Improving Bus Transit On-Time Performance through the Use of AVL Data


Jack M. Reilly

Document Type


Publication Date


Subject Area

technology - automatic vehicle monitoring, place - north america, mode - bus, operations - scheduling, operations - performance


Capital District Transportation Authority (CDTA), automatic vehicle location (AVL), fixed-route transit


The purpose of this project was to develop a set of desktop tools to analyze archived fixed-route transit automatic vehicle location (AVL) data for the purpose of measuring on-time performance and developing schedule times (running times) between timepoints. The tools were developed using data from the Capital District Transportation Authority (CDTA) in Albany, NY. The project was also intended to determine if the system developed could be exported to other transit agencies with a different AVL system that that used by CDTA. Through consultation with staff of the CDTA, we developed a set of requirements for the system including the reports to be produced, formats and user interfaces. We developed a prototype system which included a number of reports on ontime performance and running times both from originating terminals as well as intermediate timepoints on a route. Further, we prepared a set of tools which assessed the layover time at the end of scheduled transit trips. The prototype was developed using CDTA data and revisions were made based on comments from the CDTA staff and those of the Project Review Panel. In addition, we applied the software to data from the Ann Arbor Transportation Authority and the Lehigh and Northampton Transportation Authority in Allentown, PA. We determined that using archived transit AVL data could be used to provide reasonable results in running times. While it is theoretically possible to reduce the peak fleet requirement by reducing running times, we did not experience this. This is likely due to the fact that the system was applied to smaller transit agencies which have few buses to begin with on the routes they operate. The tools we did develop could improve the on-time performance of transit systems or determine the upper bound on on-time percentage given underlying variability in the transit travel times due to factors outside of the control of transit operators such as vehicle traffic. The procedure for the determination of appropriate running times consisted of two analyses: terminal to terminal times and times between intermediate timepoints. Suggested terminal to terminal times were established by finding the time necessary to assure that the subsequent trip on a vehicle assignment could depart on-time with a certain probability such as 95%. Suggested intermediate timepoint times were established by determining the specific time which would maximize the number of bus trips which would depart from timepoints between one minute early and five minutes late. We were able to apply a few statistical tools to transit AVL data to make the determination of appropriate running times but also make the system accessible to transit schedulers through the development of a simple user interface. The application of these tools to the transit systems in Ann Arbor and Allentown demonstrated that the system could be exported to develop appropriate running times on data from different AVL products. This would require some reformatting of data from these AVL systems. Our expectation is to work with firms which develop AVL products to determine the feasibility of commercialization of the desktop tools developed in this project.


Permission to publish the abstract has been given by Transportation Research Board, Washington, copyright remains with them.