Effects of Finite Element Mesh on Numerical Prediction of Ductile Tearing

Abstract

Presently, there is still no generally accepted way of calibrating material models accounting for ductile damage, and calibrating the finite element model with this material in order to obtain a representative length scale via the smallest element size. Roughly, one may identify two approaches. One is based on computational cells, where the cells accounting for damage growth are put in a layer in the prospective crack growth direction. The cell size, D, and initial void volume fraction,f0, are argued to be microstructurally based. The simulations are calibrated against a fracture test. The second approach is to a larger extent based on metallurgical observations, and is more in accordance with the objective that the material model is calibrated with simple test specimens (smooth tensile specimens). But also with this method the finite element mesh has to be calibrated against a fracture test. Hence, the finally chosen element size introduces the required length scale. The present study corresponds to the second approach. It focuses on one set of material parameters established in an earlier study. The material is a typical structural steel in the medium strength range. Results from a purely numerical study are presented for a plane strain three point bend specimen. Effects of finite element type, mesh size, mesh irregularity, and damage material layers are considered. These results illustrate the effects of mesh on ductile tearing prediction. Then 3D analyses of a 3PB test are carried out in order to compare a calibrated element size against the sizes applied in the plane strain study.