| description abstract | The present highway geometric design is governed by design speed and considers all other variables such as terrain, roadway, and driver characteristics as deterministic. The behavior of road users can be impulsive, and there is always an element of uncertainty; it requires the introduction of reliability ideas into highway design. Reliability analysis is one method that explains the uncertainty in the geometric design process and assesses the risk related to a particular design feature. The study selected 38 curves with variable geometry located along National Highway 953 (NH-953) in India’s hilly terrains. A total of 13,680 samples of speed data (360 data points for one curve) were collected using the radar gun at the curve midpoint. The first-order reliability method was used to calculate the reliability index (β) and probability of noncompliance (Pnc). The results showed that the β of a car is less than heavy commercial vehicles (HCV), and the corresponding Pnc was higher for a car than HCV. Since HCV is moving slower than a car, the safety margin (difference between design and vehicle mean speeds) is more, and Pnc is less. Further, the vehicle mean speed increases, the safety margin decreases, and the resultant β of the curves decreases (Pnc increases). Therefore, the chances of crashes are increased. Also, cars have 30% to 32% higher Pnc than HCV. Additionally, the Tobit regression approach was used to develop a β model. The results revealed that deflection angle is positively correlated while curve length, tangent length, Pnc, and total crashes negatively correlate with β. Lastly, the developed safety warrants are helpful for the highway authority to identify the unreliable highway segment and provide safety measures to increase the reliability of the segment and reduce the likelihood of crashes. | |
