MODELLING OF INTERSTATE I-465 CRASH COUNTS DURING SNOW EVENTS Mingmin Liu 10.25394/PGS.12249587.v1 https://hammer.purdue.edu/articles/thesis/MODELLING_OF_INTERSTATE_I-465_CRASH_COUNTS_DURING_SNOW_EVENTS/12249587 <p></p>Traffic safety management on interstates is crucial during adverse winter weather. According to the Federal Highway Administration (FHWA), there are over 5,891,000 vehicle crashes each year in the United States. Approximately 21% of these crashes are weather-related. INDOT spends $60 million on winter operations each year to minimize the weather impacts on driver capability, vehicle performance, road infrastructure, and crash risk. Several studies have sought to investigate the relationship of crash counts with weather, speed, traffic and roadway data during snow events, in order to help agencies, identify needs and to distribute the resources effectively and efficiently during winter weather events. The limitation of these studies is that weather variables are often correlated to each other, for example, visibility may be correlated to snow precipitation and air temperature may be correlated to net solar surface radiation. The randomness of crash occurrence also increases difficulty in such studies. In this study, a random parameter negative binomial model was used for Interstate I-465 in Indianapolis in winter 2018 and 2019.The results show that during snow events in Indiana, air temperature, wind speed, snow precipitation, net solar surface radiation, and visibility significantly impact the number of crashes on I-465. Driving over the speed limit (55 mph), especially on wet pavements are more likely to lose control of vehicles and cause crashes. Travel speed between 45 mph to 55 mph and travel speed between 15 mph to 25 mph are both strong factors. Somewhat surprising was that speeds between 25mph and 45mph were not found to be significant. The number of interchanges is also positively related to crash counts due to the high number of conflict points at ramp merging sections. Also, travelling over speed limit is a random parameter with unobserved heterogeneity which is intuitive since speeding could be more dangerous in certain areas with complex road geometry and narrower lanes. Traffic counts have a negative correlation with crash counts, likely due to faster speeds when fewer vehicles are travelling on the loop. Crash counts increased about70% during severe storm days on I-465, and visibility and air temperature are highly correlated to crash counts. These key findings can help the agency to deploy warnings when visibility is low, or temperature falls sharply.help the agency to deploy warnings when visibility is low,or temperature falls sharply. 2020-05-05 20:18:34 winter conditions crash characteristics count models Civil Engineering not elsewhere classified