CT Data Evaluation of Fibre Reinforced Polymers to Determine Fibre Length Distribution


D. Salaberger, K. Arunachalam Kannappan, J. Kastner, J. Reussner, T. Auinger - CT Data Evaluation of Fibre Reinforced Polymers to Determine Fibre Length Distribution - PPS26-2010, Banff, Canada, Canada, 2010


For short fibre reinforced polymers the mechanical properties are mainly determined by fibre content, fibre orientation and fibre length distribution. Many destructive 2D methods are available to determine these characteristics. Statistical methods are needed to transfer results of 2D analysis to a 3D volume. X-ray Computed Tomography (X-CT) is a suitable method to overcome the statistical problem since it is a non- destructive method that provides three- dimensional information. For that reason it is used to characterise fibre reinforced polymers more and more frequently. Limitations that have to be overcome with X-CT analysis arise from the dependency of resolution on sample size, limited data quality and complexity of 3D algorithms. We present a method to characterise glass fibres in injection moulded short fibre reinforced polypropylene in a way of determination of start and end coordinate for every fibre. Therefore it is possible to calculate fibre characteristics like length and orientation in a very exact way. The analysed data was generated with a sub-µm-CT device at resolutions of 3 µm, which corresponds to the quarter of the fibre diameter. Specimens with different fibre content were analysed since data quality decreases and complexity of analysis increases with increasing content. The highest fibre content we investigated was 30 % by weight. The software pipeline that was developed calculates the skeleton of each fibre. For regions, where the separation of fibres is not possible via gray value only, algorithms were developed that take morphological information into account. An estimation of error was performed in two ways: results of established methods like pyrolysis with following microscopic analysis was compared to CT results. Secondly fibre characterisation was performed manually within a small volume of the CT data using standard 3D analysis software to be able to calculate the error in terms of number of fibres that are represented correctly.