Data-driven Models For Shrinkage Porosity Prediction In Aluminium Casting
Please login to view abstract download link
Numerical simulation plays a crucial role in manufacturing processes optimization by providing valuable insights and predictions, enabling manufacturers to improve efficiency, reduce costs, and enhance product quality. Aluminium casting is one of the manufacturing processes that can be optimized through the use of advanced simulation techniques to avoid the presence of defects in the final product like porosity. Nevertheless, the computational time required for numerical simulations can be a major challenge, as the time cost can be greatly variable depending on the size and complexity of the system being modeled, the used simulation method and the available computational resources. Nowadays, Artificial Intelligence is shifting the adopted technologies in industry by offering a wealth of techniques that may help to circumvent this problem. Data-driven techniques can be used as an alternative approach for predicting porosity in aluminium casting relying on few data from a casting simulation software. In this work, Data-driven approaches are explored with a focus on its ability to build a real-time porosity predictions in casting simulations. Supervised learning methods and Optimal Transport are adapted to our application to consider the gradient of temperature in the solidification phase and the shape of the casting part. The proposed approaches are tested on a test case and the obtained results are promising compared to the simulated porosity distributions.
complas_sec@cimne.upc.edu/ Telf. + 34 - 93 405 46 94
Copyright © 2022 CIMNE, All Rights Reserved. Terms of service
