Accelerated Multi-Modal Manufacturing Optimization
Brian Giera | 20-ERD-036
The characterization of manufactured parts is often a serial process, where different post-processing steps and quality measurements are obtained in a non-co-located and non-automated fashion. Thus, the development cycle (e.g., part specification, fabrication, and qualification) is subject to bottlenecks, making part repeatability difficult and costly to achieve, quantify, and optimize. This is true for established manufacturing processes and especially true for emerging advanced manufacturing (AM) systems that are the focus of this project. The goal of this project is to develop and leverage extensive on-board sensing and adaptive tool-path planning to accelerate part and process qualification of AM parts via characterization and complementary advanced analytical techniques. Throughout this project, significant hardware and computational infrastructure was developed and deployed that will provide new capabilities with significant programmatic impact. Specifically, we created (1) an interactive database that logs all specification, production, and inspection data for parts, (2) a universal method to compare parts from ISO standard STEP files, and (3) early-stage digital twins of fabrication, photogrammetry, and structured light. Furthermore, all of these capabilities are applicable to traditional and advanced manufacturing, have already yielded programmatic impact in the form of new projects, and resulted in numerous presentations, three publications (one in preparation), and one submitted patent.
This project produced an integrated software/hardware package that recorded data and modeled key fabrication and inspection AM tasks pertinent to manufacturing at large. This project directly addresses LLNL's Advanced Materials and Manufacturing core competency, and relies on LLNL's core competency in High-Performance Computing and Simulation and Data Science. Integrated additive/subtractive manufacturing and, more broadly, the multi-modal manufacturing capability are programmatically relevant and the needs of the Stockpile Stewardship and ICF programs require rapid build qualification for large-scale production of high-quality parts. Since our approach is agnostic to the advanced manufacturing process or the specific sensor data that is collected, this approach may be applied to traditional and advanced manufacturing technologies. Specifically, the results of the project will impact LLNL's national security mission space by accelerating the development cycle of advanced manufacturing technologies and creating adaptable frameworks to deploy our approach within other systems. Our teams hired one postdoc and an Academic Graduate Appointee who was converted to staff.
Publications, Presentations, and Patents
Lee, X. Y., et al., 2020. "Two Photon Lithography Additive Manufacturing: Video Dataset of Parameter Sweep of Light Eosages, Photo-Curable Resins, and Structures." Data in Brief 32 (2020): 106119; https://doi.org/10.1016/j.dib.2020.106119.
Ojal, N. et al., 2022. "A Universal Method to Compare Parts from STEP Files." Journal of Intelligent Manufacturing 33, 2167-2178 (2022); https://doi.org/10.1007/s10845-022-01984-3.
Giera, B. et. al. "Digital Twins for Advanced Manufacturing and Inspection." Presented at AM-Cross JOWOG 2022.
Giera, et. al. "Universal Method for Part Comparison Based on Requirements Specification." Presented at Machine Learning for Industry 2021, Virtual. August 2021.
Au, et. al. "Autonomous Multimodal Manufacturing Optimization via Digital Twins." Presented at Machine Learning for Industry 2021, Virtual. August 2021.