Improved grain size prediction in aluminium-silicon alloys by thermal analysis

Grain refinement is a common practice in aluminium casting alloys, being Ti and/or B the most popular refining additions. The effectiveness of these additions may be controlled by thermal analysis consisting in analysing the cooling curve during solidification of a sample cast in an instrumented standard cup. This assessment is based on the use of a pre-established relationship between some characteristic parameters of the cooling curve and the microstructure features measured either on the cup or on a standard sample cast at the same time. Previous works showed there is still a need for improving the prediction in the range of small grain sizes in case little or no recalescence is observed. t_{f, Th-Al}, the recalescence time or the time during which temperature remains constant, has been established as an appropriate parameter to take into account high nucleation potential of the melts. This work reports a statistical analysis performed on the characteristic features of the cooling curves of 110 melts of AlSi7Mg alloy. A first analysis gave a simple linear relation between grain size and t_{f, Th-Al} with a correlation coefficient R² of 0.91, that shows a satisfactory agreement for the fine and medium grains (up to 1 mm) while being less good for very coarse grains. However, coarse grains are easily detected; thus, corrective actions can then be taken in order to improve the metallurgical quality in terms of grain size refinement before casting. Further analysis led to express t_{f, Th-Al} as a function of recalescence and of the grain size, its square and cube. The correlation coefficient is much better at 0.96 with an improved description in both small and large grain domains.