Modelling the interface in concrete-to-concrete connections between precast girders and cast-in-situ top layers

Abstract

A significant part of composite concrete viaducts was built in the 50s and 60s. This has led to an increasing need for the reassessment of the precast concrete beam bridges in the Netherlands. Many of those bridges are nowadays subjected to higher traffic loads and consequently, have a certain risk of not complying with the currently used design codes. As a result, there is a demand for the development of nonlinear finite element solution strategies, which would increase the reliability of the numerical methods in safety assessment. In this research, special consideration is given to the precast, prestressed beam bridges which were made continuous by applying cast-in-situ cross beams and a top layer above the supports. In such structures, the hogging bending moment introduces more complex stress conditions in the concrete parts and more specific at the interface between precast girders and the top slab, compared to simply supported beam bridges.
Thus, one of the key aspects of the structural performance of composite bridges is the interfacial behaviour. The focus of this research is to study the stress conditions in the vicinity and at the interface and explore methods of numerical modelling of the interface in concrete-to-concrete connections between precast beams and top layers to initiate the development of modelling strategies for this type of interfaces.
The literature review was focused on prefabricated beam bridges, the current state of knowledge on concrete-to-concrete interfaces, along with design recommendations and past experimental and numerical research. Moreover, available interface element types, material models and modelling guidelines were explored. Since DIANA FEA is used within the course of this research, the study of the available models was limited to the ones provided by this software. It was noted that the Linear Elasticity model is the simplest way of interface modelling, therefore it was utilised in the initial stage of the research. More advanced models, Coulomb Friction and Combined Cracking-Shearing-Crushing, were considered worth investigating owing to accounting for coupling between normal and tangential behaviour. The Nonlinear Elasticity material model was also recognized due to the introduction of nonlinear effects, yet being relatively simple to assemble.
The initial phase of the research was a linear, phased analysis of the continuous, composite, concrete girder. Three models were tested within this part of the research – the model without interface elements, and two with linear elastic interface elements, one having high, penalty stiffness and the other having lower, more realistic value of shear stiffness. It was verified that the models without and with penalty stiffness interface performed almost equally. The decrease in stiffness and the deterioration of the composite action caused by this, resulted in an increase of stresses in the precast element. By the support, the extreme tension raised by a factor of 1.21 and under the point of load application the compressive stresses in the beams’ web elevated by 2.26. Based on the linear analysis, no significant tensile stresses perpendicular to the interface were detected. According to the analysis of interfacial stresses interaction and assumed failure envelopes, at four chosen points - above the support, at midspan of the main span, at the local shear extreme and under the point of load application - it was observed that the point above the support is not at risk of failure, whereas the point in the midspan might be. It was concluded that the combination of stresses is relevant not only because of a possible decrease in capacity due to tension but also increase under compression. As a result, models accounting for coupling between normal and shear tractions and relative displacements are worth investigating. It was also observed, that cracking in concrete elements by the support is expected, hence nonlinear analysis is required.
The component-level experiments found in the literature were analysed in the following section to be able to perform verification study of Coulomb Friction (CF) and Combined Cracking-Shearing-Crushing (CCSC) interface material models. Based on single element FE tests it was concluded that both material models proved to be well-suited for capturing the shear-normal stresses coupling. With the same input parameters, but higher normal pressure, the shear capacity increased, representing well the reference data. The CCSC interface material model’s ability to capture both cohesion and friction softening, was also verified with the single element models. Moreover, tension softening based on mode I fracture energy can be accounted for in that material model, as well as the fracture energy’s and dilatancy’s dependency on confining stress. However, those parameters were not verified, due to, among others, limited experimental data. Element assembly with the CCSC material model for the interface, circular beam bond-slip reinforcement and nonlinear material properties of concrete, was used to analyse the specimens with rebars crossing the interface. This approach, was assumed to represent the force transfer mechanisms to the highest extent, since cohesion and friction, generated by both external pressure and reinforcing bars, along with their softening, as well as dowel action, can theoretically be represented by such model. It was observed that this type of strategy resulted in convergence issues, and due to large number of input parameters it is quite complex to analyse or further calibrate. However, the approach seemed promising since the peak loads were underestimated by only 7-15% with respect to the mean, experimentally obtained values.
In the final Chapter the Combined Cracking-Shearing-Crushing (CCSC) interface material model, with bond-slip beam reinforcements was applied in the nonlinear analysis of the previously analysed composite girder. As an alternative, the model with the Nonlinear Elasticity(NE) interface material model was also constructed, based on the analogous input parameters, to be able to compare the modelling methods. In total four models were analysed, since two sets of input, one based on Eurocode 2 and the other on best guess stemming from literature findings, were studied. What was found to be promising is that the global behaviour, assessed on the basis of crack patterns, of the beams with corresponding input, was quite similar for the analyses with the CCSC and the NE material models. With the applied numerical setup, it was not possible to obtain the total load-displacement path of the composite beams using the CCSC material model for the interface, since the models diverged. The NE material model performed more stable and allowed for the analyses to continue, which is its main advantage. Another benefit is the ease of assembly, in comparison with the CCSC model. Nevertheless, it was demonstrated that the NE might provide overestimated results due to not considering the interaction of tractions. It was highlighted that the models’ validation with experiments is needed to recommend one of the models or either of the input sets. It was recommended to simplify the approach with the CCSC material model, by for instance, simplifying the numerical setup of interface reinforcement. Moreover, according to the literature findings the scatter of cohesion and friction coefficients, as well as other input parameters, is still quite large, thus experimental research in the form of push-off tests focused on those, particular interfaces is recommended.