will propose a doctoral thesis entitled,
Multiscale Machine Intelligence Tools to Accelerate Polymer Additive Manufacturing Design
On
Wednesday, November 13th at 1 p.m.
MRDC Room 4404
and/or Virtual via Teams
Committee
Prof. Rampi Ramprasad (Advisor) – School of Materials Science and Engineering, School of Computational Science and Engineering
Prof. H Jerry Qi – School of Mechanical Engineering
Prof. Aaron Stebner – School of Materials Science and Engineering, School of Computational Science and Engineering
Prof. Victor Fung – School of Computational Science and Engineering
Dr. Ehsan Haghighat – Head of Machine Learning, C-infinity; Software Research Scientist, Carbon3D
Abstract
The modern manufacturing landscape is transforming, driven by the need for mass customization and sustainability in producing complex, multi-material structures. Traditional manufacturing methods struggle to meet these demands, especially for intricate designs. Additive Manufacturing (AM), particularly polymeric 3D printing, offers a promising solution by building products layer by layer, allowing for reduced material waste and greater design freedom. However, the vast design space in AM—including ranges of material chemistries, processing conditions, and component design—poses a significant optimization challenge.
I propose that we can use computational tools that leverage machine learning (ML) and materials informatics to accelerate the AM design process in a hierarchical manner, across the many length scales of design. In the first stage, active learning algorithms, integrated with molecular dynamics simulations, and graph neural network diffusion model surrogates, can help explore the vast combinatorial space of thermoset photopolymer acrylates. This framework will be deployed as an autonomous decision-making system for the experimental lab. The next stage is using domain-informed ML algorithms to model polymeric materials, namely the melt viscosity, across chemical and processing domains. I demonstrate that domain information in machine intelligence is crucial to model unseen spaces. Finally, the optimization of lattice structures in component design is addressed by introducing an ML surrogate that predicts mechanical responses in AM components based on finite-element simulations.
Collectively, these tools lay a foundation for multi-scale informatics-driven AM design, enabling faster and more informed decision-making. This increases AM's adaptability and scalability, paving the way for innovative, customized products in a sustainable manufacturing ecosystem.
]]>