Transit reliability has been repeatedly cited as one of the largest contributing factors to perceived accessibility in San Francisco. The San Francisco County Transportation Authority (the Authority) in conjunction with University and consulting partners are implementing an open-source, disaggregate, person-based transit accessibility and assignment tool -- FAST-TrIPs -- to be used within their SF-CHAMP activity-based model. The San Francisco FAST-TrIPs model uses General Transit Feed Specification (GTFS) schedule data, transit vehicle trajectories from the San Francisco Citywide Dynamic Traffic Assignment (DTA) model and disaggregate trip data from SF-CHAMP to assign transit person trips either stochastically (logit model) or deterministically (best path) with capacity-constraints.

The underlying simulation in FAST-TrIPs makes it an ideal tool for examining the reliability of transit paths based on changes in demand. These simulations, partnered with the San Francisco’s DTA model and extensive observed data from Automatic Passenger Counters (APCs) and Automatic Vehicle Location (AVL) systems provide a unique package to examine transit reliability from both demand and network perspectives.

This presentation will discuss how to extract reliability information from raw APC data including (1) flexible and open-source APC and AVL data processing techniques, (2) quantifying reliability as a function of demand, and (3) quantifying reliability as a function of network and service characteristics. Finally, this presentation will discuss how the team was able to simulate reliability in FAST-TrIPs and how reliability metrics from FAST-TrIPs can be used by mode- and destination-choice models.