Calibration of Safety Performance Functions for Massachusetts Urban and Suburban Intersections

2016PDFTech ReportUMTC 16.0178639CalibrationHighway safetyIntersectionsMultiple vehicle crashesSingle vehicle crashesPedestrian-vehicle crashesSuburbsUrban highwaysHighway Safety ManualNegative BinomialSafety Performance FunctionsIntersectionSafetyMassachusettsMassachusetts. Dept. of Transportation. Office of Transportation PlanningXie, YuanchangChen, ChenUniversity of Massachusetts at LowellMassachusetts. Dept. of Transportation. Office of Transportation PlanningUnited States. Federal Highway AdministrationUS Transportation CollectionThe American Association of State Highway and Transportation Officials (AASHTO) 2010 Highway Safety Manual (HSM) introduces various Safety Performance Functions (SPFs) to assist state Departments of Transportation (DOTs) with assessing the safety performance of urban and suburban arterial intersections. However, the functions included in the HSM were developed based on crash data collected from a limited number of states, and they do not account for jurisdiction-specific differences. This research first calculated the calibration factors for the SPFs in the HSM using Massachusetts data. The results show that the calibration factors for 3SG and 4SG intersections are substantially greater than 1.0, suggesting that the observed crashes at these two types of intersections are significantly higher than those predicted using the HSM SPFs. This research also developed new SPFs for urban and suburban intersections in Massachusetts. Given the limited amount of data, this research was able to generate statistically meaningful SPFs for multiple-vehicle crashes only. A simplified approach was developed for predicting single-vehicle, vehicle-bicycle, and vehicle-pedestrian crashes. The calibration factors were calculated again based on the new SPFs. The new factors are all reasonably close to 1.0. The HSM SPFs for vehicle-pedestrian collisions at signalized intersections require daily pedestrian volumes. In this research, regression models were developed to estimate daily pedestrian volumes. Additionally, Excel spreadsheets were developed to (1) implement the SPFs in the HSM and the newly developed ones for predicting crash frequencies at urban and suburban arterial intersections; and (2) identify high-risk intersections.75