Current design guides adopt a deterministic approach to the design of horizontal curves; each factor included in the design is represented by the near-worst-case value. In the context of horizontal curve design, the design procedure is based only on the driver comfort criterion, and data correspond to experiments conducted in the 1930s. Furthermore, current horizontal curve design procedures lack a quantitative evaluation for safety. To overcome those shortcomings, a new design framework is proposed to design horizontal curves; a probabilistic approach is adopted and two criteria are considered: vehicle dynamic stability and driver comfort. Reliability analysis was used to provide a quantitative evaluation for the design in regard to the probability of failure, probability of noncompliance, and reliability index. Outputs of simulation runs in a vehicle dynamics model were used to estimate demand lateral friction and lateral acceleration depending on the geometric characteristics of horizontal curves. In addition, data of an instrumented vehicle experiment were used to develop driver-level models for the distribution of the curve speed and driver comfort threshold. The first-order reliability method was used to estimate the probability of failure, probability of noncompliance, and reliability index. The proposed design framework and developed models were applied in an example to design a horizontal curve for a specific design speed.

Additional Metadata
Persistent URL dx.doi.org/10.3141/2588-03
Citation
Dhahir, B. (Bashar), & Hassan, Y. (2016). Reliability-based design of horizontal curves on two-lane rural highways. doi:10.3141/2588-03