5th International Conference on Advances in Solidification Processes (ICASP-5) 5th International Symposium on Cutting Edge of Computer Simulation of Solidification, Casting and Refining (CSSCR-5)

To unearth the fundamentals in solidification requires the use of advanced experimental methodologies and computational modelling. On the experimental side, synchrotron based high speed X-ray tomography, which can capture a tomogram (3D volume) in seconds, is a powerful tool to study solidification in real-time, allowing the visualization of solidifying alloys as 3D movies or 4D (3D plus time) images. To take advantage of the technique in solidification, we have developed a unique directional solidification furnace coupled with strong magnetic fields, which we integrated with synchrotron beamlines with high speed X-ray tomography (I12-Diamond Light Source and ID19-European Synchrotron Radiation Facility). A large amount of data (tens of TBs) were collected in a series of beamtime experiment. However, it is a challenging task to analyse and correctly interpret the data effectively and efficiently. To provide a practical approach to the problem, we have applied machine learning and computer vision algorithms to automatically process the data. One example will be presented on measuring the kinetics of β intermetallic growth in Fe-containing 319 Al alloy, which demonstrates the advantages of using machine learning in 4D X-ray imaging. with the combined use of machine learning and advanced image quantification, the work reveals the nucleation and growth mechanisms of the intermetallic phase during Al alloy solidification.