Between September 2001 and June 2002, research was carried out to determine the safety implications of flush medians at 50 sites in Auckland City, New Zealand.
'Before' and 'after' crash data were analysed using both traditional methodology and generalised linear modelling (GLM). Results showed that large crash-cost savings have been achieved by implementing flush medians despite small overall savings in accident frequency.
Flush medians decreased fatal, serious and non-injury crashes while increasing minor injury crashes at the study sites.
Crash types were affected differently by flush medians, with major savings seen in turning versus same direction, 'pedestrian' and 'loss of control' crashes, and large increases seen in some rear end crashes and the turning crashes.
Crash reductions were observed to be greatest during night-time periods, with increases in crash costs observed in peak periods.
Both narrow and wide medians produced crash savings, which highlights the benefit of installing narrow medians where carriageway width restrictions apply.
Regression analysis of the crash data has been used to form preliminary prediction models for sections of road where flush medians are to be implemented.
Preliminary guidelines have been developed to ensure flush medians result in crash savings, as half the sites in Auckland City have experienced crash increases.
Remedial measures have been included to address crashes at pedestrian refuge islands on flush medians.
Keywords: accidents, analysis, Auckland, before & after, crashes, flush medians, medians, New Zealand, pedestrian islands, roads, safety traffic