Temperature, diffusion, and stress modeling in filament extrusion additive manufacturing of polyetherimide: An examination of the influence of processing parameters and importance of modeling assumptions

Fused filament fabrication (FFF), a form of filament-based material extrusion additive manufacturing (AM), is an extremely useful technique for the rapid production of highly customized products; however, empirical evidence is heavily relied upon for understanding of the process. Initial modeling attempts have traditionally focused on predicting heat transfer and either interlayer diffusion and adhesion or stress development but have not taken a combined approach to analyze all three components simultaneously in a multiphysics model. In this study, we implement finite difference models to examine the combined heat transfer, polymer diffusion represented as degree of healing (Dh), and residual stress development in FFF of poly(ether imide) (PEI). Printing with PEI is of great interest because of its desirable mechanical properties and high use temperatures, but it also creates a more challenging modeling problem with higher thermal gradients and greater potential thermal processing window compared to traditionally modeled AM materials, such as acrylonitrile-butadiene-styrene (ABS) and polylactic acid (PLA). The larger processing window can potentially allow for more processing options but can also significantly complicate the optimization process. In this study, experimental analyses including trouser tear tests and part warpage measurements provide correlation to predicted Dh and stress levels. The models suggest that the temperature of a layer is influenced by the subsequent printing of up to at least three layers in the geometry studied. The results of this study further demonstrate the sensitivity of the molecular mobility and degree of healing to the reptation time (τrep), such that a small change in the τrep on the order of 2–3x can result in an order of magnitude difference in the time before interfacial healing can begin, culminating in significantly less healing occurring. Furthermore, the reptation time and subsequent healing predictions are highly reliant on the extrapolation method used to extend the reptation time to temperatures below those at which it was measured resulting in significantly different predictive results, even if the same experimental data is used.