Degree Type

Thesis

Date of Award

2020

Degree Name

Master of Engineering

Department

Aerospace Engineering

Major

Aerospace Engineering

First Advisor

Stephen Holland

Abstract

Typically, in finite element model analysis and automated fiber placement, composite fiber orientations are computed using either simple coordinate axis rotations or a complicated full analysis model. The former performs well on simple surfaces with single axis curvature but degrades in performance when used on doubly-curved surfaces due to fiber crossing and buckling. A full analysis model provides the accuracy of a truly physically based solution, however this process can be quite extensive and does not lend itself well to industrial use.

Therefore, this work presents a new two-step process for defining optimal composite fiber paths for use in automated fiber placement machines and finite element analysis models. Building upon work done previously in the area of discrete geodesic path generation, a method for optimizing fast approximate geodesic paths on triangular meshes using a strain energy minimization technique is presented. A comparison of the effectiveness of this process with those already discussed in the literature show how this process is fast, accurate, and ready to use in commercial applications. This study also shows how this process effectively finds optimal fiber paths on complex, non-developable surfaces, which will improve finite element analysis models and provide composite manufacturers with the ability to create components with complex geometry. Improvements to this process are discussed and have been implemented such as utilizing the parallel computing capabilities of a GPU, which speeds up the process by computing geodesics in parallel for large surfaces. The algorithms presented in this study are available freely online and have been successfully integrated into an automated finite-element model-building software called De-La-Mo (https://idealab-isu.github.io/autofiber/).

DOI

https://doi.org/10.31274/etd-20200902-137

Copyright Owner

Nathan Scheirer

Language

en

File Format

application/pdf

File Size

43 pages

Share

COinS