5 Airport and Ground Access Mode Model V4

A Model for Joint Choice of Airport and Ground Access Mode: 

A Model for Joint Choice of Airport and Ground Access Mode 11th National Transportation Planning Applications Conference May 6-10, 2007, Daytona Beach, Florida Session 14: Hot and Cool Topics in Travel Modeling Surabhi Gupta, Peter Vovsha and Robert Donnelly

Motivation for Focus on Air Passenger Demand: 

Motivation for Focus on Air Passenger Demand Problem: Regional Models lack capability to analyze changes to airports and/or ground access modes How are airports typically treated in a regional model? Employment Centers (for work trips) Special Generator type estimation (for non-work trips) Distortions for sub-areas adjacent to airports Distinctive characteristics of trips to/from airport Different from everyday activities Market includes non-residents/visitors Higher willingness to pay for some segments

FAA Regional Airport Study: Phase I : 

FAA Regional Airport Study: Phase I Preliminary Joint Airport Choice and Mode of Access model developed and tested for NY Region Developed in order to illustrate the utility of the approach for a possible subsequent more intensive phase of air passenger demand analysis, and to Provide an initial modeling tool for air passenger demand analysis with respect to airport operations, aviation service measures and pricing, and ground access characteristics Estimation: Air Passenger Survey – 2005 NY BPM travel time and costs (approximate) Airport service and cost attributes (limited)

Literature Review: 

Literature Review Pels et al.(1998) Airport-Access mode- Airline NL model Business vs leisure Hess and Polak Airport Choice (2005) - Business vs Leisure, Resident vs Visitor Airport-Airline-Access mode (2004) – Business travelers only Basar and Bhat (2004) MNL vs PCMNL model for airport choice

FAA Regional Study - 9 Airports: 

FAA Regional Study - 9 Airports John F Kennedy (JFK) La Guardia (LGA) Newark Liberty Int’l (EWR) Long Island MacArthur (ISP) Westchester County (HPN) Stewart Int’l (SWF) Trenton Mercer (TTN) Atlantic City Int’l (ACY) Lehigh Valley Int’l (ABE)

8 Ground Access Modes: 

8 Ground Access Modes Auto Drop Off/ Pick up Auto Park Taxi Cabs Local Bus Shared Van Rail Transit Rental Car Chartered Bus

Model Structure: 

Model Structure Nested Logit Model Upper Level- Airport Lower Level – Access Mode Nesting Coefficient <1 could not be estimated 72 alternatives = 9 airports x 8 access modes, out of which 68 are available and 65 were observed Explanatory Variables Airport Characteristics – size, domestic yield, delays, gauge, river crossings, distance, access mode logsum .. Access Mode Characteristics – travel time, cost, parking cost, income, age, gender, group size

Variables Used  Compared with other models: 

Variables Used  Compared with other models

Air Passenger Segmentation: 

Air Passenger Segmentation Travel Purpose Business Non Business Destination International Domestic Traveler Resident Visitor Full Segmentation Behavioral differences Partial Segmentation US Market is price comparable Domestic travel distances are comparable to international Partial Segmentation Restricted choice sets Fundamental behavior is similar (every passenger is both)

Estimation Results: Airport Choice: 

Estimation Results: Airport Choice

Estimation Results: Access Mode Choice: 

Estimation Results: Access Mode Choice

Value of Time Estimates: 

Value of Time Estimates Airport- Ground Access Choice Model (2005) Business: $62.6/hr Non-Business: $41.0/hr NYMTC Regional Model (1997) Commuter: $15.8/hr Non-Commuter: $10-$12/hr Confirmation from other research (business): Hess & Polak, 1995 ($93-$155/hr) Pels Nijkamp & Rietveld, 1995 ($120-$170/hr) Furuichi & Koppelman, 1994 ($72.6/hr)

Summary of Behavioral Observations: 

Summary of Behavioral Observations *For Non-business trips only Auto Drop-off/Pick-up is Reference point

Impact of Income : 

Impact of Income Low income groups (< 60K) Less likely - Taxis, Rental cars, Auto park Non-Availability of car or higher travel costs Prefer public transportation (rail and bus) High income groups (>140 K) Prefer Taxis Less likely to use Shared Ride, Local Buses Also prefer Auto Park, Rental Cars, Rail transit for Non-business trips

Future Development: 

Future Development The model has been applied as a sample enumeration model, meaning it adjusts (or “pivots-off”) observed or baseline forecast shares based on changes to either the ground access or the airport measures for a given planning scenario. The preliminary model has demonstrated the utility of a joint airport choice and mode of access analysis for airport ground access and operations planning Possible further development in subsequent planning phases, including: Development of additional airport related measures of capacity, service, and costs Refinement in network (skim) ground access travel times and costs Re-estimation of the model with these added variables Incorporate model as a Special Generator in the Regional NY Model