A new method to expedite the attainment of a mathematical correlation of highway accidents with traffic volumes was recently devised in Brazil in connection with a study of a four-lane highway. Hourly traffic volumes corresponding to the seven days of the week were registered during 3 months of 1975 instead of using ADT values, while the section under investigation was sub-divided into 120 segments 500-meters long. Weekly accident rates were determined for each segment according to the four restrictive elements (curves, narrow bridges, no control of accesses and lack of median barriers). It was found that the degree of safety due to the existence of guard-rails is 6 times greater than the frequency of accesses, 5 times greater than the presence of curves and 3 times greater than the existence of narrow bridges. Tabulating the accident frequencies for each segment involved, the weekly accident rates per segment were determined according to the corresponding hourly volumes and a mathematical correlation established, resulting in 3 regression curves (linear, exponential and logarithmic), all of them presenting good correlation coefficients. Other analyses of the data are also discussed.
Abstract