4/28/2017 Modeling Carbon Fiber Composite Microstructures Using Optical Microscopy David Leonhardt

Acknowledgements Dr. Ray Fertig III – Faculty Advisor Eric Jenson – Former Graduate Student Wyoming NASA Space Grant – Funding NASA Grant #NNX10AO95H 2

Presentation Overview Background Project Overview MATLAB Analysis Abaqus Analysis Conclusions Future Work 3

Background - Motivation 4/28/2017 Background - Motivation Composites desirable for use in aerospace, wind energy, and sporting goods Clear understanding of material properties critical to safety Ability to model 3D microstructures for comparison with actual broken specimens is desirable Need ability to model 3D to compare with actual failure testing. Hamid image Steve Georgiadis , Andrew J. Gunnion , Rodney S. Thomson , Bruce K. Cartwright Bird-strike simulation for certification of the Boeing 787 composite moveable trailing edge Composite Structures Volume 86, Issues 1–3 2008 258 - 268 http://dx.doi.org/10.1016/j.compstruct.2008.03.025 4

Background - Composites 4/28/2017 Background - Composites Usually consist of high stiffness fibers in low stiffness matrix Primary focus on carbon fibers in epoxy matrix Fiber-volume fraction common characterization Add FVF description 5

Background – Ideal vs Actual Ideal Microstructure Hexagonal packing Uniform FVF Actual Microstructure Semi-random packing Varied FVF Defects Voids Alignment Fibers Ideal microstructure Actual microstructure 6

Background – Finite Element Analysis (FEA) Numerical solution method for stresses, strains, and displacements Highly accurate Breaks geometry into small ‘finite elements’ Transforms system of differential equations to system of matrix equations for solution FEA analysis of airfoil N.p., n.d. Web. 6 Aug. 2013. <http://www.cbifans.com/images/research/researchfem1L.jpg>. 7

Project Overview Toolbox construction MATLAB image analysis Fiber volume fraction Abaqus™ input files Abaqus™ modeling 3D model from 2D images Actual microstructures 8 http://commons.wikimedia.org/wiki/File%3AMatlab_Logo.png http://commons.wikimedia.org/wiki/File%3ALogo_Dassault_Systemes.jpg

MATLAB – Circle Detection Pre-processing: Greyscale Intensity adjustment Processing: Circular Hough Transform Cleanup algorithms Manual adjustment 9

MATLAB – Hough Transform Raw imfindcircles results 10

MATLAB – Algorithm 1 Remove concentric circles 11

MATLAB – Algorithm 2 Remove matrix circles 12

MATLAB – Algorithm 3 Shrink touching circles 13

MATLAB – Fiber Volume Fraction 30% 40% Areal Method Actual circle diameters Requires high resolution Fiber Count Based on nominal fiber diameter More accurate Works with lower magnification 60% 80% 14

MATLAB – Abaqus Input Data 4/28/2017 MATLAB – Abaqus Input Data Append to text file Center location Radii Multiple stacks written to same file Ordered by nearest neighbor comparison Stack 2 Stack 1 15

Abaqus – Model Generation 4/28/2017 Abaqus – Model Generation Read text file Create wireframe Loft sections Generate mesh Create constraints Turn face on. 17

4/28/2017 Abaqus – Results Turn face on. 18

Conclusions Toolbox creation successful Interesting observations Working framework for future expansion Accurate measurement of FVF achieved Interesting observations Non-uniform stresses Stress concentrations due to fiber arrangement 19

Future Work Code Lab Work Change shrinking algorithm Change criterion Add shifting Improve edge handling More robust circle ordering mechanism Efficiency improvements Implement interface Lab Work Actual Specimen 20

Questions? 21