Urban and Suburban Arterial Safety Performance Functions: Final Report

This report documents findings from a comprehensive set of safety performance functions developed for the entire urban-suburban arterial road segment system on the state highway system in Washington. Conventional urban suburban safety performance functions on the basis of cross sectional classifications were developed using random parameter negative binomial models. Total crashes, as well as crashes by severity type were modeled. It was found that out of 20 statistically significant variables, number of lanes, roadway width, shoulder width, point of vertical tangent grade (PVT), vertical curve point of vertical curve grade (PVC) horizontal curve maximum super elevation (e), curve central angle (delta), horizontal curve radius (R) were found to be random parameters. In addition, derived measures such as degree of curve, absolute vertical grade difference (A), and rate of vertical curvature (K) were also found to be random. The majority of the statistically significant effects were geometric. In addition, functional class indicators such as minor arterial indicator were also found to be random.

Roadside information was not fully  evaluated due to inconsistencies in matching roadside inventories for all homogeneous segments. An alternative classification of the safety performance functions on the basis of ADT-population thresholds was also considered. Similar patterns of parameter randomness were found. In the absence of roadside and land use information, it appears from the 173 advanced random parameter models that were developed, that the treatment of geometric parameters as random is justified, due to significant unobserved heterogeneity in the urban-suburban arterial crash context.
Publication Date: 
Thursday, June 30, 2016
Publication Number: 
WA-RD 857.1
Last modified: 
10/12/2016 - 15:41
Venky Shankar, Narayan Venkataraman, Jungyeol Hong, Barad Hariharan, Daniel Kwon.
Pennsylvania State University. Department of Civil and Environmental Engineering.
Number of Pages: 
Urban areas, Suburbs, Arterial highways, Traffic safety, Highway safety, Functions (Mathematics), Traffic crashes, Crash characteristics, Highway factors in crashes, Mathematical models, Random variables, Geometric segments, Land use, Roadside, Analysis.