Accurate tuning by inverse methodology of the numerical model for efficiently simulating the high pressure die casting process

Abstract

The High Pressure Die Casting (HPDC) process allows to manufacture complex-shaped parts in a very short cycle time. Due to the high rapidity of the process cycle, tool temperatures sensibly fluctuate within each cycle, therefore the prediction of the thermal boundary conditions regulating the heat exchange is of utmost importance.

In this work, the Finite Element (FE) commercial code ProCAST is used to simulate all the steps of the HPDC, from filling to solidification, including lubrication and blowing. The manufacturing of Aluminium window brackets was chosen as the case study; the corresponding numerical model was calibrated using dies’ temperature data (acquired by an infrared camera): the unknown heat transfer coefficients (HTCs) were determined using an inverse approach. In particular, a plan of numerical simulations was arranged according to a CCD scheme and error functions (calculated as difference between calculated and measured temperature data) were used to train accurate metamodels relying on both interpolating and approximating algorithms. A virtual optimization, managed by a multi-objective genetic algorithm, was finally carried out based on the trained metamodels and optimal designs, belonging to the Pareto front, were determined.

Simulations using the results from the optimization round allowed to confirm the effectiveness of the results and, in turn, the effectiveness of the proposed approach.