Studying the geometrical models of tubular braided composite using micro computed tomography and particle swarm optimization

This paper compares the geometrical model of a 2DTBC with its actual model generated by the Micro-Computed Tomography (μCT) method. First, the geometrical models' equations are edited to simulate the manufacturing process more accurately, and the equations are incorporated into the TBC-Gen program. Then, a μ CT scan is done on a cured carbon fiber TBC, and 2D and 3D models are created. Next, two yarns of the scanned models are extracted and modelled. The segmented yarns are analyzed, and their geometrical data, including cross-section area, major and minor yarn diameters, orientation, mandrel diameter, portion angle and centre points, are extracted. Next, a yarn path simulating the scanned yarn is generated using the TBC-Gen and the extracted parameters. Then, the generated geometrical model is compared with the μCT model. To do that, a parameter named portion angle is introduced to help the geometrical model better fit the μCT model. Finally, Particle Swarm Optimization (PSO) is used to optimize the portion angle. The result of the fitting algorithm shows the accuracy of the geometrical model (less than 1% error) to simulate actual TBC. Understanding the amount of error between a geometrical model and the actual model will help to evaluate the application of geometrical models more thoroughly. Also, the more accurate geometrical model will contribute less error to the FEM simulation. The quantitative comparison between these two models can give a clear understanding of the amount of error existing in the geometrical model compared to an accurate model generated by μCT.