CT scan analysis for the characterization of fiber orientation in long discontinuous fiber composite materials

Benjamin R Denos, Purdue University


Fiber-based composites have material properties which are highly dependent on the fiber orientation and distribution of their constituent fiber and matrix. While traditional continuous fiber laminates contain highly aligned and predictable fiber orientation, discontinuous fiber composites have a microstructure that can vary locally. This is especially the case for long discontinuous, or "chopped", fiber composites that strongly maintain their initial material form as "platelets" throughout most processes. Micrographs seen in this study prove the overall homogeneity of this material, but also show a clear pseudo-laminate structure locally where collimated fiber groupings are preserved. In order to determine local material properties for long discontinuous fiber composite (LDFC) materials, local fiber orientations and other microstructural characteristics need to be ascertained. The characterization method employed in this study is nondestructive computed tomography (CT) scanning and image analysis to ascertain and quantify fiber orientation in LDFCs. The current microstructure characterization method of destructive microscopy is used for validation in this study, but is limited to small regions of interest. In contrast, CT scans on entire part volumes are used, but limit scan resolution such that fiber groupings can be seen, but individual fibers may not be discernable. This method has been applied specifically to high fiber volume fraction (~60%), large fiber aspect ratio (1800:1), transfer molded composite parts. However, this approach should apply to most fiber-based composite materials where fiber and matrix density contrasts are detectable. CT scans were taken with a North Star Imaging X-50 system with the parameters specified in Section 3.3. A script was developed in Matlab to determine fiber grouping orientation from density gradients in CT scan volumes. A unique orientation tensor is defined at every voxel of the CT scan volume. Orientation measurement verification of a part geometry with known microstructure met expectations. Next, a CT scanned part was sectioned and measured via the fiber ellipse method to ascertain actual fiber orientation for seven regions of interest. CT-measured orientations were compared with micrograph-measured values at those locations. The magnitudes and general trends of each orientation term correlated sufficiently to suggest the CT analysis method as useful microstructure mapping tool with a few limitations. Determination of proper image filter (kernel) sizes requires some preliminary knowledge and engineering judgment to capture orientation trends without loss of local microstructural details. The density gradient measurement method results in boundary effects that can affect orientation measurements at image boundaries and part surface boundaries. These boundary effects extend as far inward from the surface as the dimension of the largest filtering kernel used. Resolution of CT-measured orientations for this study are considerably less than those of micrograph measurements, but CT-measured orientations can be determined over entire part volumes. Ultimately, the new method outlined is an effective step toward non-destructive, whole-part fiber orientation and microstructure measurement that requires further improvement and validation. The orientation information from the current method, while not yet fully refined, may be used as a significant improvement in microstructural model fidelity in a number of design, simulation, and analysis applications.




Pipes, Purdue University.

Subject Area

Aerospace engineering|Materials science

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server