Analysis of pore structure, fiber distribution and strength characteristics of ultra-high-performance wet-sprayed concrete (UHPWSC) using X-CT

Ultra-high-performance wet-sprayed concrete (UHPWSC) combines the high-performance advantages of UHPC with the construction efficiency of wet-spraying. However, how the pore structure and steel-fiber architecture evolve during wet spraying, and how these features are associated with mechanical response, remain unclear. In this study, UHPWSC mixtures with steel fiber contents of 0%, 0.5%, 1.0%, and 1.5% were prepared. Rheological tests, mechanical tests, and X-ray computed tomography (X-CT) were used to investigate the rheological behavior of the fresh mixture, mechanical properties, pore structure characteristics, and fiber orientation distribution. Grey relational analysis (GRA) and a GM(1,3) model were further employed to explore the relationship between CT-derived structural parameters and strength. The results showed that, with increasing steel fiber content, the slump of UHPWSC decreased, whereas the yield stress and plastic viscosity increased. Compared with the fiber-free specimen, the compressive strength and splitting tensile strength of the specimen with 1.5% steel fibers increased by 77.0% and 63.8%, respectively. X-CT results indicated that the porosity, pore number, and pore size of UHPWSC all decreased with increasing steel fiber content, and the pore size distribution could be described by a Gamma function. Wet spraying induced steel fibers to preferentially align parallel to the sprayed deposition plane, with more than 70% of the fiber orientation angles concentrated within the range of 60°–90°. This preferential orientation became more pronounced as the steel fiber content increased. GRA showed that, within the present dataset, the proportion of pores with a sphericity of approximately 0.6 and the proportion of CT-detectable pores smaller than 100 μm exhibited relatively high correlations with strength variation. The GM(1,3) model established using these two CT-derived pore descriptors showed good in-sample fitting performance for the present limited dataset, with average relative errors for compressive strength and splitting tensile strength both lower than 5%.