Certainty And Uncertainty Analysis Of Vehicle-bridge Interaction Model

As one of the important loads born by the highway bridges, vehicle load needs to be considered in the design of new bridges, assessment of existing bridges and maintenance of old bridges. The maximum dynamic response of a bridge induced by moving vehicles is usually larger than the maximum static response subjected to the corresponding static vehicle loads. It is widely accepted that this dynamic increment is influenced by many factors, including the bridge span length or natural frequency, road surface roughness, vehicle suspension system, vehicle speed, gross vehicle weight, etc. Since the vehicle-bridge interaction (VBI) problem is complex and field tests are usually inconvenient and expensive to conduct, leading to not enough effective experimental data being available, numerical analysis has become widely applied in this area, and researchers have established different VBI models to investigate the dynamic response of the VBI system. However, to date, little work has been devoted to compare the accuracy, efficiency and suitability of different VBI models comprehensively for different circumstances. On the other hand, among these VBI numerical models, deterministic parameters of vehicles and bridges are usually assumed, and little attention has been paid to the uncertainties, especially the non-probabilistic uncertainties, inherent in the VBI system, which may bias the prediction of vehicle-induced bridge responses by VBI models.Therefore, this thesis investigated the VBI system in terms of certainty and uncertainty analysis. It attempts to provide a better understanding of different VBI models to some extent and act as a comprehensive reference in selecting suitable models. A summary of this work is given in the following:(1) Certainly analysis:different VBI models are first presented. The bridge model, based on a typical two-lane simply-supported slab-on-girder concrete bridge, is represented by a discretized Euler-Bernoulli beam, grillage, assemblage of shell and beam elements and solid elements, respectively. The vehicle model, according to the AASHTO HS20-44, is simulated as a moving-force, moving-mass and spring-damper-mass (SDM) model, respectively. The SDM model incorporates both the single-point-contact (SPC) and multiple-point-contact (MPC) tire models. The road surface roughness is generated by the power spectral density (PSD) provided by ISO 8608, and both fully-correlated (FC) and partially-correlated (PC) roughness profiles are considered. Different types of bridge responses and the tire contact forces are calculated and compared. The influence of different components of VBI models, including the bridge model, vehicle model and road roughness model, on the behavior of the VBI system are studied focusing on the bridge responses. The computational efficiency of different VBI models is also examined. The analysis demonstrates that the accuracy of different types of bridge responses calculated varies with the number of bridge vibration modes used in the simulation. In addition, the type of element used in the bridge model and the vehicle tire model both have a larger impact on the bridge acceleration than bridge deflection.(2) Uncertainty analysis:a method for predicting the bounds of vehicle-induced bridge responses with uncertain bridge and vehicle parameters is presented. The uncertainties in the parameters of the bridge and vehicle are represented with interval variables rather than conventional stochastic variables with known probability distributions. First, a three-dimensional VBI system, which has no closed-form solution and can take road roughness into consideration, is established. Then, by introducing the interval analysis method (IAM) based on the first-order Taylor series expansion, the expressions of the bridge responses, including displacement and bending moment at the midspan, can be explicitly given as functions of the interval parameters, and the lower and upper bounds of the bridge responses are determined by the particle swarm algorithm instead of direct interval arithmetic to avoid excessive overestimation of the responses. The accuracy of the IAM is further enhanced by the subinterval technique. A numerical example is provided, and the results show that, compared with the conventional Monte Carlo method, the proposed IAM is capable of obtaining the bounds of the bridge deflection and bending moment without much sacrifice of accuracy while requiring much less computational effort. This indicates that the proposed method can be successfully, effectively, and efficiently applied to predicting the bounds of the dynamic responses of complicated VBI systems with interval uncertainties. An example is also used to demonstrate the applicability of the IAM to field bridges when only limited information about the bridge and vehicle is available.