Interurban Accident Rates By Road Type and Geometric Elements
D O'Cinneide, J C Murphy, T Ryan, University College Cork, IE
Little reliable information is available internationally on accident rates by road type or on the relationship between accident rates and geometric elements. Also, the available information cannot easily be compared due to differences in the sizes of the networks studied, definitions and units used, etc.
In Ireland, UK default accident rates are usually assumed. However, the new National Road Authority Database includes geometric information, traffic volumes and accidents on each uniform section of the entire National Road Network. This provides a unique opportunity to investigate accident rates and the influence of different geometric elements on accidents. The Database allows research which previously could not have been conducted for an entire national road network to take place.
In this paper, the National Roads Authority Database is used to derive interurban accident rates for motorways, dual carriageways, three-lane, improved wide two-lane, improved standard two-lane and unimproved two-lane roads. Accidents occurring over a nine years period are examined. The derived accident rates emphasise the relative safety of motorways and dual carriageways, but raise questions on the safety of improved undivided roads. Also, it is found that a large percentage of accidents on three-lane roads result in fatalities. The derived accident rates are compared with the limited number of published international rates.
Major issues considered include checks on the reliability of databases used, the effects of the level of underreporting of accidents, the effects of the absolute traffic volumes and of the lengths of individual sections on accident rates.
Following an investigation into the relationship between geometric elements and accident rates (including lane width, hard shoulders, medians, intersections, verges, grades, curves and passing sight distance), regression analysis is used to identify the geometric elements which have the greatest impacts on safety. Of the variables examined, vehicle kilometres of travel is shown to exert by far the greatest influence on the number of accidents occurring on a given section of road. Lane width, shoulder width and the number of roadside events such as junctions and roadside developments are also shown to have a significant impact on safety.
A detailed examination into the impacts of lane width and shoulder width on accidents shows that lane widths of 3.25 to 3.50m should be avoided on undivided roads while a lane width from 3.50 to 3.75m is optimal for safety. To minimise accidents on undivided roads, the hard shoulder width should be in the range 2.50 to 3.0m.
Association for European Transport