Data-driven probabilistic braking force model for existing bridges using WIM traffic records

This study proposes a probabilistic methodology for assessing horizontal braking forces induced by vehicular traffic on bridges, addressing the limitations of traditional deterministic models commonly adopted in structural design codes. The proposed approach leverages Weight-In-Motion (WIM) data collected from a provincial road to derive probabilistic distributions of vehicle characteristics, including gross mass, length, and inter-vehicle distance. Through extensive Monte Carlo simulations, realistic traffic convoys are synthesized, incorporating variable deceleration profiles informed by naturalistic driving datasets. The resulting probabilistic braking force model allows estimation of demand levels associated with defined return periods and bridge span lengths, consistent with modern reliability-based design frameworks. Comparative analyses indicate that, depending on span and return period, conventional code-based models may either underestimate or overestimate braking demands. To support practical implementation, simplified equations are also proposed for different design scenarios. The methodology provides a robust tool for the safety assessment and design of bridges subjected to stochastic traffic actions, particularly relevant for both new constructions and the evaluation of existing infrastructure.