Calibration of inertial consistency models on North Carolina two-lane rural roads

Road crash occurrence is closely related to the geometric design consistency, which can be defined as how drivers' expectancies and road behavior fit. To this regard, the crash rate on a road segment increases as its consistency level decreases. To assess this phenomenon, inertial consistency models were recently developed. These models are based on the difference between the inertial operating speed, which represents drivers' expectancies, and the operating speed, which represents road behavior. The higher the difference between both speeds, the higher the likelihood of crash occurrence. This research aims to validate and calibrate these consistency models on American two-lane rural roads. For this, a total of 194 homogeneous road segments and 977 horizontal curves along 665 km in North Carolina (US) were used. As a result, the geometric design consistency was identified as a major factor of crash occurrence. The higher the difference between drivers' expectancies and road behavior, the higher the crash rate. Likewise, the greater the consistency level, the greater the percentage of horizontal curves without reported crashes. A Safety Performance Function was also calibrated to estimate the number of crashes on a road segment. Consistency thresholds were defined and tested to identify where these crashes are more likely to take place. Finally, the results obtained in this study were compared with those obtained previously on Spanish highways. To this regard, the crash rate on an American highway was 1.85 times greater than those observed on a Spanish highway under the same risk exposure and consistency conditions. Therefore, different tools were developed to enhance the assessment of road safety to the geometric design of both new two-lane rural roads and improvements of existing highways.