Anisotropic Flow and Fiber Orientation Analysis of Preimpregnated Platelet Molding Compounds
Process modeling of prepreg platelet molded composites (PPMC) is of interest for enabling optimized design of PPMC structures. The platelet mesoscale and the part scale are on the same order, which introduces several modeling challenges including the treatment of variability. The fiber length and concentration results in anisotropic rheological behavior during processing. Current commercial flow simulation tools do not include anisotropic constitutive models and utilize second order orientation tensor methods for the fiber orientation which requires significant interpretation to inform a structural model. A flow and fiber orientation model to predict the final orientation state in molded parts should treat the anisotropic behavior of the system and output adequate mesostructural details with which a structural analysis can be developed. Such a model is formulated and presented here with experimental validation. The platelets are modeled as a homogenized, incompressible, transversely isotropic viscous medium whose orientation is defined by three mutually orthogonal unit vectors corresponding to the fiber, transverse, and platelet normal directions. Individual orientation vectors are the state variables of the flow simulation to avoid the ambiguity associated with the second order orientation tensor. The constitutive model was implemented in a commercially available finite element code. The numerical implementation was rigorously verified by applying canonical deformations to a single finite element. A Lagrangian discretization scheme is the preferred modeling framework for the most natural treatment of the orientation state. The model was validated for a series of flow conditions and part geometries in coordination with experimental data. CT based orientation measurements were used to validate the orientation state predictions while flow front predictions were confirmed by short shot experiments. Finally, a methodology for processing and performance analysis was presented in this work. Further enhancements may be readily developed within the presented modeling framework.
Pipes, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our