Exploring the impacts of transit priority measures using Automatic Vehicle Monitoring (AVM) data

This paper measures the operational performance of a series of transit priority initiatives using an empirical analysis of Automatic Vehicle Monitoring (AVM) data on trams in Melbourne, Australia. Very little previous resear ch has modelled factors influencing the performance of priority schemes and none has explored the relative performance of space (or lane) based measures compared to time (or traff ic signal) measures. An after-before comparison of priority schemes showed that on average both space and time priority measures reduced run time (average -0.18 m ins or 1.6% and -0.05 mins or 0.005% respectively). They also reduce run time variabili ty. On average, the space based priority measures studied covered 1.97 kms or 61% of average route section lengths. Time based measures covered on average 1.91 or 25% of the junctions on each route section studied. A regression model explained 83.5% of run time and 51.8% of the variation in run time. The most influential factors affecting running time were; route length ( β=0.59), scheduled running time ( β=0.41), space priority ( β=-0.16), weekday ( β=0.09), direction of travel (β=0.07), and time priority ( β=-0.03). Results suggest a kilometre of space prio rity results in a 7.1% reduction in run time whereas a time prio rity measure at one junction yields a 1.7% decrease in run time. Results also suggest sp ace priority (over 1 km) will reduce run time variability by 10.0% while time priority (at a single junction) will reduce run time variability by 5.4%. Both space and time priority measures produce a gre ater effect on run time variability than run time suggesting impacts on service reliability are larger. The paper discusses the implications of these findings on transport policy and explores areas for future research.