Operations Research in Air Transportation Giovanni Andreatta Padova

Operations Research in Air Transportation Giovanni Andreatta Padova University Verona 18 maggio 2006 1

Air Transportation System ¨ Infrastructure v Airports v ATM (Air Traffic Management) l ATFM (AT Flow Management) l ATC (AT Control) ¨ Superstructure v Commercial Flights v Private Flights v Military Flights Giovanni Andreatta Padova University Verona 18 maggio 2006 2

Eurocontrol ATFM area and an example of Sectorization Giovanni Andreatta Padova University Verona 18 maggio 2006 3

ATM (AT Management) Motivation for improvement ¨ New technologies v CNS (Communications Navigation Surveillance) v Datalink v Better weather forecasting ¨ New economic framework v Deregulation v Liberalization v Competition Giovanni Andreatta Padova University Verona 18 maggio 2006 4

Aviation Authorities (ICAO, FAA, Eurocontrol, National Aviation Authorities) are adopting new ATM concepts: v FANS (Future Air Navigation System) v Free Flight v CDM (Collaborative Decision Making) and tools, for example: v COMPAS v MAESTRO v CTAS Giovanni Andreatta Padova University Verona 18 maggio 2006 5

SAFETY ¨ Primary objective of ATM system ¨ Same separation rules in the last 15 years (except for vertical separation …) ¨ Lack of quantitative models ¨ Need for a greater flexibility ¨ Implementation of new concepts requires careful examination (example: implications of Free Flight) Giovanni Andreatta Padova University Verona 18 maggio 2006 6

ATFM (AT Flow Management) ¨ Providers v FAA v Eurocontrol CFMU (Central Flow Management Unit) ¨ ATM is continuously evolving v Excellent information, elaboration and visualization capabilities v Lack of satisfactory decision support systems v Need for real time tools able to handle uncertainties (e. g. , weather) and dynamically changing reality. Giovanni Andreatta Padova University Verona 18 maggio 2006 7

Example of GUI Giovanni Andreatta Padova University Verona 18 maggio 2006 8

Over 95% of the total EUROCONTROL Route charges revenues Giovanni Andreatta Padova University Verona 18 maggio 2006 9

Forecasted no. of IFR flights in CFMU area Giovanni Andreatta Padova University Verona 18 maggio 2006 10

Giovanni Andreatta Padova University Verona 18 maggio 2006 11

ATFM ¨ Demand ¨ System Capacity v Scheduled flights v Charter flights v General Aviation v Military flights v State flights v… Giovanni Andreatta Padova University v Airports (number of take-off/landings) v Airways v Sectors (no. of flights entering the sector or no. of flights in the sector) Verona 18 maggio 2006 12

ATFM Giovanni Andreatta Padova University Verona 18 maggio 2006 13

ATFM Giovanni Andreatta Padova University Verona 18 maggio 2006 14

First solution: Ground Holding ¨ Ground holding safer than airborne holding ¨ Ground holding cheaper than airborne holding N. B. : Flight shift in time Giovanni Andreatta Padova University Verona 18 maggio 2006 15

Second solution: Rerouting ¨ It is sometimes directly imposed by the Aviation Authority (CFMU) ¨ Could be in-directly imposed: air traffic demand management N. B. : flight shift in space Giovanni Andreatta Padova University Verona 18 maggio 2006 16

Ground Holding ¨ Ground holding better than airborne delay ¨ History: 1981 USA AT controllers strike ¨ Dimensions (No. of daily flights): v 60, 000 domestic flights in USA v 30, 000 flights in Eurocontrol zone Giovanni Andreatta Padova University Verona 18 maggio 2006 17

Ground Holding Problem MIT Padova University Giovanni Andreatta Padova University Verona 18 maggio 2006 18

GH model (meaning) MIN Total Cost of (ground + airborne delays) subject to: ¨ Capacity constraints ¨ Assignment constraints ¨ Coupling constraints ¨ Integrality constraints Giovanni Andreatta Padova University Verona 18 maggio 2006 19

GH model (type and dimensions) ¨ The resulting model is a MIP (Mixed Integer Programming) problem with a few hundred thousands boolean variables and constraints ¨ The problem is NP-hard (i. e. , difficult) ¨ Yet, the instances using real data were easy to solve Giovanni Andreatta Padova University Verona 18 maggio 2006 20

Rerouting A rerouting is sometimes imposed by CFMU. €€€ €€ Rerouting implies different route charges! € € Giovanni Andreatta Padova University Verona 18 maggio 2006 21

EUROCONTROL route charge formula Total cost: where for each State i the charge is given by: Distance traveled in State i Unit rate in State i Giovanni Andreatta Padova University Verona 18 maggio 2006 22

Critics to EUROCONTROL formula: Should better reflect actual ATC costs: ¨ Use time rather than distance? ¨ Question the importance of weight (augment the exponent from 0. 5 to 0. 87 ? ) ¨ Efficiency in providing ATC services should be taken into account ¨ … Giovanni Andreatta Padova University Verona 18 maggio 2006 23

Air Navigation and Airport User Charges ¨ Great interest internationally ¨ Background of rising infrastructure costs and security costs at a time of global industry crisis ¨ Multiple objectives for charging systems: v Cover costs of service providers v Cost-related charging formulae v Provide incentives to improve efficiency/performance of service providers v Inform prioritization of investments v Promote operational efficiency (e. g. , reduce congestion) v Transparency and user participation v etc. Giovanni Andreatta Padova University Verona 18 maggio 2006 24

IRCS project (Trieste + Padova) Innovative Route Charging Schemes For more information see: www. eurocontrol. int/care/innovative/projects 2002/ircs/ Giovanni Andreatta Padova University Verona 18 maggio 2006 25

IRCS Specific Issues (partial list) ¨ Alternative charging formulae v Common unit rate v Two-part charge, etc. ¨ Impacts of alternative charging systems ¨ Charge according to flight plan ¨ Tax on tickets ¨ Containment of service provider costs ¨ Other? Giovanni Andreatta Padova University Verona 18 maggio 2006 26

General aim of IRCS project the final purpose of the Innovative Route Charging Schemes project is: ¨ to study the impact of changes in enroute charges ¨ with respect to the demand supply sides of ATM ¨ in a view to improve the overall efficiency of the system Giovanni Andreatta Padova University Verona 18 maggio 2006 27

Market based Flow Demand Management and Rerouting Odoni, Fan, … (MIT) Andreatta, Lulli (Padova) Giovanni Andreatta Padova University Verona 18 maggio 2006 28

Optimal congestion fee A congestion fee on a user is optimal when it is equal to the external costs that the user imposes on the other users. For a M/G/1 queue: Marginal = cost Internal + cost Giovanni Andreatta Padova University External cost Verona 18 maggio 2006 29

Idea under investigation ¨ Without congestion fee, BIG aircraft are penalized ¨ With “optimal” congestion fee, SMALL aircraft are penalized Why not consider different categories and compute the “external” cost imposed within each category? Giovanni Andreatta Padova University Verona 18 maggio 2006 30

Airports Giovanni Andreatta Padova University Verona 18 maggio 2006 31

Airport modeling ¨ Landside (Passenger terminal) ¨ Baggage handling ¨ Apron stand allocation ¨ Arrival aircraft sequencing Giovanni Andreatta Padova University Verona 18 maggio 2006 32

Airport landside modeling European sponsored projects: ¨ IV UE framework program: “TAPE” ¨ V UE framework program: “OPAL” ¨ VI UE framework program: “SPADE” Padova University Rome “La Sapienza” University Athens AUEB Toulouse ONERA Delft University Giovanni Andreatta Padova University Verona 18 maggio 2006 33

SLAM Implementation • Program: ANSI C, LEX and YACC • Graphic User Interface: JAVA • Run under UNIX and WINDOWS Output file Input file run Giovanni Andreatta Padova University Verona 18 maggio 2006 34

Apron stand allocation Romanin-Jacur, Filippi Giovanni Andreatta Padova University Verona 18 maggio 2006 35

Arrival aircraft sequencing Bianco, Rinaldi, Dell’Olmo, Lulli, … Giovanni Andreatta Padova University Verona 18 maggio 2006 36

AIRLINES Giovanni Andreatta Padova University Verona 18 maggio 2006 37

Airlines Deregulation and Liberalization drive a higher competition Need to reduce costs Need to enhance revenues Giovanni Andreatta Padova University Verona 18 maggio 2006 38

Airlines costs ¨ The Passenger Mix Model ¨ The Fleet Assignment Problem ¨ The Crew Pairing Problem, the Aircraft Routing Problem, and the Integrated Crew Pairing-Aircraft Routing Problem ¨ The Schedule Design Problem ¨ Integrated Models Giovanni Andreatta Padova University Verona 18 maggio 2006 39

Airlines revenues ¨ Revenue management ¨ Seat pricing ¨ Seat categorization ¨ Integration Giovanni Andreatta Padova University Verona 18 maggio 2006 40

Optimization Methods ¨ Integer Linear Programming ¨ Greedy algorithms ¨ Genetic algorithms ¨ Tabu search ¨ Bi-linear programming models Giovanni Andreatta Padova University Verona 18 maggio 2006 41

Where in the USA ? ¨ NEXTOR v MIT v Virginia TECH v U. of Maryland v UC Berkeley v GMU v FAA Giovanni Andreatta Padova University Verona 18 maggio 2006 42

Where in Italy ¨ Padova v Andreatta, … Brunetta, Lulli, Romanin-Jacur, ¨ Rome v Dell’Olmo, Bianco, … ¨ Trieste v Ukovich, Castelli, Pesenti, … ¨ Siena v Marti Giovanni Andreatta Padova University Verona 18 maggio 2006 43

Thank You Padova References: www. dei. unipd. it/~brunetta/at_papers/ Giovanni Andreatta Padova University Verona 18 maggio 2006 44

Giovanni Andreatta Padova University Verona 18 maggio 2006 45