| Abstract Scope |
Additive manufacturing (AM) offers unprecedented design freedom to explore solutions for demanding engineering problems. However, realising optimal solutions which are often multiscale in nature and, importantly, ensuring their first-time-right fabrication pose challenge. This talk aims to provide greater confidence in the value-proposition of, both data- and physics- based ML, for complex design and process-aware optimisation. The first part of the talk will cover the recent efforts on fusing topology optimisation and ML to enable the rapid discovery of multiscale solutions for challenging multi-physics and multi-objective problems whilst accounting for manufacturability. Insights into how ML based inverse design method help us better exploit architectured materials will be provided. Subsequently, embodiment of AM process via Scientific ML (e.g., physics informed ML) to predicted thermal history for a range of common print scenarios will be presented. Finally, how we enable first-time-right print of high-performance parts through process-aware design optimisation will be elucidated. |