An analysis of 3D fiber orientation distribution using a clustering method in image processing

摘要

The fiber orientation distribution in fiber reinforced thermoplastics (FRP) affects the strength and final shape of products. Therefore, for effective product design, it is necessary not only to estimate the fiber Orientation distribution by using computer simulation, but also to measure it. The purpose of this research is to measure the 3D fiber orientation in characteristic sections of molded parts such as the skin or core layer and the weld or gate line. We applied a clustering method used in image processing to determine the fiber orientation distribution from the measured 3D parameters and visualized the state of this fiber orientation distribution. In this experiment, we used energy absorbers molded with a liquid crystal polymer. First, we cut simples from the weld and gate sections and observed them through a scanning acoustic microscope (SAM). Next, we calculated the parameters of the 3D of the 3D fiber orientation from the SAM images, and made an image of the fiber orientation distribution visualized from the calculated parameters by the K-means method, one of the clustering methods. We confirmed that the fiber on orientation distribution in the outside and the fiber orientation in the inside were reversed, which corresponded with the expectations in product design. We could visualize the fiber orientation distribution and explain the numerical values of the fiber orientation distribution by the clustering method, and could confirm that this method was effective in the of the fiber orientation distribution.