FE modelling of the bond-slip behaviour and its effects on the seismic performance of RC framed structures

The behaviour of reinforced concrete (RC) structures is largely dependent on the strength and ductility of its beam-to-column joints. This is particularly true when the structure is subject to seismic loading. An important and essential factor that influences the behaviour of joints is the bond interaction between concrete and reinforcement steel. The performance of RC joints is commonly investigated using laboratory tests on simple subframes. This is a valuable way for assessing certain aspects but has two main drawbacks. First, it is not always possible to test full size RC joints because of cost and construction difficulties. Secondly, the performance of a joint in subframe is not representative of the behaviour of the same joint as a part of a full framed structure. The use of numerical analysis provides an efficient means for extending the scope of laboratory testing. Indeed, joint models, with various degrees of accuracy, are already described in the literature. Surprisingly, very few of these models, if any, considered the influence of bond slippage or the curtailment of reinforcement bars within beams. In this research, the effects of bond slippage on the moment capacity and ductility of RC frames has been investigated. A novel finite element (FE) approach has been developed for modelling bond behaviour. The approach is based on the use of cohesive elements to represent the interface between the main steel bars in beams and the surrounding concrete. The mechanical properties of the cohesive elements are based on assumptions made regarding possible modes of failure at the interface, and on the properties of the concrete mix. This has the advantage of determining the parameters of the cohesive elements without the need for ad-hoc laboratory tests, as suggested in the literature. The bond model has been successfully implemented using the general purpose FE software Abaqus. It has been rigorously verified and validated using the results of standard experimental tests, reported in the literature, for determining the bond strength, that is, the pull-out and beam-bond tests.The bond model has been used in a FE model for analysing RC beam-to-column joints. This model too was rigorously validated using various test results reported in the literature. However, unlike existing models, the important effects of bond slippage and curtailment of reinforcement steel have been included. As a result, a high degree of accuracy was achieved when comparing test and analytical results. As a preliminary step for studying full RC frames, an extensive parametric study was performed on a beam-to-column joint designed to EN 1998. The studies showed that slippage has a direct effect on the moment capacity and ductility of joints leading to significant reduction in both properties. In addition, moment-curvature curves were produced such that the effects of bond slippage and steel curtailment were included. To extend the analysis to full frames, an approach for converting curvatures to rotations, allowing for spread of plasticity where applicable, has been developed. The thus obtained moment-rotation curves, including effects of bond slippage, have been used for defining the rotational behaviour of joints in full RC frames. Investigation of the behaviour of full frames has been conducted within the guidelines of Eurocode EC 8. Therefore, the performance-based design approach, defined in the code, has been used to compare the response of frames, subject to lateral loads, where the rotational capacity of joints was modelled using various moment-rotation curves. The studies showed that determination of the target displacement, at which performance of a frame is to be assessed, is not sensitive to the assumed rotational behaviour of joints. However, it was demonstrated that bond slippage does influence the overall capacity and ductility of full frames. In addition, it was evident that spread of plasticity is affected in such a way that the performance region of a given frame may be influenced. In fact, in some cases the target displacement was very close to the failure region which is not recommended by the code.