Roudy Wehbe

Date of Award

Spring 2020

Document Type

Open Access Dissertation


Mechanical Engineering

First Advisor

Zafer Gürdal

Second Advisor

Ramy Harik


Automated Fiber Placement (AFP) is a manufacturing process used to fabricate large composite structures for aerospace applications. During the process, the machine head deposits multiple bands of composite material named tows over a prescribed path. Temperature, speed, and compaction pressure can be varied to obtain a good layup quality. For conventional laminated plate structures manufactured using the AFP process, fibers are laid at constant angles (0°, 90°, ±45°) in straight paths. However, to manufacture complex shell structures or variable stiffness plates, curved paths are necessary in the design leading to a length mismatch between the parallel edges of the towpath. Since finite width fiber tows are originally straight, a form of tow deformation is necessary to absorb the difference in length thus ensuring good adherence to the surface substrate. In this work, several possible tow deformation mechanisms are proposed and classified as follows: (1) elastic strain deformations (tensile, compressive, shear), (2) large in-plane deformations (fiber waviness and bunching), and (3) large out-of-plane deformations (tow wrinkling and folding). Usually, large tow deformations are unfavorable and considered as defects, thus process interruption may be necessary to perform manual repairs. The aim of this proposal is to develop models able to capture tow deformations when placed on a curved path and to validate their occurrence during the AFP process. The proposed work is tackled from three different perspectives: (1) geometrical modeling, (2) physics-based modeling, and (3) experimental investigations.

Understanding the geometry of a given layup is necessary to determine critical locations for defect occurrence. A worst-case scenario is considered where all other deformation mechanisms are suppressed in favor of out-of-plane wrinkling. Governing equations for a tow placed on a general surface are derived, and a simple deformation function is applied to the shorter edge of tow showing different wrinkle patterns along the length. A simplified form of the governing equations is provided for the special case of tows steered on a flat surface. Finally, examples are presented visualizing the wrinkles patterns and showing critical locations of wrinkling for a given layup for flat and general surfaces. Relationships between the wrinkles wavelength and the steering radius can be obtained from existing mechanics models and experimental data. This information is used to improve the geometrical model, thus obtaining more accurate results for wrinkles modeling.

In the physics-based model, the influence of the material properties and tow geometry on the deformations is studied. The tow is modeled as multiple fiber bundles laying on a stiff foundation. In a first step, only in-plane deformations are allowed, thus capturing the elastic deformation mode (tensile/compressive strains) and the large in-plane deformations (fiber waviness and bunching). In a second step, out-of-plane deformations are allowed, enabling the modeling of additional tow deformations mentioned earlier such as tow wrinkling. In a last step, the interaction between the neighboring fiber bundles is investigated by considering the transverse and shear stiffness properties of the uncured tow.

Finally, experimental investigations measuring tow deformations over steered paths are carried using the Digital Image Correlation (DIC) technique. Thermoset pre-impregnated carbon fiber tows are speckled first, then placed using an AFP machine over vi multiple paths with different radii of curvature. Shape and strain measurements of the deformed tows are acquired using a Stereo DIC setup. Quantified measurements of wavelength, width, and amplitude for tow wrinkles are obtained as a function of the steering radius. The effect of the substrate, time, and temperature on the formation of wrinkles is also studied. Other experiments using DIC are carried out to determine the deformation of neighboring tows within a course when steered along a constant curvature path. To understand the effect of AFP process parameters on tow deformations, a benchmark reference path on a flat surface with a linear increase in the curvature is proposed. Based on the location of the first visible defect along the length of the path, a critical steering radius is determined for each set of process parameters used. Finally, steering experiments are performed on a cylindrical tool with varying process parameters. The quality of the manufactured steered paths is assessed through image processing of acquired profilometry scans during manufacturing. Recommendations regarding optimal set of process parameters for future manufacturing activities are provided based on the measured defects along the path.