Fibre-reinforced composite manufacturing is highly sensitive to input variability (e.g., fluctuations in areal weight, tow misalignment, or inconsistent binder distribution) [1] which can lead to defects during preforming and subsequent steps. This sensitivity has long hindered the industrial deployment of process models, due to concerns over their predictive accuracy. However, recent work at BCI addresses this challenge by focusing on preforming process modelling, specifically how variations in the architecture and properties of dry textile reinforcements affect fabric deformation during forming. By embedding stochastic material descriptions into finite element forming models, we have demonstrated that it is possible to design robust forming processes that consistently deliver high-quality outcomes even when upstream variability is present [2]. This enables the definition of forming windows that are insensitive to material noise, thereby reducing defect rates and increasing confidence in preform quality. Crucially, one key conclusion from this work is that process models do not need to be perfectly accurate to support optimisation and that capturing the right trends is often enough.
This modelling philosophy becomes even more important when considering sustainable composite systems. In a recently completed PhD project focused on the processing of environmentally friendly materials (i.e., recycled carbon fibres and low-impact resins [3]) we demonstrated that upfront digital design accounting for manufacturing constraints, can significantly accelerate the development of viable processing conditions for new materials (see Figure 1). The same study also confirmed that recycled feedstocks tend to exhibit inherently higher variability. While this challenge remains unresolved for now, it is clear that the modelling strategies developed in the aforementioned study will be valuable here too.
The concept of using physics-based modelling not only to predict outcomes but to enable variability-aware process design will be advanced further in one of the workstreams of a recently announced EPSRC Prosperity Partnership. We strongly believe that wider industrial adoption of new composite systems (which is critical to deliver the net-zero agenda) will not be possible without a much greater reliance on process simulation. This, however, requires a paradigm shift whereby we stop chasing perfect models and start embrace “good enough” models [4].
References:
[1] Chen S., Talokder T., Mahadik Y., Thompson A. J., Hallett S. R. and Belnoue J. P.-H. (2025). Preform variability propagation in non-crimp fabric (NCF) forming, Composites Part B: Engineering, 299:112418, https://doi.org/10.1016/j.compositesb.2025.112418.
[2] Chen S., Thompson A. J., Dodwell T. J., Hallett S. R. and Belnoue J. P.-H. (2025). A comparison between robust design and digital twin approaches for Non-Crimp fabric (NCF) forming, Composites Part A: Applied Science and Manufacturing, 193:108864, https://doi.org/10.1016/j.compositesa.2025.108864.
[3] Yavuz B.O., Hamerton I., Longana M.L. and Belnoue J. P.-H. (2025). Modelling the tensile behaviour of aligned discontinuous carbon fibre thermoplastic matrix composites under processing conditions, Composites Science and Technology, 269:111252, https://doi.org/10.1016/j.compscitech.2025.111252.
[4] Belnoue J. P.-H. and Hallett S. R. (2024). Process models: A cornerstone to composites 4.0, Composites Part B: Engineering, 283:111621, https://doi.org/10.1016/j.compositesb.2024.111621.
