Predicting induction heating of tape-based thermoplastic composites - The role of orthotropic electrical conductivity
Yannick Buser is a PhD student in the Department of Production Technology. (Co)Promotors are prof.dr.ir. R. Akkerman and dr.ir. W.J.B. Grouve from the Faculty of Engineering Technology.
Induction welding offers efficient assembly of large and complex structures from individual thermoplastic composite (TPC) parts. Despite the commercial success for woven fabric-reinforced TPCs in aerospace industry, the adoption of induction welding for unidirectional (UD) ply-based parts lags behind in comparison. Induction heating is based on the electromagnetic induction of eddy currents in the carbon fibre network. For UD ply-based TPCs, these currents depend, in part, on coincidental fibre-fibre interactions, resulting in a high sensitivity of the process to small variations in the material and its processing history. This sensitivity does not align well with the present approach to process window development and part and tool design, which relies on empirical procedures and practical know-how. Predictive induction welding simulations are essential for enhancing process understanding and can also support the iterative adjustment of process parameters in response to changes in material systems and tooling. This thesis therefore aims to enable prediction of the induction heating in UD ply-based composites through physics-based numerical simulation.
The work first investigates the induction heating of carbon fibre-reinforced PEEK and PEKK laminates using a stationary coil, with heating patterns monitored through a thermal camera. The results show that eddy currents are primarily induced along the different fibre directions in the laminate lay-up, with heating mainly originating from the Joule effect. Additionally, the ply interfaces are found to significantly influence the distribution of current density within the plies.
Thereafter, the characterisation of orthotropic electrical conductivity of UD plies and laminates is explored, examining the applicability of the rule of mixtures to predict conductivity in the fibre direction, and investigating variations in the electrical properties driven by stochastic effects in the transverse directions of the plies. The conductivity in the fibre direction is found to exhibit predictable behaviour and demonstrates excellent conductive properties. In contrast, conductivity in the transverse directions is four to five orders of magnitude lower and shows significant variability, with fluctuations spanning (more than) an entire order of magnitude.
The material data is then used in a numerical model to predict the induction heating response of flat laminates. The simulations are validated through comparison with the results of the prior induction heating experiments. The simulations prove effective for providing an accurate estimate of the induction heating response of tape-based TPCs. However, only an average induction heating response is predicted. The transverse electrical conductivity of UD plies varies to such a degree that any simulated induction heating profile carries significant uncertainty. Therefore, predictions currently serve as initial estimates and will require further fine-tuning on the shop floor, ideally with support from in-line monitoring and control technologies.
Finally, the validated numerical modelling framework was used to conduct a sensitivity analysis on the induction heating patterns. The analysis shows that improving the through-thickness electrical conductivity of UD plies leads to a more stable heating rate during induction heating, enhancing the robustness of the welding process. Ultimately, this work thus contributes to enabling both predictions and robustness of the induction welding process for tape-based TPC components.
