Gradient-based fibre detection method on 3D micro-CT tomographic image for defining fibre orientation bias in ultra-high-performance concrete

摘要

Ultra-high-performance fibre reinforced concrete (UHPC) is a class of advanced cementitious composites characterized by its high compressive and flexural strengths, toughness and enhanced durability. The mechanical properties of the UHPC are to a great extent dependent on fibre volume fraction, orientation and distribution within the cementitious matrix. However, determination of the true three-dimensional fibre orientation and distribution is challenging. In this paper micro-computed tomography (micro-CT) is used to determine these parameters. Cylindrical samples of UHPC were extracted from a dogbone tension test specimen and from a pretensioned bridge girder, and high-resolution micro-CT images were then acquired. Using the scanned data, and following three-dimensional image reconstruction and image processing, quantitative fibre information was obtained via a novel image post-processing technique based on local-intensity gradient significantly improving cross-fibre detection compared to existing techniques. The estimated fibre volume fraction is close to design and experimentally measured values. Straight and hooked end fibres in UHPC sample were successfully identified and segmented using this novel technique. The test results show the fibre arrangement to be highly anisotropic with fibres aligned predominantly in one direction, which is attributed to the casting processes and flow.