Meta-modelling for fast prediction of the effective elastic properties of cement mortar doped with CNT

Abstract

This article presents an innovative metamodeling approach for predicting the effective elastic properties of cement grouts containing carbon nanotubes (CNTs), for the purpose of structural rehabilitation. The study addresses the homogenization of nano- and micro- scales through a Representative Volume Element (RVE). The numerical model consists of a cement matrix doped with randomly oriented CNTs, yet the computational load associated with this model poses a significant processing limitation. To overcome this challenge, a metamodeling approach based on a Kriging surrogate model is proposed. This approach provides a fast and accurate alternative to bypass the timeconsuming numerical homogenization process. The surrogate model establishes a connection between the microstructural characteristics of CNTs, cement, and the effective elastic properties of the composite grouts. The accuracy and efficiency of the proposed metamodel are demonstrated through detailed analyses.