Measuring Bus Performance using GPS Technology

Assessing the running times of bus services has traditionally been a difficult and expensive task for the majority of bus operators in Australia, and in other parts of the world. Up until recently, travel times have been collected by time keepers positioned at key points along a given route or service corridor who record bus arrival and departure times. These data then need to be manually collated before any kind of analysis can be undertaken. The time consuming nature of this process restricts the ability of operators to collect large and meaningful samples of data. Furthermore, it is difficult, if not impossible, to identify congestion points from such data, and to evaluate the impact that they might have on overall service levels. Passive Global Positioning System (GPS) technology offers a low-cost means of collecting large amounts of highly accurate data, which can be used in an on-going performance assessment program. Although raw GPS track points can be viewed on most standard GIS packages, on-screen visual analysis is extremely time consuming for even small amounts of data. Programming skills are therefore required to break continuous GPS data into records that are more meaningful to an operator. A number of important tasks need to be undertaken before analysis can take place. Firstly, periods of in-service or out-of-service running need to be defined, and routes need to be identified. This can be a complicated task because operators often design shifts so that buses may switch between different areas and routes, from run to run, to maximise vehicle utilisation. Once routes are identified, records must then be separated into individual runs and matched with a timetable to compare scheduled and actual running times. This paper provides an overview of a number of software applications developed for processing and analysing large GPS data records collected by a bus operator in Sydney in late 2002/early 2003. The data collection process is described, and some examples are presented of output produced by the main trip processing and timetable query program. It is concluded that passive GPS is a highly attractive method of collecting data on performance, even for very small operators