Premium
Springback Reduction in Tailor Welded Blank with High Strength Differential by Using Multi‐Objective Evolutionary and Genetic Algorithms
Author(s) -
Nguyen NgocTrung,
Hariharan Krishnaswamy,
Chakraborti Nirupam,
Barlat Frédéric,
Lee MyoungGyu
Publication year - 2015
Publication title -
steel research international
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.603
H-Index - 49
eISSN - 1869-344X
pISSN - 1611-3683
DOI - 10.1002/srin.201400263
Subject(s) - blank , twip , multi objective optimization , metamodeling , structural engineering , welding , bending , finite element method , genetic algorithm , pareto principle , reduction (mathematics) , engineering , algorithm , materials science , computer science , mathematical optimization , mathematics , mechanical engineering , composite material , crystal twinning , geometry , microstructure , software engineering
The springback behavior on two sides of TWIP (twinning‐induced plasticity) and mild steels tailor welded blank (TWB) is considerably different due to the large difference in strength. To reduce springback defects in a U‐draw bending process of such TWB, different non‐constant blank holding force (BHF) profiles on two sides of the blank are applied in this study. A systematic approach to obtain optimal BHF‐stroke profiles, which helps to reduce the springback is proposed. Control of the variable BHF aims at increasing the uniformity in through‐thickness of the stress component along the stretching direction, while decreasing the risk of failure due to over‐stretching. Therefore, the optimal condition would require satisfying two conflicting objectives simultaneously: (i) minimize springback deformation and (ii) minimize the forming severity, leading to a Pareto‐optimal problem. The optimization procedure consists of sampling design, finite element (FE) simulations, metamodeling, and finally the calculation of a Pareto‐frontier. Generated outputs from FE simulations on statistically significant sampling points are used for the construction of metamodels of optimum accuracy and complexity, which, in turn, are used to evaluate the output for any set of inputs, replacing the computing intensive FE simulations. A novel genetic algorithms based multi‐objective optimization technique is subsequently applied for optimization.