KSU Researcher Develops Crash Prediction Model for Rural Highways
Date Posted: April 18, 2016
Posted In: Ideas,
From Traffic Technology Today: A Kansas State University (KSU) researcher is partnering with the Kansas Department of Transportation (KDOT) to develop a state-specific crash prediction model for rural multilane highways.
Syeda Rubaiyat Aziz (below), a doctoral candidate in civil engineering at KSU, started her research by calibrating the American Association of State Highway and Transport Officials’ (AASHTO) Highway Safety Manual methodology. She is using Google Maps and KDOT video logs to examine each roadway segment and intersection to obtain the number of crashes and a variety of other factors, including the presence of lighting posts, driveway density and roadside hazard rating. The roadway geometric data was provided by KDOT, which is funding the research project.
Aziz has also developed a new Kansas-specific crash prediction tool. Her research indicates that this model performs better than the existing High Safety Manual in predicting crashes in the state.
Aziz categorizes crashes into three severity levels: fatal crashes, injury crashes and property damage-only crashes. She further divides injury crashes into three subcategories: incapacitating, non-incapacitating and possible injuries. Aziz foresees a two-step use for her research results. First, her findings are useful in identifying the most hazardous or unsafe segments and intersections of rural roadways. Then, the conclusions can direct KDOT in finding suitable countermeasures, and thereby prioritizing requests within the state’s transportation budget. Aziz plans to complete the KDOT-funded project and obtain her doctoral degree in May. Upon completion, KDOT will use Aziz’s research findings and readjust the model to current data for future projects.