Ultra-efficient Bus Rapid Transit timetabling

Abstract

Bus Rapid Transit (BRT) systems are increasingly used, particularly in the developing world, to provide low-cost, high-capacity urban mobility. An example of this trend is Bogotá’s TransMilenio BRT system, the test site for this thesis, which uses an homogeneous fleet of 18 metre long articulated buses to service 1,450,000 passenger trips per day, and which reaches a peak passenger load level of 45,000 passengers per hour per direction. The computational tools and techniques used to plan the timetables of such BRT systems are largely the same set of tools and techniques used to plan non-BRT transit systems.Unlike other transit systems, high-load BRT systems commonly run simultaneous express services, a situation that the tools developed to timetable non-BRT transit systems were not specifically designed for. Due to the running of simultaneous express services, the timetabling of highload BRT systems becomes a combinatorial problem, a far more complex class of problem than normal non-combinatorial timetabling. The thesis is advanced here that high-load BRT systems could be timetabled far more efficiently via a software tool built around a recognition of the problem’s combinatorial nature.This thesis is tested by building such a software tool, using that tool to develop an alternative timetable for the Américas Line of the TransMilenio BRT system, and then comparing that timetable’s performance to the performance of the existing timetable. Data used for the Américas Line was from the period Monday the 23rd to Friday the 27th of May 2005. The software tool developed to process this data changes express service stopping patterns as quickly as passenger load changes, leading to a great many express patterns over the course of a one day timetable. Due to this rapid tracking of passenger load, the resulting timetable is referred to as an “ultra-efficient timetable.”The ultra-efficient timetable produced for the Américas Line has 88 unique stopping patterns, compared to the existing timetable’s three, and is shown to be tracking passenger load far more precisely. Bus fleet size under the ultra-efficient timetable is 9% lower than for the existing timetable, indicating an estimated capital cost saving for the Américas Line of US$1.8 million. Bus kilometres travelled under the ultra-efficient timetable are 40% lower than for the existing timetable, indicating an estimated annual operating cost saving for the Américas Line of US$6 million.The ultra-efficient timetable delivering these performance improvements is shown to be approximately ten times more complex than could be reasonably deployed using paper timetables. Consequently, an ultra-efficient timetable would need to be deployed in conjunction with a fully-automated passenger information system. With this caveat, the thesis that high-load BRT systems can be far more efficiently timetabled using a combinatorial software tool is confirmed here. As an alternative to deploying ultra-efficient timetables, the combinatorial timetabling technology developed for this thesis could also be used to produce more efficient versions of normal paper-based BRT timetables.