Keywords

Unsignalized intersection, 3-legged intersection, 4-legged intersection, severity, safety

Abstract

According to documented statistics, intersections are among the most hazardous locations on roadway systems. Many studies have extensively analyzed safety of signalized intersections, but did not put their major focus on the most frequent type of intersections, unsignalized intersections. Unsignalized intersections are those intersections with stop control, yield control and no traffic control. Unsignalized intersections can be differentiated from their signalized counterparts in that their operational functions take place without the presence of a traffic signal. In this dissertation, multiple approaches of analyzing safety at unsignalized intersections were conducted. This was investigated in this study by analyzing total crashes, the most frequent crash types at unsignalized intersections (rear-end as well as angle crashes) and crash injury severity. Additionally, an access management analysis was investigated with respect to the different median types identified in this study. Some of the developed methodological techniques in this study are considered recent, and have not been extensively applied. In this dissertation, the most extensive data collection effort for unsignalized intersections was conducted. There were 2500 unsignalized intersections collected from six counties in the state of Florida. These six counties were Orange, Seminole, Hillsborough, Brevard, Leon and Miami-Dade. These selected counties are major counties representing the central, western, eastern, northern and southern parts in Florida, respectively. Hence, a geographic representation of the state of Florida was achieved. Important intersections' geometric and roadway features, minor approach traffic control, major approach traffic flow and crashes were obtained. The traditional negative binomial (NB) regression model was used for modeling total crash frequency for two years at unsignalized intersections. This was considered since the NB technique is well accepted for modeling crash count data suffering from over-dispersion. The NB models showed several important variables affecting safety at unsignalized intersections. These include the traffic volume on the major road and the existence of stop signs, and among the geometric characteristics, the configuration of the intersection, number of right and/or left turn lanes, median type on the major road, and left and right shoulder widths. Afterwards, a new approach of applying the Bayesian updating concept for better crash prediction was introduced. Different non-informative and informative prior structures using the NB and log-gamma distributions were attempted. The log-gamma distribution showed the best prediction capability. Crash injury severity at unsignalized intersections was analyzed using the ordered probit, binary probit and nested logit frameworks. The binary probit method was considered the best approach based on its goodness-of-fit statistics. The common factors found in the fitted probit models were the logarithm of AADT on the major road, and the speed limit on the major road. It was found that higher severity (and fatality) probability is always associated with a reduction in AADT, as well as an increase in speed limit. A recently developed data mining technique, the multivariate adaptive regression splines (MARS) technique, which is capable of yielding high prediction accuracy, was used to analyze rear-end as well as angle crashes. MARS yielded the best prediction performance while dealing with continuous responses. Additionally, screening the covariates using random forest before fitting MARS model was very encouraging. Finally, an access management analysis was performed with respect to six main median types associated with unsignalized intersections/access points. These six median types were open, closed, directional (allowing access from both sides), two-way left turn lane, undivided and mixed medians (e.g., directional median, but allowing access from one side only). Also, crash conflict patterns at each of these six medians were identified and applied to a dataset including median-related crashes. In this case, separating median-related and intersection-related crashes was deemed significant in the analysis. From the preliminary analysis, open medians were considered the most hazardous median type, and closed and undivided medians were the safest. The binomial logit and bivariate probit models showed significant median-related variables affecting median-related crashes, such as median width, speed limit on the major road, logarithm of AADT, logarithm of the upstream and downstream distances to the nearest signalized intersection and crash pattern. The results from the different methodological approaches introduced in this study could be applicable to diagnose safety deficiencies identified. For example, to reduce crash severity, prohibiting left turn maneuvers from minor intersection approaches is recommended. To reduce right-angle crashes, avoiding installing two-way left turn lanes at 4-legged intersections is essential. To reduce conflict points, closing median openings across from intersections is recommended. Since left-turn and angle crash patterns were the most dominant at undivided medians, it is recommended to avoid left turn maneuvers at unsignalized intersections having undivided medians at their approach. This could be enforced by installing a left-turn prohibition sign on both major and minor approaches.

Notes

If this is your thesis or dissertation, and want to learn how to access it or for more information about readership statistics, contact us at STARS@ucf.edu

Graduation Date

2009

Advisor

Abdel-Aty, Mohamed

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

Civil and Environmental Engineering

Degree Program

Civil Engineering

Format

application/pdf

Identifier

CFE0002900

URL

http://purl.fcla.edu/fcla/etd/CFE0002900

Language

English

Release Date

February 2010

Length of Campus-only Access

None

Access Status

Doctoral Dissertation (Open Access)

Share

COinS