Mechanical design of ring tensile specimen via surrogate modelling for inverse material parameter identification

Abstract

The mechanical characterization of anisotropic thin walled-tubes along hoop direction is not a trivial task. It is necessary to develop experimental techniques, numerical methods and design test samples, which enable to determine the real tube properties along hoop direction without any external influences. In this study, first we propose a surrogate based-model for the mechanical design of the ring hoop tensile test (RHTT) specimen, in order to obtain the effective homogeneous stress and strain distribution of the uniaxial tensile test along hoop direction. Second, the optimized sample is used to carry out RHTT and to obtain the actual flow stress curve and the anisotropy coefficients of AA6063-O extruded tube. However, the experimental curve measured from RHTT (force –displacement) is a degenerate response, since it suffers from intermixture effects of the effective material behaviour with the friction between the sample and the sample-holding tool. Hence, we developed an inverse parameter identification method, which uses design of experiments, finite element analysis and artificial neural network to separate out the tubular material parameters from the friction coefficient. The assessment of the developed method is achieved by comparing the predicted material parameters and the identified flow stress curve obtained by artificial neural network algorithm. The finite element simulation results corroborate the obtained findings.