The Relationship Between Geometric Design Consistency and Safety on Rural Single Carriageways in Ireland
P Watters, M O'Mahony, Trinity College Dublin, IE
A design consistency study was done. Speed prediction and accident prediction models were expanded. Highway Improvements can be concentrated on sections where the models predict that accidents are high, hence making rural single carriageways safer.
The majority of rural single carriageways in Ireland follow historical routes where their alignment is determined by land boundaries. They tend to fall below current design standards and have higher accidents rates (per vehicle kilometre) than roads designed to modern standards. The resources are not available to improve all rural single carriageways, so sections need to be identified as priority sections for improvement. Geometric design consistency studies can be used to identify inconsistent sections on highways, which can then be targeted for improvement. Consequently local authorities can make optimal use of available resources and can considerably improve the safety performance of the highway.
Accurate operating speed models are needed to calculate the measures of geometric design consistency. No geometric data exists for rural single carriageways in Ireland. A method of estimating geometric data from digital maps was implemented on some 70km of highways. Thirty curves and thirty tangents were then selected to represent the overall geometric makeup of the highway. Numerous geometric indices were measured on site. This paper will present the results of a spot speed survey which was conducted, using traffic counters, at the midpoint of each tangent and curve. Regression analysis was used to develop operating speed models which related driver speed to the geometric characteristics of the highway. These models tied in adequately with previous research.
Accident data, from 2000 to 2005, was collected for sections of road examined in this paper. The Garda (Irish police force) accident reports were examined for all accidents. The accidents recorded were plotted on digital maps using their Irish Grid coordinates and were assigned to their appropriate curve or tangent. The accidents were subdivided by severity and number of vehicles, creating five accident groups. Accident rates were calculated for each road section. Generalised linear modelling was used to expand accident prediction models for each accident group. The models linked traffic flow, vehicle speed and several consistency measures to accident rates. The R2 values of the models presented in this paper compared favourably to other studies.
These models are able to predict sections of road where accidents could conceivably be high due to the geometry of the road. Any improvement works can therefore be concentrated on these sections and hence rural single carriageways can be made safer.
Association for European Transport