Automatic Identification of Integral Ductile Fracture Parameters Using Autonomously-driven Finite Element Analysis and Application to Tailor-welded Blanks of High-strength Steel Plates

The fracture of tailor-welded blanks (TWBs) fabricated from high-strength steel plates was predicted using integral-type ductile fracture prediction models, specifically, Ayada's model and Cockcroft & Latham's model. The test materials with tensile strengths of 1470 MPa and 440 MPa were laser-welded, and small tensile test specimens with a width of 1 mm were cut from five areas on or near the weld bead. Autonomously-driven finite element analysis (ADFEA) was used to identify the flow curves and critical damage values from the results of small tensile tests. ADFEA is a parameter identification system based on inverse analysis that combines finite element analysis with optimization algorithms driven by machine learning. It is highly effective in reducing the cost of determining simulation conditions for TWBs, which requires the identification of multiple parameters. The accuracy of the identified critical damage value was confirmed through notched small tensile tests. The fracture behavior of the material with a tensile strength of 440 MPa near the weld bead was well reproduced in the FEA.