Travel demand models have been used for more than half a century to determine the need and estimate the usage of proposed new highway and transit systems. The majority of such models use Traffic Analysis Zones to aggregate demographic data and estimate interzonal travel demand for large time blocks (such as morning peak period). The interzonal demand is assigned to a link and node network to estimate likely roadway volumes. Activity-based travel demand models are based on the disaggregate travel activity of individual travellers, not the aggregate behaviour of all the travellers in a zone. They have the potential to better simulate behaviours such as time-of-day choice, route choice, mode choice, and trip chaining. Because they are disaggregate and based on individual behaviour, there may be potential to borrow model structures and parameters to reduce model development costs in new locations. In this project the DaySim activity-based demand model developed in Sacramento, California, was transferred to both the Jacksonville and Tampa, Florida, regions. The structure and parameters (coefficients) of the original Sacramento DaySim model were applied in Jacksonville and Tampa using local demographic and land-use data. Then, local data were used to re-estimate parameters and coefficients, effectively creating new activity-based model sets for the Jacksonville and Tampa regions. Statistical and model performance tests were conducted between the model pairs, revealing significant differences that varied by model component and the regions being compared. The analysis was hampered by small sample sizes or absence of data for certain variables required in the Sacramento DaySim specifications, leading to the observation that the complexity of a borrowed model specification should be supported by the data available at the destination site. In addition, spatial distribution of activity centers is region specific, which can lead to differences in mean trip lengths. (Author/publisher)
Samenvatting