Effect of mineral admixtures on the properties of steel fibre reinforced SCC proportioned using plastic viscosity and development of regression & ANN model

Abstract

In the present study, fresh and mechanical properties of self-compacting concrete (SCC) mixes made up of quartz sand, micro silica, Ground Granulated Blast-furnace Slag (GGBS) and fibers have been evaluated for three grades of concrete (M40, M50 and M60). Plastic viscosity based mix proportioning is adopted in the present study. Further, steel fibers have been added to the respective mixes in various proportions (0.5%, 1.0% and 1.5% by volume of concrete) to examine the effect of fibers on fresh and mechanical properties. The fresh properties include slump flow, V-funnel, T50 cm slump, and L-box. It is found that the fresh properties for all the mixes are within the limits mentioned by EFNARC. Compressive strength, split tensile strength and flexural strength were evaluated for 3, 7, 28, 56, 90 and 180 days for M40, M50 and M60 grades with and without fibers. It is observed that the workability of all the mixes corresponding to M40, M50 and M60 is decreased with the increase of steel fibre volume fraction. The mechanical properties, split tensile strength, flexural strength increased with the increase of percentage of fibers. The compressive strength is increased up to 1.0% fibre volume and marginally decreases for 1.5% fibre volume and however, it is higher than the mix with 0.5% fiber volume. The increase in mechanical properties may be due to additional formation of CSH. Multiple linear regression analysis has been performed by considering about 75% of the mixed experimental mechanical data and three equations are proposed and validated with the remaining dataset. It is found that the predicted mechanical properties are closely matching with the related experimental observations. Further, Artificial neural network based model has been developed to predict the compressive strength of various SCC mixes. The back propagation training technique and Levenberg-Marquardt algorithm was employed to develop ANN model. It was found that the model could predict the compressive strength of various SCC mixes +-12% compared to experimental observations.