MIT International Center for Air Transportation Scalability of Air Transportation Networks through the Development of Multi- Systems: A Worldwide Perspective Philippe A. Bonnefoy bonnefoy@mit.edu & Prof. R. John Hansman rjhans@mit.edu Massachusetts Institute of Technology Global Airline Industry Program Industry Advisory Board Meeting October 25 th 2007
Introduction Air Transportation System Demand/Capacity Inadequacy Problem Growth of demand for air transportation FAA forecast growth rate (2005-2017 forecast):(enplanements air carrier: +3.1% per year, regional carriers: +4.3%, general aviation turbojet operations: +6.0%) Factors adding pressure to the system: Entry of Very Light Jets and Unmanned Aerial Vehicles Key infrastructure constraints in the air transportation system Demand/capacity inadequacy problem: leading to the generation and propagation of delays throughout the system Record of 22.1 million minutes of flight delays in 2006 Implications: Degradation of the quality of air service Economic impacts Need: Understand how the air transportation system evolved to meet demand in the past and how it will to so in the future * Data source: ICAO traffic data & FAA OPSNET data Revenue Passenger Kilometers (billion) Delays (in min.) Millions 1400 1200 1000 800 600 400 200 3.5 3 2.5 2 1.5 1 0.5 0 1970 1980 1990 2000 2010 0 North America Europe Asia and Pacific Latin America & Caribbean Middle East U.S. Flight Delays from 1990 to 2007 2
Analysis of the U.S. Air Transportation Network Analysis of the U.S. Air Transportation Network Air transportation system is fundamentally a network system Described and represented using network abstractions Use of tools from network theory Recent theories of scale free and scalable networks U.S. Air Transportation Network in 2005 * 3
Scale Free & Scalable Networks: Definitions & Properties Background on Network Theory Large Scale Network e.g. www in 1999 Notations and basic network characterization concepts: Un-weighted networks: Degree (k) = number of connections in and out of a node Weighted networks: - Flight Weighted degree (weighted by the frequency of flights on each arc) - Passenger weighted degree (weighted by the number of passenger traveling through each arc) fi fj pi pj fj fj pj pj e.g. kin = 2 kout = 2 k = 4 Number Frequency of of airports nodes p(k) 1.0 0.8 0.6 0.4 0.2 Degree distribution p ( k) α k γ linear-linear scale 0.0 0 200 400 600 800 1000 Degree (k) Networks that exhibit power law degree distribution are also called scale free networks Cumulative Frequency Number of airports nodes with with p(k>k) degree greater than 1.E+00 1.E-01 1.E-02 1.E-03 1.E-04 1.E-05 log-log scale time 1.E-06 0 200 400 600 800 1000 Degree (k) Degree Distribution Scaling up Scale free networks are represented by an affine function on a log log scale Networks that can grow without constraints are scalable (i.e. scale up ) 4
Analysis of the U.S. Air Transportation Network Analysis of the U.S. Air Transportation Network: ( Level) * The air transportation network exhibits a partial power law distribution Power law distribution for small and medium size nodes Non power law distribution (between 250,000 and 970,000 flights) Limits to scale in this network and capacitated nodes (capacity constrained airports) are present in the non-power law part of the distribution Number of airports with flight weighted degree greater than 10000 1000 U.S. Air Transportation Network (airport level) 100 10 1 Power Law Distribution Distribution with correction applied* 100000 1000000 1000 10000 100000 1000000 Flight weighted degree (i.e.number of flights) size Non Power Law Distribution + 5
Analysis of the U.S. Air Transportation Network Analysis of the U.S. Air Transportation Network: ( Level) * s in the non power law part of the distribution: 29 airports Some of these nodes are clearly constrained by capacity: Slot restricted airports (i.e. ORD, LGA, DCA) s that exhibit high level of delays that are indicative of congestion and capacity constraints (i.e. J.F. Kennedy JFK, Newark EWR, Philadelphia PHL, Boston Logan BOS and San Francisco SFO). Flight weighted degree distribution Number of airports with flight weighted degree greater than 10000 1000 100 10 1 code Power Law Distribution Distribution with correction applied* 100000 1000000 1000 10000 100000 1000000 Flight weighted degree (i.e.number of flights) Flight weighted degree (i.e. annual number of operations) code Non Power Law Distribution 29 airports 29 airports in the non power law part of the distribution Flight weighted degree (i.e. annual number of operations) ORD 964360 LGA 386589 ATL 949708 BOS 381064 DFW 795974 MIA 366561 LAX 629735 IAD 361754 DEN 556178 SEA 354658 CVG 512830 MEM 343970 MSP 498523 SLC 339080 DTW 498053 SFO 331498 IAH 494410 PIT 324287 LAS 490290 JFK 311827 PHX 484252 MCO 292520 PHL 430218 MDW 280940 EWR 420197 STL 274770 CLT 417485 BWI 266166 DCA 264784 6
Analysis of the U.S. Air Transportation Network Regional System Analysis * Hypothesis: Emergence of secondary airports in the vicinity of primary airports, leading to the development multi-airport systems, allows the network to scale Need to shift the focus from an airport level perspective to a regional airport system perspective 29 airports in the non-power law part of the distribution: basis for a case study analysis of regional airport systems. SFO LAX SEA Regional airport systems in the U.S. SAN LAS SLC PHX DEN DFW MSP STL MEM IAH ORD ATL CVG TPA DTW PIT CLT MCO MIA BOS LGA/ JFK/ EWR PHL DCA/ IAD/ BWI Definitions: Regional airport system: all airports within 50 miles of one of the identified airports. Primary airport: airport serving between 20% and 100% of the traffic in the region Secondary airport: airport were defined as serving between 1% and 20% of traffic Other airport: serving less than 1% 7
Analysis of the U.S. Air Transportation Network Primary & Secondary s in the United States Total of 16 primary and 16 secondary airports were found in the 11 multiairport systems in the United States, 14 remaining regional airports systems were single airport systems. SEA MSP ORD DTW MHT BOS OAK SFO SJC LAX BUR LGB LAS ONT SNA SAN SLC PHX DEN DFW DAL STL MDW MEM CVG ATL PIT PHL BWI IAD DCA CLT PVD ISP LGA / JFK / EWR IAH HOU TPA PIE SRQ SFB MCO MLB FLL Legend Primary airport MIA Secondary airport * 8
Analysis of the U.S. Air Transportation Network Analysis of the U.S. Air Transportation Network: (Regional Level) Primary and secondary airports in each of the regional airport system serve the demand for air transportation within the region s part of multi-airport systems are aggregated into single nodes U.S. Air Transportation Network ( level) Multi-airport system node New network composed of 11 multiairport nodes and 2159 single airport nodes Power law flight weighted degree distribution across the entire range of flight weighted degree. Number of nodes (single airport & aggregated multi-airport systems) with flights weighted degree greater than Mechanisms by which airports emerged 10 in a region are key to the ability of the system to scale and to meet demand. 1 10000 1000 100 Legend: Flight frequency (flights per month) over 1250 901 to 1250 31 to 150 601 to 900 5 to 30 301 to 600 2 to 4 151 to 300 0 to 1 Power Law Distribution (across entire range of flight weighted degree) 1000 10000 100000 1000000 10000000 Flight weighted degree (i.e. number of flights) 9
Worldwide Case Study Analysis of Multi- Systems Extension of the Case Studies to other Regions of the World Overall analysis of multi-airport systems covers: 63 case studies of multi-airport systems, accounting for 175 airports, in 31 different countries, across all regions of the world. Legend North America Europe Latin America & Caribbean Middle East Africa Asia/Pacific * World regions defined based on the IATA statistics regions 10
Worldwide Case Study Analysis of Multi- Systems Fundamental Regional Level Scaling Mechanisms Multi-airport systems evolve according to two fundamental mechanisms; Construction of new high capacity airports Emergence of secondary airports Emergence of Secondary Strengthening of secondary airport role Emergence of a secondary airport Failure to transfer traffic to an external primary airport Failure of a Secondary airport to emerge Re-emergence of a primary airport Closure / failure of primary to re-emerge Construction of new airport and transfer of traffic Legend Failure to transfer traffic to an external primary airport Construction of New High Capacity Primary airport Primary airport (Emerged) Closed primary airport - Secondary airport 11
Emergence Existing MIT Worldwide Case Study Analysis of Multi- Systems Mechanisms by which s Emerge in Multi- Systems Mechanisms by which primary & secondary airports emerge by world regions: North America Europe Construction New Construction New Emergence Existing Emergence Existing Middle East Emergence Existing Asia / Pacific Latin America & Caribbean Emergence Existing Africa Emergence Existing Construction New Construction New Legend Construction of new airport Emergence of existing airport 12
Model of Scaling Dynamics in the Air Transportation System System Dynamics Model Demand Population Socio Econ Factors Demand Generation Regional Regional Latent Latent Demand Demand Passengers Regional Latent Demand Attractiveness Alternative Choices Passenger Mode/ Choice System A Enplanements Substitute/ No Travel Secondary/ Other Regional Economic Impacts (GRP) Airlines Forecast Demand attractiveness set / Alternatives regulatory constraints Airline Network Planning / Scheduling Process Flights (acft. mvts.) Fare Freq, Secondary/ Other A. Revenue A. Cost A. Profit Attractiveness to Passengers Infrastructure Provider Forecast demand Access to Capital Environmental constraints Land availability Regulator Infrastructure Enhancement Process Capacity Utilization Ratio Need for Enhancement Delays Externalities: Cost of operations Pressure to reduce congestion Attractiveness to Airlines Regulatory framework Regulatory Process Implement Demand Mgt Need for Construction Need to manage demand 13
Key Factors Influencing the Dynamics of Multi- Systems Availability of Existing Infrastructure Analysis of regional airport system capacity coverage charts: (active civil and jointly operated airports with at least one runway longer than 5000ft) United States: Extensive set of existing under-utilized airport infrastructure Europe: Limited existing under-utilized civil airport resources (presence of military airfields) Explaining the conversion of existing military airfields into secondary airports Asia: Very limited existing under-utilized civil airport resources Data source: DAFIF (Category A & B airports) excluding existing primary & secondary airports 14
Key Factors Influencing the Dynamics of Multi- Systems Entry of Low Cost Carriers at Secondary s Low Cost Carriers (LCCs) generally concentrate at secondary airports Entry of low-cost carriers changes the dynamics at the airport and regional level: Entry of one LCC, followed by of entries of both LCCs and other airlines Increased competition at the airport and regional level Yield decrease followed by stimulation of demand and traffic shifts Presence of Low Cost Carriers vs. Other Airlines at Primary & Secondary s 36% 10% * Data source: websites 15
Key Factors Influencing the Dynamics of Multi- Systems Construction of s: Drivers & Constraints Historically, airports in Europe and the United States were built prior to or during World War II in Europe & North America. Meet the needs of military activity and growing demand for commercial traffic Environmental constraints limiting the construction of new airports in the three decades In Asia, phase of construction of primary airports in Asia is more recent (1970s and 1990s/2000s) Demand forecasts (double digit growth rates) Weaker environmental constraints (than in Europe and United States) World War II * Data source: websites and other sources 16
Key Factors Influencing the Dynamics of Multi- Systems Projected Rate of Growth of Demand for Air Transportation GDP driving air transportation activity (and conversely) Projected rate of growth of RPKs related to type and capacity of airport infrastructure developed Secondary airports (small incremental capacity strategy) in Europe and United States (small to medium average annual rate of growth) New high capacity airports in anticipation of medium to high average annual rate of growth 100000 RPK per capita vs. GDP per capita Singapore Passenger-Kilometers per Capita 10000 Cyprus Malta Mauritius Suriname Trinidad Malaysia Portugal Thailand Saudi Chile Arabia 1000 South Africa Lebanon Russian Fed. Czech Rep Gabon Costa Rica Turkmenistan Namibia Brazil Mexico Turkey Hungary Tunisia Philippines Columbia Uzbekistan Kenya Bolivia Indonesia Cuba Morocco Azerbaijan Poland Yemen Vietnam Syrian Egypt A.R. Venezuela Algeria Kazakstan Ethiopia Romania 100 Bangladesh China India 10 Ireland US New Zealand Netherlands Australia Europe U.K. Germany Canada France Switzerland Finland Austria Spain Japan Greece Italy Belgium Legend Population 1 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 GDP per capita * Data source: ICAO traffic data & CIA World Fact book data 17
Key Factors Influencing the Dynamics of Multi- Systems Future Needs for Infrastructure: Projected Demand/ Capacity Adequacy Analysis of future needs for airport infrastructure: Population / Number of airports used as a proxy for latent demand vs. existing infrastructure capabilities China and India: high population/airport infrastructure ratios will require significant future development of airport infrastructure United States and Europe: Significant number of existing airports that can accommodate future growth Future emergence of secondary airports Millions 9 8 India Population / # s (with runways longer than 5000ft) * Data source: CIA World Fact book 7 6 5 4 3 2 1 0 China Japan Italy Netherlands Brazil Spain Hungary South Africa Ireland Germany Portugal Belgium Singapore Poland Luxembourg France Europe Czech Rep Malta U.K. Greece Canada United States Finland Australia Cyprus Austria 18
Conclusions Air transportation network is not entirely scalable at the airport level due to airport capacity constraints, scales to meet demand through regional level mechanisms. Fundamental regional level scaling dynamics; construction of new high capacity airports, emergence of secondary airports. Differences & similarities across the world United States & Europe Concentration of traffic at primary airports Construction of new large airports: pre/during WWII Significant limitations to the development of new large airports; environmental barriers & constraints Emergence of secondary airports (Southwest effect) 1970s to 2000s More recent trend in Europe, especially with the growth of LCCs after deregulation in the early 1990s Asia Pacific Construction of multi-primary airport systems due to lack of existing infrastructure and projected demand Future Evolution of the System and Needs Regional level scaling mechanisms will be key to meeting future demand Need to protect exiting airport infrastructure (both civil and military airports) in the United States and in Europe Need to develop a phased development approach for future primary airports in Asia taking advantage of the lessons learned in the United States and in Europe 19