Robustness of deep-drawing finite-element simulations to process variations

Robustness of numerical models paves the way for efficient compensation of perturbations resulting in deviations from the nominal conditions. This is critical if the numerical simulations will be used to determine closed-loop process control adjustments to assure the final part quality. This work details the procedure to establish and validate numerical process models, through an investigation of deep-drawing of AA1100-O blanks using 3D Servo Press. Of particular interest is the robustness of the deep-drawing simulation models to different process variations and off-design conditions. The experiments are performed on a 3D Servo Press, used as a conventional press, and equipped with a spring-loaded blank holder. From the experiments, the punch force-displacement as well as local features, i.e., flange draw-in and wall-thinning, are obtained. Two types of finite element models of the drawing process are created, one using shell and the other using solid elements. Correspondingly, the plastic anisotropy of the blanks is modeled using the Yld2000-2d (2D) and Yld2004-18p (3D) yield functions. The friction coefficient between the blank and tooling is inversely identified by comparing the simulated punch force-displacement response, flange draw-in and thickness variations with the experimental ones. The robustness of the numerical and material models is confirmed by process variations on the geometry of the blanks, i.e., an initial offset of blank center and elliptical blanks. However, the wrinkling of the flange due to variation of the blank holder force is not captured by the model. A modification to the model, i.e., by introducing appropriate geometric imperfections to the blank, enables it to predict the flange wrinkling. This work investigates the robustness of numerical models to different types of process variations, which is vital in model-based control analyses.