The segregation tendency of coarse aggregates during transportation, placement, and setting remains the least understood property of self-consolidating concrete (SCC) mixtures. Previous research provided recommendations to either reduce segregation in SCC by reducing the total volume of coarse aggregates (increasing the volume of mortar) and/or by adding viscosity-modifying admixtures. Both approaches usually lead to increases in production cost, therefore hindering a wider implementation of SCC. This paper presents an extensive experimental program to evaluate the segregation potential of a wide range of SCC mixtures. A total of 123 SCC mixtures coveting a large scope of mixture designs were made. The properties of fresh SCC mixtures such as slump flow, clump 50, V-funnel flow, and L-Box flow in addition to compressive strength at various ages were evaluated. A simple yet accurate test method for quantitatively. evaluating the segregation of SCC was proposed and its results were compared to that of other test methods in the literature. The large experimental database thus created was also used to train an artificial neural network (ANI) that was able to accurately predict the segregation resistance of new SCC mixtures unfamiliar to the model. Finally, a sensitivity analysis using ANN predictions was performed. ANN responses in simulating the effect of basic mixture ingredients of SCC on its resistance to segregation were compared to those obtained using actual experimental results, and a critical analysis of the effect of mixture design parameters on segregation is provided. The study contributes to an improved understanding of segregation of SCC and should assist in producing SCC mixtures with high resistance to segregation.
