Motor vehicle accidents were analysed as related to highway classes and design elements to determine accident rates for these classes and elements. A one-year study was conducted to determine the accident and severity rates for various highway types and ADT's. A two-year study was undertaken after the completion of the first phase to extend the accident and severity rates to specific geometric roadway features. Data were collected in the first phase in California, Louisiana, and Ohio. Data, processed to be analysed, were composed of the following segments of highways: (1) known segment lengths, (2) known ADT, (3) each homogeneous with respect to number of lanes, median, and access control, and (4) to each was affixed the record of the accidents which had occurred upon it during the period of years under study. Traditional accident and severity rates were calculated for: accidents per million vehicle- miles, one vehicle accidents per million vehicle-miles, injury accidents per mile of highway, etc. Regression analysis techniques were applied to the data by the use of a model. Data on specific geometric elements of the roadway were obtained from Ohio, Connecticut and Florida. These data were used to extend accident and severity rates to geometric elements. These data, as processed for analysis, consisted of the following: (1) 0.3-mile highway segments, each with known ADT, (2) each homogeneous with respect to number of lanes, access control, and median, (3) each containing known geometric elements (curvature, gradient, intersections, and structures), and (4) to each was affixed the record of the accidents which had occurred upon it. Statistical analyses were applied to each of the several highway types subdivided by the geometric elements for each accident type. Results are presented as accident rates -- expected number of accidents per 0.3-mile segment -- on pure segments (no curve, no grade, no intersections, and no structures) of a given highway type with multiplication coefficients for segments containing geometric elements or combinations of elements. Results from the first study show that 4-lane highways had higher accident rates than 2-lane highways when there was no median and no access control. Access control had the most powerful accident reducing effect, and partial control of access was partially effective. Medians seem to tend to decrease the number of accidents. The number of one- vehicle accidents per million vehicle-miles (MVM rates) decreased with increasing ADT and the MVM rate for multi- vehicle accidents increased with increasing ADT.
Samenvatting