# Stochastic Vehicle Load Simulation for Small- and Medium-Span Bridges Based on Weigh-in-Motion Monitoring

## TL;DR

This paper introduces a new method to simulate vehicle loads on small- and medium-span bridges using real data and Monte Carlo techniques to improve bridge safety assessments.

## Contribution

The novel contribution is a stochastic vehicle load simulation method based on Monte Carlo sampling and weigh-in-motion data for improved regional load modeling.

## Key findings

- Vehicle weight on the bridge follows a multi-peak Gaussian mixture distribution with similar peak magnitudes but varying frequencies.
- Axle load distribution is single-peak Gaussian with minor differences in peak values and frequencies.

## Abstract

Vehicle loads constitute the dominant source of dynamic excitation for small- and medium-span bridges, exerting a critical influence on bridge safety and service performance. However, vehicle load characteristics exhibit pronounced temporal variability and strong regional heterogeneity, which poses challenges for accurately characterizing the in-service loading conditions of bridges in specific regions using conventional dynamic load models. Therefore, this study focuses on the actual operational characteristics of vehicles on the Lieshihe bridge and the effects of vehicle loads and proposes a stochastic vehicle load simulation method based on the Monte Carlo sampling technique and weigh-in-motion (WIM) measured data. Initially, the recorded vehicle data are classified into representative vehicle models, and statistical analyses are conducted to characterize lane-dependent traffic flow variations and the occurrence patterns of vehicle overloading. Subsequently, axle number and axle spacing are selected as the core indicators for vehicle classification, based on which vehicles are categorized into five representative vehicle types. The changing patterns of axle load, vehicle weight, vehicle speed, etc., for each vehicle type are studied, and corresponding probability density distribution models are established to describe the stochastic nature of vehicle characteristics. Finally, using the Monte Carlo method combined with important attributes of vehicle flows, a stochastic vehicle load model is established based on the spatial–temporal characteristics. The results demonstrate that the vehicle weight on the bridge exhibits a Gaussian mixture distribution with multi-peaks, characterized by similar peak magnitudes but markedly different occurrence frequencies; axle load shows a single-peak distribution of Gaussian distribution with small differences in peak values and frequencies.