Efficient parameter identification for macroscopic fiber orientation models with experimental data and a mechanistic fiber simulation

Abstract

Advanced macroscopic fiber orientation models depend on a variety of phenomenological parameters. The prediction quality is closely related to the choice of those parameters. Therefore, the aim of this research is to propose an efficient method for parameter identification. First, a macroscopic fiber orientation model for concentrated short fiber-reinforced polymers with a minimum number of parameters has to be identified. To define the macroscopic model a comparison with experimental data is used. A sliding-plate experiment with repeatable initial conditions is conducted for obtaining fiber orientation evolution under controlled shear and temperature conditions. Then the fiber orientation models are fitted to the experimental validation curve. Since the experimental curve generation for parameter fitting is time and cost consuming, a more efficient method is exploited: a mechanistic direct fiber simulation. The simulation can then be used to generate fiber orientation curves for varying physical descriptors (fiber length, fiber length distribution, volume fraction, viscosity, shear rate).