The paper describes recent enhancements to the transit side of the Chicago Activity-Based Model (ABM) to incorporate premium transit attributes beyond travel time and cost. These new attributes are included in transit path building and mode choice. The following model components are discussed with validation results:

• Transit mode labels (like bus, rail) and associated non-behavioral constants are eliminated. Instead, the multi-modal transit system is considered with individual preferences (age, income) for finding a certain path by access mode (walk, drive).

• The entire ABM system works with 17,000 Micro-Analysis Zones (MAZs). The transit network is modeled with 6,000 physical stops. Access and egress details are modeled using detailed street networks. The transit path for each MAZ-to-MAZ pair is constructed on the fly using stop-to-stop assignments and pre-calculated access-egress components.

• Stops are classified into 5 categories (pole, shelter, plaza, station, and major station), and wait time convenience factors, real-time information factors, internal walk times, and other parameters are differentiated by stop type. Proximity to commercial activities, cleanliness categories, and other variables are introduced to affect individual wait time perception.

• Perception of in-vehicle time is parameterized by using variables such as seating convenience, productivity, cleanliness, and temperature control. Additionally, factors like ease of boarding and payment are used to parameterize perceived boarding and transfer penalties.

• Transit capacity constraints are introduced through an iterative recalculation of effective headways, and in-vehicle crowding effects are introduced through perceptional weights on in-vehicle time as a function of vehicle crowding and seating availability.

• Calculation of wait time is improved by applying non-linear curves reflecting how transit users normally arrive at stops as a function of service frequency. Transit service reliability is incorporated through an extra wait time component associated with schedule non-adherence.

• Calculation of transit fares is improved to account for actual cost structures in transit path building and subsequent mode choice. Actual transfer policies, discounts, and distance-based fares are incorporated.

• Sub-models for a probabilistic classification of users by mobility attributes (such as using a transit pass) and modality style (willingness to consider transit) were incorporate and were very significant in mode choice.