A study on methods to improve X-ray CT data quality of composites

  • Christina Fraij

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)

Abstract

Better characterisation of the meso-structure of carbon-fibre reinforced polymer composites (CFRPs), including features such as ply-drops, tape gaps and overlaps, in-plane fibre orientation and out-of-plane wrinkling can only be achieved with high-quality non-destructive testing (NDT) data. X-ray micro-computed tomography (CT) is a popular NDT technique used for characterising small coupon-sized CFRPs but optimisation for high-fidelity characterisation of larger components is challenging. High contrast-to-noise ratio (CNR) is required between fibre and resin whilst voxel sizes need to be at least 20 microns (for scanning larger components). Due to the similarity in their chemical composition, achieving high contrast levels between the fibre and resin is difficult and requires careful optimisation of the acquisition system. Novel X-ray CT data-quality metrics were created which are relevant to the requirements of inversion methods that can analyse the CT data and map the important features. These optimisation metrics, which vary with material properties and CT-acquisition parameters are: effective Fibre-Volume Fraction (FVF) Contrast Sensitivity (CS) and the effective FVF CNR. Their significant dependence on the incident photon-energy distribution was investigated theoretically and experiments were conducted with a range of source voltages, target materials and filters to understand this influence. Total exposure was also studied to understand its effect on total noise, which is a combination of both temporal noise due to scattering and spatial noise due to FVF variation contribution. The optimisation metrics were then measured from experiment and compared with theory and modelling work to understand the trend in the variation of contrast and noise with varying parameters. Finally, a more rigorous framework was established for measuring the optimisation metrics required for determining X-ray CT data quality including the decomposition of the CT-scan histogram into its constituent material distributions. In conclusion, a procedure has been developed, backed by theory, modelling and experiment, which allows optimisation of CT data-quality, resulting in a significant improvement in the mapping of the meso-structure of CFRPs.
Date of Award25 Sept 2018
Original languageEnglish
Awarding Institution
  • University of Bristol
SupervisorRobert A Smith (Supervisor)

Cite this

'