A comprehensive modelling framework for defect prediction in automated fibre placement of composites

Automated fibre placement (AFP) has become one of the mainstream techniques in advanced composite manufacturing, bringing much higher efficiency and reducing waste compared to hand layup. However, one critical issue in the AFP technique is manufacture-induced defects, such as wrinkles, when the materials are deposited along a curved path or on a doubly-curved surface. These defects can lead to the degradation of the finished part’s mechanical performance by up to 25%. Currently, industry largely depends on costly physical trials to tackle the issues. Process simulation of the manufacturing provides a feasible alternative way by conducting the trials in virtual space. This is, however, challenging due to the complexity of the AFP deposition process, i.e. large numbers of machine processing parameters (such as deposition path, processing temperature, layup speed, pressure, etc.) coupled with the nonlinear behaviour of the prepreg tape material. Current academic research is mostly limited to qualitative prediction under a single condition, which is quite far from real-world industrial applications.
In this work, recent advancements achieved in process simulation of AFP, are presented. The work starts with the characterisation of material under key mechanisms that are known to affect defect formation. Constitutive models for each mechanism are then derived and implemented in the form of material subroutines for commercial finite-element (FE) packages, which are further integrated into a full FE simulation platform. The new framework was validated against real-world data and good agreement was observed. An investigation of the processing parameters on deposition quality demonstrated the capability of the platform to work in varied processing conditions rather than single isolated ones. The work provided insights into how manufacturing parameters of AFP affect the part quality and pave the way to replace physical trials with virtual tests, thus it can effectively reduce cost and increase production rate.