Robust Design and Simulation-Based Optimization of Secondary A356 Aluminum Alloy Gravity Die Casting Progettazione robusta e ottimizzazione basata sulla simulazione della colata in conchiglia per gravità di leghe di alluminio secondarie A356

Secondary aluminum can substantially reduce the energy demand associated with primary production, but its variable composition makes repeatable processing difficult. This thesis developed and experimentally assessed an integrated workflow for the robust design of gravity die casting of secondary AlSi7Mg (A356-type) alloys. The aim was to produce cylindrical blanks suitable for machining tensile specimens and extracting metallographic samples, while limiting turbulence, air entrapment, shrinkage, and thermal variability in the central evaluation region. Approximately 36 mold variants were developed in SolidWorks and evaluated in MAGMASOFT. After the initial near-net-shape dog-bone concept showed incomplete filling and feeding, it was replaced by a cylindrical cavity mold manufactured from H13 steel. The numerical workflow combined MAGMASOFT with a transparent Python calculation model, a Design of Simulation Experiments campaign, and genetic-algorithm-based ranking. Fe, Si, Mg, and Mn contents and the initial melt and mold temperatures defined 600 possible combinations; an orthogonal-array design reduced this space to 106 balanced simulations. Predicted ultimate tensile strength and secondary dendrite arm spacing were considered jointly. Within the investigated ranges, mold temperature had the strongest influence on predicted SDAS, whereas Mg and Si showed the clearest positive effects on predicted tensile strength. Thermal measurements were used to refine the casting-mold boundary condition. Six thermocouples positioned 0.5 mm from the AlSi7Mg-H13 interface recorded mold-temperature histories during replicated castings at nominal melt and mold temperatures of 720 °C and 150 °C. A one-dimensional inverse calibration identified a single global temperature-dependent heat-transfer coefficient with effective solid and liquid-side limits of approximately 594 and 614 W m⁻² K⁻¹. The calibration fit produced a global RMSE of 8.25 °C after a +12.70 s time alignment. The identified curve is therefore treated as an effective coefficient for the tested mold, coating, and process window rather than as a universal material constant. The final mold produced fully filled cylindrical castings without visible external short shots or flow interruptions. Optical microscopy of corresponding radial and longitudinal sections yielded mean SDAS values of 11.50, 13.76, and 16.76 µm for representative first-, middle-, and last-ranked scenarios. The 45.7% increase from the first to the last group reproduced the predicted ordering and provided trend-level experimental support for the simulation-guided selection procedure, although it did not constitute point-by-point model validation. Experimental tensile validation remains to be completed. Overall, the study establishes a traceable route from mold design and boundary-condition calibration to simulation-based optimization and microstructural measurement, providing a controlled platform for future tensile testing, broader validation, and data-driven prediction of secondary-alloy behavior.