Safety performance functions for motorcycle crashes at interchange ramp segments

Compared to other facilities (e.g., road segments and intersections), ramp segments pose more challenges for motorcycles due to their higher operating speeds (especially for merging and weaving maneuvers) and sharp curves. This study investigates motorcycle crash frequency at interchange ramp segments by developing motorcycle crash-specific safety performance functions (SPFs). Data on motorcycle-involved crashes in Kentucky over eight years (2015 to 2022) were used. Extensive ramp-specific information at 147 interchange ramp segments was manually collected.

The crash dataset exhibited marginal over-dispersion (i.e., crash variance is marginally greater than its mean); therefore, both negative binomial (NB) and Conway-Maxwell-Poisson (CMP) modeling approaches with fixed and varying dispersion parameters were explored.

The results showed that the CMP model with varying dispersion parameter (i.e., the heterogeneous CMP or HTCMP model) outperformed the other models based on various goodness-of-fit and prediction performance measures. The SPF model results revealed that higher ramp annual average daily traffic "AADT" and presence of roadside barriers were associated with increased motorcycle crash frequency along ramp segments. On the other hand, presence of right shoulder rumble strips, wider ramp lanes, and relatively longer gore lengths were significantly associated with reduced motorcycle crash frequency along ramp segments. The predicted crash frequencies from the HTCMP model were used to identify and rank high motorcycle-crash ramp segments using the empirical Bayes (EB) method.

Several safety countermeasures were proposed to reduce motorcycle crashes at interchange ramp segments, such as installing grooved rumble strips on the shoulders to alert drifting motorcyclists.