Determination of Johnson-Cook material constants for Copper using traction tests and inverse identification

Abstract

Copper and its alloys have specific properties - electrical and thermal conductivity, ductility, mechanical strength, corrosion resistance - which make it a material used in various industrial applications. However, these materials are difficult to weld by conventional processes due to their high thermal conductivity and high rate of oxidation at temperatures close to melting. To be able to join copper-based materials, friction stir welding seems to be a promising possibility. The quality of the assembly obtained strongly depends on the parameters of the process chosen to generate a temperature close to the optimum value. So, to find these parameters, it is necessary to carry out experimental test campaigns which are generally very expensive in terms of time and equipment. An alternative will be to use the numerical simulations of the FSW process to find the optimal parameters. Numerical simulation contributes also to a better understanding of the influence of input parameters on the phenomena in the process and the connections made. To develop a numerical model, it is necessary to use a constitutive equation that defines the behavior of the material throughout the process. The most commonly constitutive equation used for FSW modeling is Johnson-Cook. The paper describes the strategy of identifying the constants of the material, respectively the method of inverse identification of the parameters. The constants of the thermo-mechanical behavior (both elastic and plastic) are identified by coupling the finite element method with Abaqus and Matlab software. This strategy tries to minimize the quadratic deviation between the response of the model and the experimental tests. Tensile tests were carried out on quasi-pure copper CU-DHP samples, at temperatures of 22°C, and 500°C, for speeds of 3 mm/min and 30 mm/min, respectively. The validation of the model thus identified was carried out on tensile tests at a temperature of 300°C, by comparing the curves obtained with the Johnson-Cook model and those obtained experimentally.