Modelling porosity in composite liquid infusion processes

联合收割机结合增强纤维和液体树脂的复合材料制造可以以不同的方式进行。最流行的方法涉及在压力下注入树脂，使得其被迫渗透通过保持在封闭工具（树脂传递模塑- RTM）中或在真空膜（真空注入- VI）下的干燥织物。随后的固化在零件被移除进行最终加工之前进行。不管灌注方法如何，基本要求是完全浸渍复合材料预制件而没有缺陷。在实践中，这是难以实现的，并且主要问题是空隙和孔隙的产生。一定的孔隙率总是存在的;然而，它应该被限制在1-3%以下，因为即使是这个量也会使复合材料的强度性能降低高达20%，并且对长期疲劳性能特别有害。灌注过程中的孔隙率几乎完全是由于树脂流过纤维增强材料时形成的气泡造成的。在流动前沿发生双相流动，涉及树脂通过开放织物结构中的间隙的组合快速流动，与单个紧密纱线的延迟渗透相结合，导致空气截留和气泡的产生。空气可能被永久地捕获，或者可能从纱线中排出并通过表面张力粘附在纱线上。在高压力梯度的情况下，气泡可以与树脂一起流过织物结构中的通道，或者在某个点处被捕获，或者在出口处被排空。此外，气泡可能聚结或可能塌陷（扩散）。最后，它们的尺寸将根据它们在形成时的内部压力和在树脂固化期间的最终施加压力而变化。这些气泡的最终尺寸和分布决定了孔隙率的分布。本项目将通过实验研究产生气泡的加工条件、气泡迁移的机制以及决定固化复合材料中最终空隙尺寸和孔隙率的加工条件。这项工作将得到数值方法的支持，使用有限元（FE）和解析解进行预测，以获得孔隙度生成的标准。最后一项任务是利用人工智能（人工神经网络- ANN）解决方案开发FE和替代模型，以预测两个演示器部件的孔隙率。人工神经网络的方法是特别感兴趣的，因为这样的技术是计算速度快，可以想象，在“真实的时间”，以监测传感器在输液制造的一部分，并控制流速在最小的空隙含量。

Composites manufacturing to combine reinforcing fibres and liquid resin can be undertaken in different ways. The most popular method involves infusing the resin under pressure so that it is forced to permeate though a dry fabric that is held in closed tooling (Resin Transfer Moulding – RTM) or under vacuum membranes (Vacuum Infusion - VI). Subsequent curing then takes place before the part is removed for final machining. Regardless of the infusion method an essential requirement is full impregnation of the composite preform without defects. In practice this is difficult to realise, and a major problem is the generation of voids and porosity. Some porosity is always present; however, it should be limited to under 1-3% since even this amount can reduce composite strength properties by up to 20% and is especially detrimental to long term fatigue properties. Porosity in infusion processes is almost entirely due to the formation of air bubbles as the resin flows through the fibre reinforcement. At the flow front dual phase flow occurs involving combined fast flow of resin through gaps in the open fabric architecture, combined with delayed infiltration of the individual compact yarns, causing air entrapment and the creation of bubbles. Air may be permanently trapped or may evacuate the yarns and cling to the yarn by surface tension forces. With high pressure gradients bubbles may flow with the resin through channels in the fabric architecture, either to be trapped at some point, or to be evacuated at an outlet vent. Furthermore, bubbles may coalesce, or possibly collapse (diffusion). Finally, their size will change depending on their internal pressure at formation and final applied pressure during resin cure. The final size and distribution of these bubbles determines the porosity distribution.This project will investigate experimentally the processing conditions that create bubbles, the mechanisms of bubble migration and the processing conditions that determine final void sizes and porosity in cured composites. The work will be supported by numerical methods for prediction using finite element (FE) and analytical solutions to obtain criteria for porosity generation. A final task will develop FE and surrogate models using artificial intelligence (artificial neural network - ANN) solutions to predict porosity of two demonstrator parts. The ANN approach is of special interest, since such a technique is computationally fast and could, conceivably, be used in ‘real time’ to monitor sensors in infusion manufacture of a part and control flow rates in for minimum void content.