Effects of Detectable Defects (EDD) – Influence of production related defects in automated fiber placement processes in thin walled carbon fiber structures

点击翻译按钮获取中文摘要

Automated Fiber Placement (AFP) is a well-established manufacturing process for the fabrication of high quality composite structures using pre-impregnated carbon fibers. The main research hypothesis proposes the possibility of performing a qualified structure-mechanical evaluation of the cured structure based on production defects in the uncured component. The necessary real time information to detect and classify production defects is provided by a thermographic process monitoring. This newly acquired knowledge describing the mechanical impact of production defects such as gap, overlap, fuzzball or twisted tow provides an unprecedented basis of decision-making on the type and necessity of corrective measures. Compared to conventionally methods of non-destructive failure analysis (e.g. ultrasonic testing, active infrared thermography, eddy current measurement, X-ray or computed tomography), it is possible to prove the defect during the production process. Therefore, the defect correction takes place much earlier and is less expen-sive. Furthermore, the current 100% ultrasonic testing rate after of the cured structures can be significantly reduced or replaced by a specific testing at critical points. This improvement in productivity and process reliability of the AFP technology leads to a significant increase in efficien-cy along the entire AFP process chain.The developed models to describe defects of a cured structure and the resulting mechanical properties based on data available during the production process exceed significantly the currently available process knowledge. The parametric modeling of different production-related defects and the associated experimental investigation of failure mechanisms generate knowledge about the influence of the curing process on production defects, while the thermal modeling of the AFP process as well as the development of machine learning algorithms characterizing and quantifying the production defects will generate further process knowledge.