This report presents the analysis conducted to identify the factors that contribute to severe and fatal crash occurrence on multilane corridors. The authors’ preliminary investigation using simultaneous ordered probit model provided enough evidence that a fixed influence area of intersections for all of the corridors is not justified. With the increase of an intersection’s influence area, crash types that are more specific to segments get included and change the crash pattern for the overall intersection. Therefore, for investigation purposes, to treat the corridors in their entirety will result in much more insightful results than when treating the segments and intersections separately. The crash data were grouped into six major types as follows: 1) rear-end, 2) head-on, 3) angle/turning, 4) sideswipe, 5) crashes involving slow moving vehicles (e.g. cycles, mopeds, etc.), and 6) crashes involving single vehicles. Binary severity classification models were developed by using non-parametric conditional inference trees. Parameters like alcohol/drug use came out to be significant across all crash types and clusters. Lane changing on corridors with high truck traffic was found to be risky from a severity point of view. Poor pavement conditions and high permitted speed limits increased the likelihood of severe rear-end crashes. Non-use of safety equipment also increased the severity level provided the crash had occurred. Based on the results of the overall investigation certain recommendations were made taking the 4 Es (Engineering, Education, Enforcement and Emergency Management) into consideration. (Author/publisher)
Abstract