For the past decades many studies have focused on the operation and safety of limited access roadways. However, the accessibility of these facilities are made possible by interchanges which are systems of interconnecting roadways with one or more grade separations providing movement for traffic between two or more roadways of similar or different characteristics such as speed characteristics, traffic characteristics and roadway class. Most studies have focused on the total number of crashes that occur on these facilities but a few studies have been conducted on severity crashes. The main question that researchers have not answered is whether there are differences in crash severity level distributions when comparing them with each other and further when comparing them with the total crashes distribution.
Severity of crashes depend on a number of characteristics on a given network such as distribution of traffic, speed, driver behavior, environmental conditions, lighting conditions and pavement conditions. Since interchanges are roadway sections which are different from the basic freeway segment there are different safety characteristics that could be encountered. The hypothesis is that different types of interchanges have different safety characteristics depending on the type of traffic serves and the configuration. Moreover, different types of interchanges will have different distribution of severity crashes which are also a function of the configuration and the traffic it serves. Another hypothesis is that the different levels of severity have different distribution on interchanges which could also impact the total number of crashes on limited access roadways. The objectives for this research were; to compare the crash characteristics of the types of interchanges in Florida. Secondly, to compare the probability distribution for the different severity crashes on these interchanges. The third objective was to produce a crash prediction model for the types of interchange and a general crash prediction model for the all crashes and make a comparison between them for total crashes and severity crashes. Forth objective was to perform a time series and forecast analysis for total interchange crashes From the models that were produced, simulations were used to produce the comparison of the severity level crash distributions and the interchange type crash distributions with the total crash distributions and the most common distribution used for rare events which is Poisson distribution.
The results from this study showed that there were mathematical differences in the distributions but since statistics is one of the major evaluator of traffic characteristics and conditions, statistically there were no differences among the distributions. However, the regression analysis performed showed that interchange types have an impact on the occurrence of crashes. Furthermore, the time series analysis showed that there was a time dependency in the occurrence of crashes and also when other time related variable were considered in the analysis.
