Speeding and red light running are among the major causes for road crashes. Respectively, speeding is more crucial for rural areas, whereas red light running is more frequent in urban areas where most of the traffic lights are located. To adequately address both issues consistent actions are needed in three main directions (also known as the three Es): engineering, education and enforcement. This study treated, in a way, the engineering and the enforcement. On the one hand, it reviewed the effect of road design and road side environment on speed and red light running that was reported in previous studies. On the other hand, the study aims at developing a model for prediction of the percentage of offenders depending on several road characteristics. It was found that age and gender is an important factor for both types of violations. The road characteristics reported to have an influence on speed are among others curve radius, curvature change rate, road, lane and shoulder width, shoulder type, type of objects and lateral distance, traffic signs and road marking, access points density and street lighting. In regard to red light running the most important factors are related to traffic signal characteristics (control type, cycle length, yellow interval duration), intersection characteristics (traffic volumes, grade, width, speed limit, number and width of approach lanes) and traffic flow characteristics (percentage of trucks). The prediction model was based on speed and road data for the N roads in the Province of Gelderland in the Netherlands. The study was aimed at tangent sections and considered both driving directions together. When calculating the percentage of offenders a speed of 87 km/h was used as a limit. It was found that the percentage of offenders is considerably lower in horizontal curves than on tangent sections. Additionally, it was found that the horizontal curve effect disappears about 100 m from the beginning or end of the curve. After reduction, 64 sites were used for model calibration and 12 sites for validation of the resulted model coefficients. A linear and a nonlinear model were developed. Tangent length, median width and the presence of barriers were found to have a positive effect on the percentage of offenders. The presence of street lighting, on the other hand, had a negative correlation to the percentage of offenders. In the end, the only difference between both models was the logarithmic transformation of the tangent length. This transformation was necessary to capture the effect of relatively shorter tangents. This also explains the higher R2 for the nonlinear model. The R2 for the linear model is 0.428 and for the nonlinear model it is 0.502. Both models were equally well validated with the different sample of road sections. It was found that both models perform better for percentages of offenders between 15 % and 45 %. (Author/publisher)
Samenvatting