A Bus Rapid Transit (BRT) station with multiple loading zones tends to have a longer passenger-bus interface and, thus, lead to longer passenger walking times and longer bus dwell times than ordinary bus stops. As a way to reduce bus dwell times in a BRT station, this study focuses on eliminating delays in passengers’ reaction to their desired bus by designing an improved passenger information system (PIS) that can increase passengers’ certainty about the bus stopping location. This study develops an agent-based simulation model based on observations from a BRT station in Brisbane, Australia to reflect a real BRT operations and passenger flows. The input parameters for the simulation model are calibrated with actual data including smart card records, field measurements, and video recordings. After mapping passenger moving and waiting patterns, and allocation logic of bus loading areas, various what-if analyses can be performed to design better passenger information systems.
Several studies have applied simulation approaches to model traffic conditions and operational scenarios at public transit stations. Widanapathiranage et al. (2014) used microscopic simulation to analyze the relationship between station queuing and capacity. Seriani and Fernandez (2015) experimented the effect of pedestrian traffic management in the boarding and alighting time of passengers at metro stations by using simulation modelling technique. The simulation results from this study show that pedestrian traffic management measures can have significant impacts on the passenger service time, passenger density in cars and on platforms as well as passengers’ dissatisfaction in metro stations. The existing studies, however, rely mainly on field measurements and video recordings, which have a limitation in accurately identifying passenger demand for each bus route at different time-of-day periods. To overcome this limitation, this study uses smart card data to extract the detailed information on bus supply and passenger demand at a given BRT station. Using smart card transaction records collected in April 2013, the number of non-transfer/transfer passenger per route per service in 15-minute intervals during the peak hours was identified to build an origin-destination matrix for assigning boarding and alighting passengers. An example of time-dependent passenger demand used to sample boarding passengers is presented in Figure 1. Transfer passengers shown in Figure 1 are identified from alighting passengers extracted from smart card transaction records.
Figure 1. Boarding passengers at Cultural Centre Station, Brisbane, Australia, outbound in April, 2013.