Calibrating the HSM Crash Prediction Models for Urban Three-Legged Signalized Intersections

Calibrating the Highway Safety Manual Crash Prediction Models for Urban Three-Legged Signalized Intersections

The crash prediction models provided in the Highway Safety Manual (HSM) predict crashes using traffic and geometric data for various roadway facility types. The calibration of the HSM crash prediction models would increase the accuracy of crash prediction since the models in the HSM were developed using data from Minnesota and North Carolina for urban 3-leg signalized intersections (3SG). In this study, data from 89 3SGs from Kansas were used to estimate the calibration factor for better prediction of crashes. Out of the 89 intersections, 68 had actual AADT values available for the minor approach, and a multiple regression model with natural log transformation of continuous variables was developed to estimate minor AADT for the remaining intersections.  A calibration factor was estimated to be 0.64 when considering all crashes and 0.52 when considering only fatal and injury crashes, showing that the HSM methodology overpredicted the crashes at 3SG in Kansas. The reliability of the calibration factors was checked with the help of Cumulative Residual (CURE) plots and coefficient of variation (CV). To increase the reliability further, calibration functions were developed and assessed. The results showed that the calibration functions have a better reliability as compared to calibration factors.