Driving behavior and risk analysis in urban tunnels with super-large cross-sections

With the development of urban roads, super-large cross-section urban tunnels (SCUT) with four or more lanes appear. However, there is a lack of research regarding the impact of super-large cross-section urban tunnels on driver behavior and associated risk augmentation, leading to insufficient evidence for optimizing traffic facility configurations for SCUT. This paper investigates driving behavior and risk factors in SCUT by driving simulation. The virtual environment for variety urban tunnels is constructed. By leveraging the collected driving behavior data, a driving behavior risk index (DBRI) tailored specifically for tunnels is developed. Moreover, machine learning methods are employed to extract the non-linear relationships among the driving behavior risk index, the number of lanes in the tunnel, the illumination luminance, the color temperature of the tunnel, and the width of the tunnel. The results of the study indicate the following: (1) Compared to tunnels with three lanes or fewer, the risk associated with driving in SCUT is reduced. (2) The color temperature of the tunnel environment has a significant impact on driving behavior, followed by tunnel brightness. Maintaining a high-level range of color temperature in the tunnel environment is beneficial for reducing driving risks. (3) Wider lanes may potentially increase the risk of unsafe driving behavior in SCUT. Finally, based on the results, pertinent recommendations are put forward.