Advanced Experimental Data Processing for the Identification of Thermal and Strain-Rate Sensitivity of a Nuclear Steel

The knowledge of material mechanical behaviour in different physical conditions is necessary to accurately simulate structural response using finite element methods, especially when complex physical processes, such as strain-hardening, large strains etc., are involved. In this context, the material characterization at different temperatures and strain-rates is indispensable, but it is equally essential to properly transform the test data into efficient constitutive equations capable to accurately reproduce the material response. As an alternative to the conventional analytical approach of the stress–strain curve fitting, this investigation examines the adoption of an inverse method that exploits a FEM model to accurately keep account of the specimen stress, strain, and temperature fields. The material parameters of the selected constitutive model are then obtained by using an optimization algorithm that iteratively changes the parameter values to minimize a target function. The algorithm has been implemented in MATLAB using the LS-DYNA FEM solver. In the paper, this method has been applied to the experimental data produced in a test campaign (EU project LISSAC) for a ferritic steel normally employed in nuclear pressure vessels. These data refer to tensile testing under several strain-rate and temperature conditions and include both smooth and notched cylindrical specimens. The constitutive models of Johnson–Cook and Zerilli–Armstrong have been considered for the demonstration of the methodology. The efficiency of the approach in determining the model parameters is critically assessed.