Computational thermodynamics in ferrite content prediction of austenitic stainless steel weldments

In this paper, four computational approaches using Thermo-Calc and DICTRA have been used to calculate the ferrite content of a set of austenitic stainless steel welds with different solidification modes and ferrite contents. To evaluate the computational approaches, the calculations were compared to the experimental results. It was found that for each solidification mode, there is one computational approach that predicts ferrite with better accuracy. For ferritic-austenitic alloys, the best accuracy is obtained when considering the peritectic model, with deviations of 1.2-1.4% ferrite. In the case of austenitic-ferritic alloys, the solidification analysed through the eutectic approach showed an accuracy of 0.6-1.6% ferrite, whilst in alloys with fully ferritic solidification, starting calculations, not from the liquid state but from fully ferritic below solidus, was the best approach, showing 2.3% ferrite deviation from the experimental measurements. Computational thermodynamics has proved to be a promising tool to explore simulation and calculation of ferrite content phase fractions in welding. However, further investigation is still needed to correlate the real microstructural features with the computational parameter cell size. The feasibility and accuracy of computational thermodynamics when predicting ferrite in low-heat-input welding processes such as laser welding is also another aspect for additional investigation.