Contour Fitting of Fused Filaments Cross-Section Images by Lemniscates of Booth: Application to Viscous Sintering Kinetics Modeling

A method for image analysis was implemented to determine the edge pixels of two biopolymer-based thermoplastic filaments during their hot melt isothermal sintering at 120 °C. Successive inverted ellipses are adjusted to the contour of the sintered filaments and lead to the identification of the parameters of the corresponding lemniscates of Booth. The different steps of the morphological image analysis are detailed, from 8-bit coded acquired images (1 frame/s), to the final fitting of the optimized mathematical functions describing the evolution of the filaments envelope. The complete sequence is composed of an initial pure viscous sintering step during the first minute, followed by viscoelastic swelling combined with melt spreading for a longer time, and then the stabilization of the sintered filaments shape for over 2 min at high temperatures. Using a master curve obtained from Hopper’s abacus, the characteristic viscous sintering time is assessed at tvs = 78 s, confirming the one previously found based on the measurement of the bonding neck length alone. Then, the full description of the evolution of the thermoplastic filaments envelope is assessable by image analysis during sintering trials as a result of its digital modeling as successive lemniscates of Booth, reflecting geometry changes in the molten state.

Advanced image analysis of thermoplastic filaments’ viscous sintering kinetics: Contour fitting with a Lemniscate of Booth

Viscous sintering kinetics of thermoplastic polymers has been studied for powders using models refining the Frenkel-Eshelby approach. It is usually based on the measurement of the bonding neck between two molten particles submitted to thermo-microscopy trials. Recently, specific experimental setups have been described for studying the viscous sintering of filaments used in additive manufacturing by FDM. The description of their coalescence by models developed for particles is a rough approximation. However, the evolution of the shape of their section can be modelled by lemniscate curves. In the present work, we present an advanced image analysis approach allowing the fitting of the contour of the filaments by a Lemniscate of Booth. It is based on the automatic assessment of the coordinates of their edge pixels and the adjustment of lemniscates to match their evolving shape as a succession of inverse ellipses. We apply this procedure to a model-biopolymer recently shown as 3D-printable, the plasticized zein, a corn protein extruded as cylindrical filaments. Their sintering is recorded at 120°C as 8-bits coded raw images. After segmentation, a numerical mask is applied to follow the filaments outline. Using Matlab® as computer algebra system, the adjustment and the identification of lemniscates parameters leads to determine the viscous sintering characteristic time, similar to those of standard polymers. Then, the full monitoring of sintering kinetics is achievable and makes possible a better modelling of such experimental trials and their application to enhance the control of the welding between layers in additive manufacturing.

Skorohod-Olevsky viscous sintering model sensitivity to temperature distribution during the sintering process

This paper investigates the influence of temperature field non-uniformity on sintering simulation results using the Skorohod-Olevsky viscous sintering model. As a difference to previous studies, here a thermal transient analysis is performed to provide a detailed temperature field over the component within sintering time. Results obtained using uniform temperature distribution are compared to those obtained using a nonuniform distribution derived from a transient thermal analysis. Results are compared for different geometry sizes, that lead to different temperature non-uniformity levels. The study has shown that the temperature nonuniformity cannot always be neglected and should be considered as a possible source of modeling error.