| In the US in 2009, 5.5 million collisions occurred in which over 2.2 million people were injured and over 33,000 people died due to highway collisions. Over half of these total crashes were intersection and access point—related. Most collision reporting systems do not provide the necessary level of information to identify access—related collisions but collision data, where available, indicate a high incidence of access—related collisions.;The objectives of this research were to develop a valid statistical model to estimate the number of access point—related collisions occurring at access points near signalized intersections and providing checklist for site planners and decision-makers to distinguish higher collision sites from lower collision sites and avoid constructing higher collision sites.;Geometric, traffic, and access—related collision data over 5 years, from January 2005 to December 2009 were collected for 108 sites. Out of the 15 independent variables tested, only AADT, driveway width, and Synchro through movement 95% queue at the intersection near the access point were statistically significant in developing the collision prediction statistical model. This model could be used by state DOTs and municipal traffic engineers to address access management requirements and to predict problems likely to result from site traffic impacts.;To provide checklist for site planners to distinguish the higher collision sites from lower collision sites, the data that were previously collected and some new information such as demographic and socio economic data were used. The higher collision sites were investigated one by one. Quantitative, binary, and categorical variables, and demographic and socio economic information, were analyzed and compared between the higher and lower collision sites. Statistical tests were used to find the contributing factors and provide checklist to certify no access points will be constructed before the safety issues are considered. The proposed policy checklist stated that an access point most likely would be a higher collision site if it was operating full movement, had a driveway peak hour volume of over 120 vehicles per hour, had an intersection peak hour through movement Synchro 95% queue of over 230 feet, and had a driveway left turn proportion (100 × left turns from driveway per peak hour /AADT) of over 0.2. |
