Image-based methods for the identification of in-plane composite moduli at high strain rates

Abstract

This thesis describes the development of new image-based methods to characterise in plane moduli for off-axis composites at high strain rates. It is difficult to accurately identify these properties with current experimental methods such as the split-Hopkinson bar, because the assumptions that the technique is based on can be violated at strain rates above a few hundred s−1. With this new approach, transverse and shear moduli of Carbon Fibre Reinforced Polymer (CFRP) composite samples have been identified at strain rates in the 500 to 2000 s−1 domain. The off-axis composite modulus identification methods were developed within the Image-Based Inertial Impact (IBII) test technique. Using the Virtual Fields Method (VFM), full-field measurements and rigid body virtual fields were included in the principle of virtual work to derive stress averages on a test sample. Transverse and shear moduli were then identified from linear fits to the averagestress-strain response. This thesis details the numerical implementation of the VFM theory used to derive these moduli for unidirectional (UD) and multi-directional (MD) composite samples with configurations of UD90◦, UD45◦ and MD45◦.
An image deformation study was undertaken to verify the identification methods and to assess smoothing parameters for processing experimental images. Two full-field measurement techniques for calculating the displacement fields were evaluated in the image deformation simulations: the Grid Method and Digital Image Correlation (DIC). The first major goal of the study was to verify that DIC displacements could be used within the IBII test methodology. This was achieved by comparing the moduli identified from unsmoothed strain and acceleration fields with no simulated camera noise overlayed on the images. Here, the moduli identified with DIC were similar the Grid Method values. The second goal of the study was to obtain optimised smoothing parameters that gave the best trade-off between systematic and random errors on the identified moduli. This analysis showed that different optimised smoothing parameters were obtained for the UD45◦ samples, because of their more-complex kinematic fields compared to the UD90◦ samples. Optimised smoothing parameters that resulted in systematic and random errors of less than 1% were obtained for both grid and DIC images. IBII tests were performed at the University of Southampton’s high speed impact laboratory, where a 50 mm bore gas gun was used to launch aluminium projectiles at composite samples with impact speeds around 40 m.s−1. A Shimadzu HPV-X ultra-high speed video camera operating at 2 MHz recorded images of the impacted samples and moduli were obtained from the deformed image sets. Validation of the modulus identification methods was achieved by comparing the moduli from different sample configurations. The transverse modulus E22 obtained from the UD45◦ samples was within one standard deviation (SD) of the UD90◦ sample result. Shear strain waves were detected in the UD90◦sample strain maps, which were generated by a slight pitch-angle misalignment between the impacting components in the test. As a result, shear moduli were also obtained from the UD90◦ samples, with G12 values within on SD of the UD45◦results. Shear moduli obtained from MD45◦samples were around 6 % lower than the UD values, which was consistent with the reduced laminate density of the MD samples. Comparison of the high strain rate moduli obtained in this study suggests that values reported in literature could be over-estimated, including both a material and a testing system response. Comparison of the identified moduli from the two full-field measurement techniques showed that theE22 and G12 values were lower for DIC to the Grid Method. Voids were detected in thecomposite plates that the UD DIC samples were cut fro m, which provided an explanation for the lower DIC values. Shear moduli identified from MD45◦ samples using DIC and the Grid Method were within 1%, which was expected given that both samples were cut from the same plate. Results from this study suggest that both the Grid Method and DIC are suitable for high strain rate moduli identification within the IBII test methodology.

More information

Identifiers

ORCID for Samuel Parry: orcid.org/0000-0002-0215-7365

ORCID for Fabrice Pierron: orcid.org/0000-0003-2813-4994

Catalogue record

Export record