Studies Of Existing RC Bridge Safety And Durability Evaluation Methods

Recently, bridge deterioration with time and ever increasing traffic loads raise concerns about safety and durability of aging bridges. The management and optimum repair strategy of highway bridges have earned world-wide attention. In order to keep pace with the growing demand for deteriorating bridges with financial resources, decision makers have to evaluate the safety and durability of the aging bridges correctly. It is based on this, optimum repair strategy can be made. Against the puzzles in safety and durability evaluation and repair decision of existing bridges, incorporating the key technical item of the ministry of communications of Liaoning Province (Grant No. 0101), this paper performed the following studies:1. 8 old RC beams replaced from existing bridges are tested under three-point loading. In order to research the difference between old and new RC beams, contrast test of 5 new beams is complemented. The section strains, deflection, maximum crack width, maximum crack height of old and new RC beams under different levels of loads are analyzed. Reduction coefficient of bearing capacity of old RC beams having worked 35 years is obtained through comparing old RC beams with new RC beams.2. Based on the shear testing data, the influence of cycle loading after presplitting on shear capacity and shear crack width of RC beams is analyzed. The corresponding relationship between visual damage characteristic (crack width ) and loading rate (ratio of load against failing load) of RC beams under cycle loading after presplitting is also be analyzed. The relationship can provide evaluating shear bearing capacity of RC beams through visual damage with reference.3. The relationship between apparent damage characteristic and loading-bearing rate is very complex for reinforced concrete beams. It is very difficult to be described by mathematical function. This paper applies Artificial Neural Networks technology to build the complex non-linear relationship between apparent damage characteristic and loading-bearing rate. BP Neural Network model was built to inversing the loading-bearing rate of reinforced concrete beams using apparent damage characteristic. Emulation results approach test results. It is proven that the artificial neural network approach is a new applicable method to evaluate the loading rate and safety of reinforced concrete beams.4. This paper establishes a safety evaluation model of existing reinforced concrete bridges based on Analytic Hierarchy Process. In order to combine quantitative factors evaluation with qualitative factors evaluation, Fuzzy Synthetic Evaluation method is used to evaluate bridges safety, in which experts' experience can be used sufficiently. The mainprocess and essential steps are introduced in detail. A practical application is illustrated herein, which shows that the proposed method is feasible and reasonable. The proposed safety evaluation model can be used to evaluate the safeties not only of the whole structure, but also of every component. And thus, Fuzzy Integrated Evaluation Method can be used to evaluate bridge safety.5. By merging fuzzy inference, neural networks and genetic algorithms, this paper constructs an expert system to evaluate durability of bridges based on ANFIS and genetic algorithms. In the proposed expert system, the inference mechanism used by experts for evaluating bridges are implicit in the connections and weights of the neuro-fuzzy networks. Witch successfully avoid the neural networks becoming a "black box" and is easy to be comprehended by users. Furthermore, the proposed expert system has the positve attributes of adaptation and learning. In addition, GA-BP hybrid method is used to train neural networks, this method takes advantage of the strength of genetic algorithms and back propagation (the fast initial convergence of genetics and the powerful local search of back propagation). Finally, the evaluation results of 13 existing bridges in Lioaning Provice demonstrate the effectiveness and practicability of the proposed neuro-fuzzy expert system.