Proxy-Based Models for Bridge Scour Prediction and Safety Evaluation

Abstract

　　In this thesis two models were developed for the bridge scour issue: a bridge scour prediction model and a safety evaluation model Bridge scour is a long-term process that occurs under water and cannot be observed directly This results in sudden bridge failures and disasters 　　A reasonable model for predicting bridge scour and preventing bridge failures is important However because measuring scour is a difficult task there is a shortage of scour records Therefore to establish a bridge scour prediction model this thesis uses rainfall events and variance in riverbed elevations extracted from the literature as proxy indexes and integrates genetic algorithms artificial neural networks and the Monte Carlo method A real case is illustrated and can provide a reference for bridge management authorities 　　Scouring around bridge piers is an important safety issue of bridge management since it can lead to bridge slanting and collapse Since the water flow mechanism around a pier structure is complicated it is very difficult to develop a generic model to determine the scour depth Many researchers have tried to estimate the scour depth around bridge piers by simulating a bridge model with a consideration of various factors such as the depth of the water the average flow velocity and the diameter of the sand However most models require predefined conditions and can only be applied to certain types of bridges 　　In addition the bridge natural frequency as another proxy index is introduced to evaluate the bridge scour conditions and reduce the difficulty in measuring the bridge scour A complicated finite element model and a simple soil spring model which is an improved form of the complicated finite element model are implemented to simulate the bridge structure and compute the bridge natural frequency along the scour depth when considering various environmental variables The complex relationship among factors is modeled using support vector machine and the level of safety can be efficiently determined 　　The results of this study will provide bridge management authorities with efficient and effective methods to predict bridge scour and evaluate the bridge status using proxy indexes to reduce the rate of disaster occurrence from bridge failures

Date of Award	2014 Sept 4
Original language	English
Supervisor	Chung-Wei Feng (Supervisor)