This paper outlines the methodology and analysis used in a series of Transport and Road Research Laboratory (TRRL) studies of accident occurrence at urban junctions and road links. It reports and compares some recent results for junctions. At the time of writing, TRRL had published the results of two major studies to develop accident predictive relations for urban junctions: (1) accidents at four-arm roundabouts; (2) accidents at four-arm single carriageway urban traffic signals. Various other TRRL studies were in progress. The major studies conducted a cross-sectional investigation of injury accidents across a selected sample of sites, and measured the determining variables and factors for a large sample of junctions of the specified type, to provide a sound basis for obtaining statistically significant results. Accident data were collected generally over a period of 4 to 6 years. Tabulations were obtained of accident frequency and severity and accident rates. Relations predicting accidents were established using the statistical techniques of generalised linear modelling. The 'full' relations derived have several important applications: (1) identification of generic design improvements; (2) evaluation of the safety of individual junctions; (3) examining trade-offs between safety and delay at individual junctions, then optimising safety/mobility for all road users. Most important, TRRL research is developing a tool to appraise traffic management schemes across whole towns and cities.
Samenvatting