Improved treatment of special attractors Kevin Stefan May 2017
Context Some major facilities not represented well in destination choice models: Airport, stadium, hospital, major mall, etc. Unique characteristics Often policy targets
Context Traditional approach usually to add to size or add k-factor Alternative here: change travel disutility structure Consistent with theory – higher order goods have a larger market areas
Brisbane model form Aggregate trip-based model Fully nested structure: Number of trips Destination choice Mode choice Hour of peak Time of day AM Mid PM Off SOV PnR HOV 2 HOV 3 KnR Transit Walk Bike 6-7 7-8 8-9
Number of trips (generation) Destination choice TAZ TAZ TAZ TAZ TAZ TAZ TAZ Mode choice Hour of peak Time of day AM Mid PM Off SOV PnR HOV 2 HOV 3 KnR Transit Walk Bike 6-7 7-8 8-9
Adult nonworker age < 55 40 model segments 8 person types Primary student Secondary student Tertiary student White collar worker Blue collar worker Adult nonworker age < 55 Adult nonworker 55-74 Adult nonworker 75+ × 5 purposes each Home to primary* Primary to home Home to other Other to home Non home based * Primary: School for students, work for workers, escort for ANW < 55, shop for ANW 55+
Model form: base Logsum of mode choice / time of day choice model Additional distance function Size term using population, school enrollment, employment (total and by ANZSIC)
Model form: Size treatment Base model plus: Additional size term reflecting total employment in special attractor zones Calibrated to match total trips to SAs (as part of this study)
Model form: Distance treatment Base model plus: Additional parameters modifying the distance function Not calibrated specifically for SAs
Distance treatment functions
Airport trips: size treatment
Airport trips: distance
Size Distance More from further suburbs Too many clustered by airport
Conclusions Behavioural basis Attracts trips well Better respects trip length distribution
Thank You! Thanks to my coauthors: JD Hunt, Paul McMillan and Alan Brownlee, HBA Specto Inc.