Between July 2004 and February 2006, research was continued on the safety implications of flush medians in Auckland, New Zealand. A site-by-site benefit/cost analysis found that 38% of studied sites achieved a negative benefit/cost ratio, proving that separate analysis is required for every proposed flush median site. The width of a flush median was found to have no effect on overall benefit/cost ratios.
Crash types that increased as a result of installing a flush median were investigated. JA, FD and LB-type crashes are the crash types that were shown to increase as a result of installing a flush median.
It was concluded that JA-type crashes increased because installing a flush median reduces visibility. To mitigate the expected increase in crashes that a flush median is likely to cause, visibility from side roads should be improved. It was concluded that FD and LB-type crashes increased as a result of more congestion, which caused longer queue lengths and longer peak periods. The probable increase in FD and LB-type crashes when a flush median is installed can be minimised by ensuring adequate capacity at all intersections along the route.
A methodology was formed to predict crash patterns at proposed flush median sites. The Preliminary guidelines for safer flush medians were updated to include findings from this research. A general process to follow when implementing flush medians was formed, including the identification of site characteristics, prediction of crashes using the methodology, and remedial measures to improve safety. The use of these guidelines should ensure safety increases and crash savings at all flush median sites.