The Use of Mixed Element FEA for Predicting Tubular Frame Behaviour
Pushover analysis of offshore framed structures is usually performed on simplified structural models, using finite element analysis. To assess the typical assumptions made concerning joint and member behaviour, various numerical models of increasing complexity are considered in this paper. These range from simple beam models, to models including a detailed replica of the critical joint connected to a beam model, to a complete shell model of the critical compression member and end connections. Phenomenological models of member behaviour, in this case a Marshall Strut representation, are also examined. Numerical results are compared with experimental data from two studies [1, 2]. It is found that detailed modelling of the joint stiffness is sufficient to accurately model the peak frame response, while a detailed model of the complete compression member, which includes cross section distortion, is needed to match the residual strength of the frame. Marshall strut elements are found to provide a robust solution, provided that the variables which describe the model can be defined with sufficient accuracy. Detailed modelling of the frame is also found to be able to provide the required input data for such special purpose elements.