Issues on the Correlation between Experimental and Numerical Results in Sheet Metal Forming Benchmarks

Abstract

The validation of numerical models requires the comparison between numerical and experimental results, which has led to the development of benchmark tests in order to achieve a wider participation. In the sheet metal-forming research field, the benchmarks proposed by the Numisheet conference series are a reference, because they always represented a challenge for the numerical codes within the state of the art in the modeling of sheet metal forming. From the challenges proposed along the series of Numisheet benchmarks, the springback prediction has been frequently incorporated, and is still a motivation for the development and testing of accurate modeling strategies. In fact, springback prediction poses many challenges, because it is strongly influenced by numerical parameters such as the type, order, and integration scheme of the finite elements adopted, as well as the shape and size of the finite element mesh, in addition to the constitutive model. Moreover, its measurement also requires the definition of a fixture that should not influence the actual springback and the proper definition of the measurement locations and directions. This is the subject of this contribution, which analyzes the benchmark focused on springback prediction, proposed by the Numisheet 2016 committee. Numerical results are obtained with two di erent codes and comparisons are performed between both numerical and experimental data. The di erences between numerical results are mainly dictated by the ambiguous definition of boundary conditions. The analysis of numerical and experimental springback results should rely on the use of global planes to ensure the objectivity and simplicity in the comparison. Therefore, the analysis gives an insight into issues related to the comparison of results in complex geometries involving springback, which in turn suggests some recommendations for similar future benchmarks.