These slides were presented at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

This paper presents research conducted at the University of Maine (UMaine) Advanced Structures and Composites Center (ASCC) on the effects of material model temperature and time dependence on numerical model predictions for polymer material extrusion additive manufacturing processes. This information is aimed to give insight to designers, analysts, and engineers so that informed decisions can be made considering time, funding, and resource allocation. The numerical finite element analysis (FEA) models were built in Abaqus 2024 with a sequentially coupled thermo-mechanical analysis combined with an Additive Manufacturing (AM) module. Three benchmark models, namely Ring, Thermometer, and W-Shape, were chosen for this study in order to accentuate different modes of deformation. The material selected was neat polyethylene terephthalate glycol (PETG) as it is a matrix material that have been used in additive manufacturing applications and its properties could be readily obtained. The choice of a neat material was justified to simplify the analysis by neglecting direction effects on the material properties. The material model utilized in this work was obtained from experimental characterization results. The temperature-dependent material property data were fitted by bilinear curves derived from a least square function available in a Python library (pwlf). The time-dependent material property data were given in terms of master curves and shift factors. The derivation of those curves set up the parameterization of the material model, enabling the study of model sensitivity as the components of those curves could be varied. Specific heat, elastic modulus and coefficient of thermal expansion (CTE) temperature dependence and elastic modulus time dependence (viscoelasticity) were the primary parameters of interest. The remaining material properties and process parameter values, such as deposition and ambient temperature, were kept constant. Maximum deformation and maximum von Mises stress were the key performance indicators (KPI) evaluated and used to provide recommendations for modeling considerations. When applicable, SIMULIA iSight 2024 was employed in the sensitivity studies conducted on the benchmark models. Direct deflection comparisons resultant from the use of different material models were made as well as the evaluation of the impact of the material properties on the KPI values through the derivation of Pareto plots. In terms of sensitivity investigations, comparison between models using constant and temperature-dependent material models were made. Similarly, some parameters that are part of the viscoelastic model, namely relaxation terms and time constants, were varied to gauge its impact on the same KPI values. In this case, because of the constraint imposed by Abaqus on the use of viscoelasticity with AM module, a User Material Subroutine (UMAT) written in FORTRAN was developed so that the viscoelastic material property could be used with progressive element activation. Finally, a more comprehensive study was also conducted, in which temperature and time dependent combined effects were compared across the benchmark models.