Calibrating the Highway Safety Manual Crash Prediction Models for Urban and Suburban Arterial Intersections in Kansas

US Transportation Collection2020PDFTech ReportDissanayake, SunandaKarmacharya, RijeshKansas State University. Transportation CenterKansas. Department of Transportation. Bureau of Materials and Research. Research UnitKansas State University. Transportation CenterKansasArterial highwaysCalibrationCrash risk forecastingHighway safetyIntersectionsRegression analysisSuburbsTraffic crashesUrban areasCrash avoidance systemsSignalized intersectionsUnsignalized intersectionsK-TRAN: KSU-17-3 P1C2100Final Report, November 2016\u2013December 2019Kansas experienced about 60,000 crashes annually from 2013 to 2016, 25% of which occurred at urban intersections. Hence, urban intersections in Kansas are one of the critical locations in terms of frequency of crashes. Therefore, an accurate prediction of crashes at these locations would help identify critical intersections with a higher probability of an occurrence of crash, which would help in selecting appropriate countermeasures to reduce those crashes. The crash prediction models provided in the Highway Safety Manual (HSM) predict crashes using traffic and geometric data for various roadway facilities, which are incorporated through Safety Performance Functions (SPFs) and Crash Modification Factors. The primary objective of this study was to estimate calibration factors for different types of urban intersections in Kansas. This study followed the crash prediction method and calibration procedure provided in the HSM to estimate calibration factors for four different urban intersection types in Kansas: 3-leg unsignalized intersections with stop control on the minor approach (3ST), 3-leg signalized intersections (3SG), 4-leg unsignalized intersections with stop control on the minor approach (4ST), and 4-leg signalized intersections (4SG). Following the HSM methodology, the required data elements were collected from various sources. The Annual Average Daily Traffic (AADT) data were extracted from the Kansas Crash Analysis & Reporting System (KCARS) database and GIS Shapefiles were downloaded from the Federal Highway Administration website. For some of the 3ST and 3SG intersections, minor-street AADT was not available. Hence, multiple linear regression models were developed for the estimation of minor-street AADT. Crash data were extracted from the KCARS database, and other geometric data were extracted using Google Earth. The HSM requirement for sample size is 30 to 50 sites, with at least 100 crashes per year for the study period for the combined set of sites. In this study, 2013 to 2015 was chosen as the study period for 3ST, 3SG, and 4SG intersections, and 2014 to 2016 was chosen for 4ST intersections, based on the availability of recent crash data at the beginning of the calibration procedure for each facility type. The sample size considered for calibration was 234 for 3ST, 89 for 3SG, 167 for 4ST, and 198 for 4SG intersections. Out of the 234 3ST intersections, minor-street AADT was estimated using multiple linear regression models for 106 intersections. For 3SG intersections, minor-street AADT was estimated for 21 out of the 89 intersections. The calibration factors for these facility types were estimated to be 0.64 for 3SG, 0.51 for 3ST, 1.17 for 4SG, and 0.61 for 4ST when considering crashes of all severities. Considering only the fatal and injury crashes, the calibration factors were estimated as 0.52 for 3SG, 0.40 for 3ST, 2.00 for 4SG, and 0.73 for 4ST. The calibration factors show that the HSM methodology underpredicted crashes for 4SG intersections, and overpredicted crashes for the other three intersection types. The reliability of the calibration factors was assessed with the help of Cumulative Residual plots and coefficient of variation. The results from the goodness-of-fit tests showed that the calibration factors were not reliable and showed bias in the prediction of crashes. Hence, calibration functions were developed, and their reliability was examined. The results showed that calibration functions had better reliability as compared to calibration factors, with more accuracy in crash prediction. The findings from this study can be used to identify intersections with a higher probability of having crashes in the future. Suitable countermeasures can be applied at critical locations which would help reduce the number of crashes at urban intersections in Kansas, thus increasing the safety.120