COMPLAS 2023

Microstructure Generation and Full-field Multi-scale Analyses for Short Fibre Reinforced Thermoplastics

  • Praud, Francis (Arts et Métiers Institute of Technology)
  • Schneider, Konrad (Hamburg University of Technology)
  • Chatzigeorgiou, George (Arts et Métiers Institute of Technology)
  • Meraghni, Fodil (Arts et Métiers Institute of Technology)

Please login to view abstract download link

In the actual ecological context, short fibre reinforced thermoplastics, due their lightweight and good mechanical properties, appear more and more attractive for structural applications in the automotive industry. However, their use remains hampered by the difficulty to predict their complex behaviour arising from complex microstructural effects combined with damage mechanisms as well as the rheology of the thermoplastic matrix. Aiming at providing a better understanding regarding their mechanical behaviour, a full-field multi-scale approach dedicated to short fibre reinforced thermoplastics is proposed. This approach is based on a mesh generation tool for matrix-inclusion Representative Volume Elements (RVE) featuring periodicity by construction. Basically, fibres are randomly positioned within the RVE while their orientations are distributed according to an Orientation Distribution Function (ODF). At the microscopic scale, while the fibres are assumed to be elastic, the behaviour of the thermoplastic matrix is described by phenomenological multi-mechanism constitutive model accounting for viscoelasticity, viscoplasticity and ductile damage. The proposed approach thus enables to visualize the local deformation and degradation mechanisms occurring at the microscopic scale while analysing their influence on the macroscopic response of the composite upon monotonic, persistent and cyclic loadings. Although such a full-field multi-scale model is highly computationally expensive and therefore poorly attractive for direct structural analyses, it gives access to valuable information towards enriching means-field multi-scale models. These latter indeed required reasonable computational resources making them more suitable in industrial applications.