A Hybrid Digital Twin Framework for Modeling Machine Process Parameters for Automated Fiber Placement-based Manufacturing of Composite Structures

The Navy seeks a physics-based, artificial intelligence (AI) driven digital twin (DT) framework to tailor the automated fiber placement (AFP) process for manufacturing complex composite structures, aiming to improve production rates and part quality. While AFP is a widely used technology for fabricating large-scale composite components, its benefits are often constrained by challenges such as insufficient process planning, reliance on manual inspections, and the intricate interaction of physical phenomena inherent to the process. These limitations often lead to inconsistencies in part quality and diminished manufacturing efficiency. To address these challenges, Global Engineering and Materials, Inc. (GEM) and its team member propose a physics-informed, AI-driven DT solution that intelligently tailors AFP process parameters in real time. This framework integrates three key modeling components: high-fidelity physics-based simulations to represent the thermomechanical and material behavior during fiber placement; data-driven models trained on empirical and synthetic datasets; and machine learning-based reduced-order models that enable rapid predictions with minimal computational overhead. The DT will function as a real-time decision-support and control tool by ingesting data from in-situ monitoring systems such as temperature sensors, compaction force measurements, and imaging systems. This observational data will be assimilated on-the-fly, allowing the DT to dynamically update its internal models, detect early signs of process anomalies or defect formation, and autonomously recommend or apply corrective actions. The proposed effort also includes the development and execution of virtual experiments to validate the DT's core capabilities. These simulations will be used to explore optimal processing conditions, evaluate sensitivity to disturbances, and assess the DTís ability to achieve optimal layup characteristics and minimize defects, demonstrating its potential for enabling intelligent, self-adaptive AFP manufacturing.