Multi-Variable and Bi-Objective Optimization of Electric Upsetting Process for Grain Refinement and Its Uniform Distribution

Abstract

It is significant to adjust the microstructures of preforms in pursuit of high-quality exhaust valves. This work is a novel attempt to identify the optimum process parameters in electric upsetting of 3Cr20Ni10W2 high-alloy to achieve grain refinement and uniform distribution by multi-objective genetic algorithm (MOGA) optimization. A finite element (FE) model on basis of electric-thermal-mechanical and macro-micro sequential multi-physics analysis methods was developed in software MSC. Mar. And different schedules of four independent process variables (heating current (I), clamping length (L), upsetting pressure (Pset) and velocity of the anvil cylinder (v)) were performed aiming to achieve two objective indicators (average grain size (dav) and inhomogeneous degree of grain distribution (σd)). Then, two objective response surfaces were constructed as the functions between the two indicators and the four independent process variables. As per the criterion that simultaneously minimize dav and σd, the processing parameters (Pset, L, v, I) were optimized by MOGA, and corresponding numerical simulation were performed. The results show that both dav and σd are improved significantly at the optimal process condition as verified by the trial-manufacture experiments, which validated the optimal design and corresponding simulation based on grain refinement and uniform distribution by MOGA was credible and effective.