SAFETY EVALUATION OF CURVE DELINEATION IMPROVEMENTS: EMPIRICAL BAYES OBSERVATIONAL BEFORE-AND-AFTER STUDY.

The objective of this study was to evaluate the safety effectiveness of treatments aimed at improving horizontal curve delineation. Different target crash types were considered in the analysis: total, nighttime, daytime, rainy, non rainy, run-off-road, non run-off-road, injury, property damage only. The treatment sites for this study were 15 curves of the motorway A16 Naples-Canosa, in Italy. All the curves are characterized by low radius, high deflection angle, available sight distance smaller than the stopping sight distance, and superelevation smaller than the demanded one at both the design and the operating speed. Treatments typologies were: installation of chevron signs; installation of curve warning signs and chevron signs; installation of curve warning signs, chevron signs and sequential flashing beacons along the curve. An empirical Bayes observational before-after study was performed. Results indicate statistically significant crash reductions in total, nighttime, daytime, rainy, non rainy, run-off- road, and property damage only crashes. Estimate of total crashes reduction is equal to 39.4%. Treatment effectiveness is greater for curves with radius less than or equal to 300 m and for curves with deflection angle greater than 60 gon. Estimate of total crashes reduction is equal to 52.2% for curves with radius less than or equal to 300 m and equal to 51.0% for curves with deflection angle greater than 60 gon. Most effective treatment resulted the installation of curve warning signs, chevron signs and sequential flashing beacons along the curve. Statistically significant total crashes reduction estimate for this treatment is equal to 47.6%.