Dr. Aleena Baby, from the Access Process Simulation Group, presented the latest work, “Smart Casting: How AI-Driven Defect Detection Is Redefining Manufacturing,” at the AI Meetup in Basel, highlighting how artificial intelligence is transforming defect prediction in metal casting.
The presentation introduced PorosAI, an AI-based system designed to predict casting porosity during the design phase itself, enabling engineers to identify failure-prone regions before tooling, pouring, or physical trials take place. Although this approach is still under development and has so far focused on a specific application, it could represent a significant change in pre-engineered, data-driven decision-making. Conventional porosity prediction relies on computationally intensive simulations that can require one hour to an entire day per design iteration, significantly limiting design exploration. PorosAI overcomes this bottleneck by delivering porosity predictions in under one second, allowing engineers to evaluate hundreds of design alternatives in real time.
By integrating AI directly into the casting process chain, this technology enables:
- Early detection of defect-prone regions prior to production
- Reduced scrap rates, rework, and late-stage design changes
- More sustainable and resource-efficient manufacturing practices
The research was carried out within several major funded and industrial initiatives, including:
- BMWK project “InDiPro” (Nr. 20N2203D)
- Siemens PLM cooperation (Agreement Nr. 60068580)
More details about the project can be found on paper: link to paper