Publication: Multi-objective optimization of tensile and hardness properties for austempered ductile iron via NSGA-III and XGBoost
10
0
Issued Date
2025-06-01
Resource Type
eISSN
23524928
Scopus ID
2-s2.0-105002033911
Journal Title
Materials Today Communications
Volume
46
Rights Holder(s)
SCOPUS
Bibliographic Citation
Materials Today Communications Vol.46 (2025)
Suggested Citation
Wiangkham A., Ariyarit A., Aengchuan P., Thammachot N., Wattikornsirikul N., Homjabok W., Sriphen S., Homsantie A. Multi-objective optimization of tensile and hardness properties for austempered ductile iron via NSGA-III and XGBoost. Materials Today Communications Vol.46 (2025). doi:10.1016/j.mtcomm.2025.112458 Retrieved from: https://hdl.handle.net/20.500.14740/20378
Corresponding Author(s)
Other Contributor(s)
Abstract
Austempered ductile iron (ADI) is a versatile cast iron known for its excellent machinability, toughness, fatigue and wear resistance, and high strength-to-weight ratio, making it ideal for applications in the automotive and machinery industries. The mechanical properties of ADI can be optimized through heat treatment processes, where key parameters such as austenitizing temperature, austenitizing time, austempering temperature, and austempering time play critical roles. Balancing these parameters is essential to enhance tensile strength and hardness, while minimizing variability in ductility. This study employs the Non-dominated Sorting Genetic Algorithm III (NSGA-III) to optimize heat treatment parameters for ADI across various conditions. An XGBoost model, trained using experimental data, serves as the multi-objective function for the NSGA-III, facilitating the identification of optimal heat treatment conditions. The findings reveal that increasing the austenitizing temperature and soaking time enhances strength and hardness, while lower austempering temperatures combined with longer soaking times also improves these properties. Statistical analysis indicates that temperature has a more significant influence on mechanical properties than time. The XGBoost model demonstrated high accuracy, achieving an average R² of 0.9686 and a Mean Absolute Percentage Error (MAPE) of 4.90 %. The NSGA-III optimization highlighted trade-offs among strength, ductility and hardness in ADI. For practical applications, it is recommended to austenitize at 850–900°C for 1.5–2 hours and austemper at 250–350°C for 2.5–3.5 hours to achieve optimal mechanical performance.