Journals Proceedings

Abstract

Non-linear finite element method (FEM) allows to take into account material and geometrical non-linearities in the simulation of the behaviour of reinforced concrete structures. However, the accuracy of the numerical solution with respect to experimental tests is often questionable, especially in the case of 2D and 3D structures. Several competitions showed in the past significant scatter of the predicted results with respect to the correct ones. Even though internationally well-known computer softwares can be used to predict the structural response, the uncertainty of the numerical simulation cannot be neglected. Therefore, the application of finite element models to the assessment of concrete structures requires a proper investigation of the uncertainty related to the results of the simulations. This paper presents a comparison of numerical simulations of sixteen case studies taken from past experimental tests and modelled with three commercial non-linear softwares. The purpose of the investigation is to show how significant could be the difference between the experimental and numerically evaluated failure load and displacement in function of the code used and the variation of only one material parameter.