Abstract: The viscosity of molten metals is a critical parameter influencing melt fluidity, alloy forming quality, and casting performance. It is therefore essential to maintain melt viscosity within an optimal range to enhance the fluidity and mold-filling capacity of molten metals. However, acquiring accurate data on liquid metal viscosity poses significant challenges due to the measurement difficulties and uncertainties in verifying the accuracy of theoretical calculations, hindering in-depth viscosity research. In this review, a systematic introduction of experimental methods for viscosity measurement and theoretical models for viscosity prediction was first conducted, together with an analysis on the advantages/disadvantages of each method/model. Then, all the experimental melt viscosities in unary, binary, ternary, and multi-component systems of light alloys especially aluminum alloys were summarized, and employed to verify the prediction accuracy of different theoretical models, ranging from physical, empirical, to geometrical ones. A comprehensive comparison between the predicted viscosities due to the theoretical model and the experimental data indicated that the CALPHAD approach and the machine learning method should be an effective strategy for predicting accurate viscosities in target alloy melts. Finally, the conclusions were drawn, and the future development directions for efficiently acquiring accurate viscosities in target alloy melts were also pointed out.

Key words: Viscosity, Light alloys, Alloy melt, CALPHAD