Open Collections

Nazemi, Amir

University of British Columbia

The superior strength-to-weight ratio of woven fabric composites, along with their high formability, is one of the primary design parameters defining their increased applications in modern manufacturing processes, including aerospace and automotive. However, complex geometries of finished components continue to bring several challenges to designers as they need to cope with manufacturing defects with a minimal overhead cost. Wrinkling is a common defect in forming and handling semi-finished textile composites. This defect is due to weak bending stiffness of yarns in fabric preforms, causing an excessive relative motion between fibers during out-of-plane deformation. This challenge can be further exacerbated when using specialized blank holders in forming set-ups. For forming simulations of fabric composites, Finite Element (FE) method has been a longstanding means to predict and mitigate manufacturing defects. Such simulations are predominately intended to not only predict the onset, growth, and shape of wrinkles, but also determine the best processing conditions that can yield an optimized positioning of fibers upon forming. However, small-time step requirements in explicit FE codes, numerical instabilities, and large computational time are among some drawbacks of the current FE models in composites forming research, hindering their extensive use as fast and efficient digital twins in pertinent industries. This thesis presents a novel woven fabric simulation technique through a so-called material point method (MPM), which enables the use of much larger time steps along with fewer numerical instabilities, hence the ability to run significantly faster and yet efficient forming simulations. As a preliminary case study, both standard 2D deformation modes and 3D hemispherical forming setups are employed using a plain weave test material. The MPM results are compared to the conventional FE simulations, as well as to the physical experiments. Overall, the MPM method showed a runtime around 20 times faster than its FE counterpart in the present study, yet with comparable reliability in forming parameters predictions as verified by different metrics during experiments. Such a fast simulation tool is believed to have the potential to sizably enhance the development of emerging automated fiber handling and preform processes under the emerging AI-based smart manufacturing paradigm.

Text

2023-07-31

Attribution-NonCommercial-NoDerivatives 4.0 International

http://hdl.handle.net/2429/83439

Mechanical Engineering

University of British Columbia

UBCO

http://creativecommons.org/licenses/by-nc-nd/4.0/