In 2011, the author presented a simplified tour-based model for Brunswick, GA, pop. 77,000. This model borrowed relationships from an aggregate four-step model, but applied them in disaggregate form. Although this method is by no means a true activity-based model, it appears to provide many of the same benefits at a small fraction of the resource requirements.
Charlotte, NC (pop. 2.2 million) is also implementing this approach. Charlotte is a fast-growing area with a new light-rail system and many planning issues of larger areas, but the resources to create an ABM were unavailable. After performing a new home-interview survey, the staff sought an approach that would improve on the existing four-step process, fit available resources, allow them to learn more about tour-based models, and be a stepping-stone towards a future ABM.
The Brunswick approach was improved by adding a household synthesizer, which estimates size, workers, income group, and life cycle. Tours are generated using logit equations and Monte Carlo simulation and a separate attraction model is used. Tour distribution uses a doubly-constrained destination choice model, with probabilities from a conventional gravity formula. For intermediate stops, a logit model estimates the probability of 0/1/2/3 stops for each half-tour. Key variables are HH size, income, area type, and employment density. Another destination choice model estimates the stop locations from all possible destinations based on detour time, employment, and area type.
Another improvement involved a nested logit mode choice model. This is similar to the four-step version, but applied in disaggregate form. Time of day uses fixed fractions by purpose, which are converted to the probability of 4 time periods for each half-tour. A trip accumulator converts the tours plus stops into trip tables for assignment. The model runs in a few hours on a standard computer within a TransCAD environment.
The conclusions from this work are that a simplified tour-based process developed originally for a small area can be successfully applied in a larger area. The results (still under development) should meet the staff’s expectations for a disaggregate tour-based model that improves upon the four-step process but is understood by its users.