoT airport threshold vector airline threshold fractionBs old bid for slot s in previous roundif airline bids for slot sotherwiseBs set of monetary bidsPs airline expected profit by using a slotVariables: Network Model used to Evaluate Auction EffectivenessLGAORDMSPDTWATLDFWLAXIA
24、DBWIDENPHXSFO11-node networkdeparture separationarrival separationRunway capacity determined by Wake Vortex Separation Standards (nmiles/seconds) (M. Hanson) and a scale factor to account for runway dependencySimulation scenariosqAssumptions:-Aircraft can arrive within allocated slots with Required
25、Time-of-Arrival errors of 20 seconds (using Aircraft RTA Capabilities)-Auction items: Metering Fix Arrival Slots-No combinatorial package bidding-Bid values and minimum increments are relative to the value of initial bidqInput: -Summer 2000 OAG schedule of arrivals to ATL (1160 flights)qScenario 1 (
26、Baseline):-OAG scheduleqScenario 2 (Simple auction):-Monetary Offer is the only determining factor-Auction-produced scheduleTraffic levels and estimated queuing delays during VMC45 min maximum schedule deviation allowed, no flights are reroutedScheduled arrivals (#operations/quarter hour)Estimated A
27、verage Runway Queuing Delay (min)ATL reported optimum rateResults : Flight Deviationsmin15-min max allowed 30-min max allowed45-min max allowed Bell-shaped curves are consistent to the model assumption about airline bidding behavior Curves are skewed to the right due to optimum sequencing that shift
28、s aircraft toward the end of 15-min intervals70%Results : Auction metrics#Flights to be rerouted#Seats to be rerouted#RoundsAverage Auction Revenue Per Flight (x $Initial Bid)Maximum schedule deviation allowed (min)23Average cancelled arrivals in summer 2000: 23#seats of rerouted flightsObservations
29、 on Research to DateqSimple Auctions could Exclude small airlines and/or small markets from Hub Airports-Simple Bidding Rules can Prevent this ProblemqNumber of flights to be rerouted is comparable to the number of cancelled flightsqCombinatorial Clock Auctions Offer a Promising Market-Based approac
30、h to Demand ManagementqAuction Proceeds could be used as Incentives to the Airports for Infrastructure Investments and to the Airlines for Avionics InvestmentsAirlines Could bid with Avionics Investment Promissory NotesqIncreased Hub airport capacity is Dependent on Aircraft being able to maintain A
31、ccurate Time-Based Separation (ROT and WV safety constraints)qData Links, ADS-B, FMS-RTA and New Operational Procedures will be requiredqAirlines could Bid with Script that constituted a contract to equip their Aircraft with-in X years (i.e. bid price)qCash Bids could be used to replace PFCs and go
32、directly to the Capacitated Airports Infrastructure Investment AccountsFuture workqMore airline and airport inputs-Experimental auction ParticipationqInclude Efficiency RulesqInclude combinatorial biddingqInclude pricingqConduct experimental auctionsqBackupObserved Runway IncursionsOne formal simult
33、aneous runway occupancySeveral “near” simultaneous runway occupanciesOut of 364 valid data points-14 secATL and LGA Aircraft Inter-arrival TimesLGA Arrival Histograms Normalized by Arrival RateDisplaying Positive or Negative Deviation from WVSS AdherencePerfect WVSS Adherence = 0 ATL Arrival Histogr
34、ams RW 27 Normalized by Arrival RateDisplaying Positive or Negative Deviation from WVSS AdherencePerfect WVSS Adherence Value = 0ATL Arrival Histograms RW 26 Normalized by Arrival RateDisplaying Positive or Negative Deviation from WVSS AdherencePerfect WVSS Adherence Value = 0Aircraft Wake Vortex Se
35、paration Violations : LGA & BWIPerfect WVSS Adherence Value = 0FAA Barriers to ChangeqFAA has an Operational and Regulatory Culture-Inclination to follow training that has seemed to be Safe in the PastqFAR has NOT Changed to Provide Operational Benefits from Introduction of New TechnologyqAssumption
36、 that Aircraft Equipage would be Benefits Driven did not account for Lack of an ECONOMIC and/or SAFETY Bootstrapping RequirementFAA Investment Analysis Primarily focus on Capacity and DelayqOMB requirement to have a B/C ratio 1 leads to a modernization emphasis on Decreasing DelayqIn an Asynchronous
37、 Transportation Network operating near its capacity margin, Delay is InevitableqDelay Costs Airlines Money and is an Annoyance to Passengers BUT-is Usually Politically and Socially AcceptableHypothesis: Most Major Changes to the NAS have been due to Safety ConcernsSafety Concernsq1960s Mandated Intr
38、oduction of Radar Separationq1970s Decrease in Oceanic Separation Standards Required a Landmark Safety Analysisq1970s Required A/C Transponder Equipageq1970s Required A/C Ground Proximity Equipageq1990s Required A/C TCAS Equipageq1990s Required A/C Enhanced Ground Prox. Equipageq1990s TDWR & ITWS Introductionq1990s Mandated Development of GPS/WAAS
