Sensitivity of a Multi-Physics DC Casting Model to Material Database Selection: Implications on Hot-Tearing and Cold-Cracking Prediction in an AA7050 Alloy

Abstract Numerical process simulation is increasingly used to optimize direct-chill (DC) casting of aluminum alloys, yet crack-susceptibility predictions depend critically on the constitutive–property databases employed. This study presents a sensitivity analysis of the ALSIM DC casting model with respect to material database selection, using AA7050 billet (315 mm diameter) casting as a benchmark. Three casting scenarios—hot tearing (HT), cold cracking (CC), and healthy billets—were simulated using five databases spanning semi-solid and fully-solid regimes: Al-2 wt.% Cu, AA7050, and artificially strengthened AA7050XTR semi-solid databases, alongside AA1050 and AA7050 fully-solid (hardening) databases. The results show that the fully-solid database dominates the first principal-stress field, which governs the critical crack size (CCS) used as the CC susceptibility indicator, while the semi-solid database has limited effect on the integrated critical strain (ICS) used as the HT indicator. This insensitivity may arise because accumulated strain in DC casting is governed primarily by process conditions rather than material properties, revealing a fundamental limitation of strain-based HT criteria for alloy-to-alloy comparison. To address this limitation, an improved version of reserve strain value (RSV) concept is discussed as a complementary, material-sensitive HT indicator that may support future development of coupled HT-CC assessment methods. Practical recommendations for industrial database development are provided.