This paper describes a multi-level strategy with increased complexity through four levels of structural analysis of concrete bridges. The concept was developed to provide a procedure that supports enhanced assessments with better understanding of the structure and more precise predictions of the load-carrying capacity. In order to demonstrate and examine the multi-level strategy, a continuous multi-span prestressed concrete girder bridge, tested until shear failure, was investigated. Calculations of the load-carrying capacity at the initial level of the multi-level strategy consistently resulted in underestimated capacities, with the predicted load ranging from 25 % to 78 % of the tested failure load, depending on the local resistance model applied. The initial assessment was also associated with issues of localising the shear failure accurately and, consequently, refined structural analysis at enhanced level was recommended. Enhanced assessment using nonlinear finite element (FE) analysis precisely reproduced the behaviour observed in the experimental test, capturing the actual failure mechanism and the load-carrying capacity with less than 4 % deviation to the test. Thus, the enhanced level of assessment, using the proposed multilevel strategy, can be considered to be accurate, but the study also shows the importance of using guidelines for nonlinear FE analysis and bridge-specific information.

No abstract

Full-scale failure tests of bridges play an important role in the better understanding of the behaviour of bridges and in the development of assessment methods. Such experiments are challenging and often expansive and, thus, are rare. This paper provides a review of failure tests on concrete bridges with a focus on the learning outcomes from the tests. In total, 40 failure tests have been identified for 28 bridges. Various types of bridges have been studied, involving bridge decks composed of slabs, girders and combinations thereof. There are examples of both reinforced concrete (RC) and prestressed concrete (PC) decks. In general, the tests indicated conservative estimates of the load-carrying capacity using theoretical calculations based on methods traditionally used for design and assessment. It can also be concluded that almost a third of the experiments resulted in unexpected failures, mainly shear instead of flexure. It also seems that inaccurate representation of geometry, boundary conditions and materials often explains the differences between the theoretical and the tested capacity.

No abstract