Capacity Constrained Park And Ride In Trip-Based And Activity Based Models
Corresponding Author: Kevin Stefan, HBA Specto, Inc.
Presented By: Paul McMillan, HBA Specto
Abstract
Park and ride transit is an important mode in a number of cities, but it faces a relatively “hard” capacity constraint; the number of stalls. In areas of high demand, users travel earlier to try to beat their neighbours, are displaced to less popular stations, and even give up on the mode if they cannot reliably find a space.
Both Brisbane, Australia and Calgary are cities where park and ride is important – they have dense downtowns with expensive parking served by substantial rail networks. This leads to high transit use and substantial park and ride demand. A capacity constrained park and ride system has been developed for and implemented in models in both of these cities – an advanced trip-based model in Brisbane and an activity-based model in Calgary.
The capacity constrained park and ride (PnR) model uses multiple time periods to represent lots filling up as the day progresses; spaces filled up in the first time period are unavailable in later time periods. For each time period, a lot choice process is performed that represents the utility of travelling from each origin I to each destination J via every possible lot P. Trips are assigned based on this utility to each lot appropriately. An additional “shadow cost” is calculated for each lot that is over capacity; the PnR model iterates within the time period until all trips are allocated to an available stall, representing displacement from one lot to another and permitting multiple “best lots” for a given IJ pair.
The resulting logsums represent the utility of park and ride considering all available lots. These logsums are then used in the models that select modes and times of day; a lot that fills up by 8 will result in users travelling earlier to get a space. This PnR model also produces assignment of the auto and transit portions of the trip separately, and permits the representation of “kiss and ride”, where users are dropped off at transit facilities. An example using the Brisbane model shows additional park and ride capacity leading to mode, time and destination changes.