Modelling residual stresses in sand-cast superduplex stainless steel

• Published in: Journal of materials processing technology 2015-03, Vol.217, p.253-261
• Main Authors: Palumbo, G., Piccininni, A., Piglionico, V., Guglielmi, P., Sorgente, D., Tricarico, L.
• Format: Article
• Language: English
• Subjects: Casting; Cooling; Creep modelling; Creep tests; Finite element method; Mathematical models; Modelling; Molds; Residual stress; Residual stresses; Sand; Sand casting; Superduplex stainless steels (ASTM A890 Gr. 5A)
• ISSN: 0924-0136
• DOI: 10.1016/j.jmatprotec.2014.11.006

•Avoiding to model the sand mould negligibly affects FE results.•The creep behaviour is the most influencing factor.•The proposed methodology is robust. A numerical/experimental procedure is proposed for calculating the residual stress state during the cooling phase of the casting process of a superduplex stainless steel (ASTM A890 Gr. 5A). The experimental activity consisted of casting, tensile and creep tests. Casting tests were used to set (by acquiring temperature data at different points) and validate (by measuring displacements after releasing residual stresses by cutting) the Finite Element model. Tensile and creep tests were used to determine the material properties from room temperature to 1200°C. A fully coupled thermo-mechanical analysis was conducted by neglecting the presence of the sand mould; instead, attention was focused on the material by modelling the creep behaviour using the Bailey–Norton formulation. Measurements of the of the displacements due to the stress release after EDM wire cuts revealed to be in good agreement with the numerical model and confirmed the key role played by viscosity during the cooling phase. Neglecting the viscous strain led to an overestimation (more than twofold) of the stress level in the cast part after the cooling phase in the sand mould. The good matching between experimental and numerical data indicated that a numerical model that incorporates the creep behaviour is able to accurately capture the investigated phenomenon, despite the simplification in the modelling of the casting process (sand mould replaced with virtual convection) which does not substantially affect its accuracy. The robustness of the methodology, which is characterized by small computational cost and good quality of results, was further proved simulating and comparing numerical and experimental results concerning a second casting geometry.