A data-based approach for predicting hardness in friction stir additive manufacturing of EN AW-7075-T6

Abstract Friction stir additive manufacturing (FSAM) is an emerging additive manufacturing process for fabricating large aerospace components made from aluminum alloys (e.g., EN AW-7075-T6). In the studies presented in this article, the hardness losses typical for this process were addressed. The objective of the investigations was to model these hardness losses using non-numerical methods, thereby increasing their applicability to industry. It was demonstrated that the hardness of the topmost layer was similar across parts with different numbers of layers due to the application of a temperature control system. A time-temperature metric (called cumulated hold time) was defined to predict the mean hardness of the underlying reheated layers (correlation coefficient of − 0.951). Using the lumped capacitance method, the cumulated hold time was modeled from the energy per length during welding. The developed submodels enable prediction of hardness profiles with an RMSE of 2.3 HV 0.2 and identification of suitable process parameters without measuring the part’s temperatures during fabrication. Experimental investigations showed that the welding temperature had a small positive effect on the hardness of the top layers and a negative one on the bottom layers; however, the positive effect of the welding speed on the hardness of all layers was dominant.