A QUANTIFICATION OF MODEL UNCERTAINTIES IN NLFEA OF R.C. SHEAR WALLS SUBJECTED TO REPEATED LOADING

The non–linear finite element method allows to perform the prediction of the complex behaviour of reinforced concrete members from serviceability to ultimate conditions. The non-linear finite elements analysis can be successfully employed to solve a wide field of problems in structural engineering. In this context, in order to adopt NLFEM software for structural safety verification the level of uncertainties presented by such kind of analysis have to be deeply quantified. Appreciable results may be achieved simulating the response of structural components subjected to progressive incremental loading process. However, as in the case of seismic assessment of R.C. members, the same results may not be reached under repeated loading configuration. In this paper two commercial non-linear finite element codes have been considered in order to assess the uncertainties in the simulation of non-linear concrete behaviour under repeated loading by adopting plane stress models. The resisting model uncertainties (intended as the ratio between experimental and NLFEA outcomes) in reproducing the experimental achievements related to six laboratory tests on shear walls with results known from the literature have been assessed. The overall capability of the two codes to predict the actual structural behaviour under cyclic loading is commented and mean value, variance and coefficient of variation of model uncertainties is quantified and discussed.