COMPREHENSIVE PROPERTY DETERMINATION FOR FIBER-REINFORCED POLYMER COMPOSITES IN EXTRUSION DEPOSITION ADDITIVE MANUFACTURING—BAYESIAN VS DETERMINISTIC This work introduces both deterministic and Bayesian methodologies to simultaneously determine the elastic constants of the constituent polymer and the fiber orientation state in a short fiber-reinforced polymer (SFRP) composite based on a small number of experimental measurements of the composite properties. The ability of the

Abstract

This work introduces both deterministic and Bayesian methodologies to simultaneously determine the elastic constants of the constituent polymer and the fiber orientation state in a short fiber-reinforced polymer (SFRP) composite based on a small number of experimental measurements of the composite properties. The ability of the Bayesian approach to calibrate uncertainties makes it a promising tool for enabling a probabilistic framework for composites manufacturing digital twins. The two methods that enable the reverse engineering of the orientation of the fibers and the in-situ polymer properties are compared. For the extrusion deposition additive manufacturing (EDAM) process and other SFRP composites processes (e.g. injection molding), extensive characterization efforts are currently required to develop composites manufacturing digital twins. To circumvent the extensive characterization required, Digimat© provides a suite of tools to reverse engineer material properties of SFRPs. However, Digimat© lacks a methodology to inversely determine the fiber orientation state and the constituent polymer properties simultaneously. To that end, this work presents both a deterministic and hierarchical Bayesian approaches to determine the polymer properties and the fiber orientation state simultaneously. The results indicate that both approaches provide a reliable framework for the reverse engineering process. The deterministic approach provides a more rapid, point estimate methodology, whereas the Bayesian approach provides a more comprehensive methodology that includes uncertainties in the reverse engineering process.