A review and framework for modeling methodologies to advance automated fiber placement

Abstract

Accurate and reliable modeling techniques are required to properly understand and predict manufacturing processes and the quality of the final product. The Automated Fiber Placement (AFP) process in its entirety is commonly difficult to predict due to the complex and interconnected phases of the manufacturing lifecycle. Currently, modeling within AFP utilizes physics-based models (PBM) to understand the design of a structure and its translation to a manufacturing plan. Although this is an excellent choice for simpler problems, adding in multi-scale or transient properties can render PBM incapable of detecting minute variations and hidden patterns. Data-driven models (DDM) can be employed to understand and utilize manufacturing and inspection related data which presents a clear and successful option in complex cases that are not easily represented by PBM. This paper outlines a systemic review of the use of PBMs and DDMs in AFP along with methods for their combination into hybrid models (HM). The review concludes with identifying gaps in current modeling techniques and demonstrating the efforts being undertaken to further advance modeling efforts for AFP manufacturing.

Keywords

• Automated fiber placement
• Physics-based modeling
• Data-driven modeling