Degree Type

Thesis

Date of Award

2017

Degree Name

Master of Science

Department

Civil, Construction, and Environmental Engineering

Major

Civil Engineering

First Advisor

Omar Smadi

Abstract

The United States Road Assessment Program (usRAP) is a powerful tool for conducting Systemic Safety evaluations. The level of safety of the roads can be assessed through the usRAP Star Rating method, giving one star to least safe and five stars to safest roads. As part of the Star Rating data collection process, a comprehensive list of 40 road attributes are recorded for each 100-meter segment using StreetView imagery. Some of the challenges that are associated with usRAP data collection protocols are human error, inaccurate measurements, and the coder’s subjectivity. To examine the effects of these errors on Star Rating results, this study has leveraged the Second Strategic Highway Research Program Roadway (SHRP 2) Information Database (RID) to complement the existing dataset. The RID includes a variety of safety-related roadway attributes collected by a mobile data collection vendor and meets high accuracy requirements by implementing a quality assurance plan. Using benefit-cost analysis, this study aims to compare the objective data collection approach of utilizing a mobile data collection vendor with high quality assurance processes versus the subjective approach of coding data manually. Star Ratings are calculated for a sample of two lane rural roads in North Carolina using the RID and the manually coded dataset.

usRAP uses the risk-based non-crash measure of Road Protection Score (RPS) for assessing the level of safety of the roads by a 1-5 Star Rating scale. The previous validation studies have been mostly limited to the comparison of crash rate and Star Rating averages and have failed to establish a comprehensive statistical relationship. In order to investigate such relationship, this study develops a crash prediction model using a sample of two lane rural roads in North Carolina. The crash frequency was estimated as a function of Road Protection Score and Annual Average Daily Traffic using a negative binomial model. The results of this study showed that the crash frequency consistently increases with Road Protection Score. The safety performance function showed that moving from a 3-star road to a 2-star road would result in 47% more crashes. These findings confirm that Star Rating is a valid risk measure for crash frequency on two lane rural roads.

Copyright Owner

Zahra Parvinashtiani

Language

en

File Format

application/pdf

File Size

48 pages

Share

COinS