Automated generation of carbon nanotube morphology in cement composite via data-driven approaches

Abstract

Electrified cement composite has attracted considerable attention in major scientific and engineering fields due to its excellent functional characteristics. With increasing interest in this functional material, the need for an advanced theoretical approach has also increased significantly. In the present study, a data-driven model based on hierarchical micromechanics and particle swarm optimization is proposed to estimate the morphological characteristic of conductive nanofiller of cement composites. Experimental data needed for the simulation are acquired by fabricating cement specimens with various contents of multi-walled carbon nanotube (MWCNT), carbon fiber, and water-to-cement ratios, and measuring their electrical resistivity, porosity, and aspect ratio by relevant experimental and computational techniques. Based on the proposed framework, a series of numerical simulations including the experimental comparisons of the electrified cement composite are carried out to clarify the potential of the present model. The number of model parameters is reduced to the curviness of MWCNT, which is the most influential model parameter, and the process of collecting and simplifying the pattern is included.