Determination of interfacial heat transfer coefficients in a sand mould casting process using an optimised inverse analysis

Determination of the interface heat transfer coefficients in casting processes represents a fundamental step in the creation of a reliable numerical model that can predict some of the most common defects (for example hot tear or residual stress) which may affect the process. This work focuses on the most appropriate methodology to determine the heat transfer coefficients for the numerical modelling of the casting process of a super duplex stainless steel (ASTM A890 Gr. 5A) using a silica sand mould. Experimental instrumented castings were used to acquire (both in the casting and in the sand) temperature changes for a number of points of interest by means of thermocouples; in addition, the entire process was simulated using the finite difference method commercial software package MAGMASOFT® (v. 5.2) in order to calculate the temperature at the same points at any step of the process. Firstly, the most influential input parameters were chosen in order to determine the factors to be investigated using a reduced factorial scheme. Further simulations, in which the value of the chosen factors was changed, made it possible to create response surfaces using as a response variable the value of the mismatch between the experimental and numerical temperature changes in the same points. Finally, the optimisation procedure using a multi-objective genetic algorithm was performed, with the goal of finding optimal values for the input parameters, i.e. those with which the mismatch between experimental and numerical temperature changes is minimised. As confirmed by the numerical simulations using the results of the optimisation procedure, the methodology proposed enabled us to determine the correct values of the input variables for modelling the casting process of the ASTM A890 Gr. 5A.