Advanced nano-concrete characteristics materials prediction using Taguchi design of experiment and artificial neural networks

Abstract

The incorporation of nano-materials into concrete has emerged as a promising approach to enhance its mechanical performance, durability, and microstructural characteristics. However, predicting nano-concrete mix designs is complex due to the interaction of multiple influential factors, which has led to significant advancements in nano-concrete mixtures. This study presents a hybrid methodology that integrates the Taguchi design of experiments (DOE) with artificial neural networks (ANNs) to predict the performance characteristics of advanced nano- concrete, providing high prediction accuracy and valuable insights into the synergistic effects of nano-additives or fine particles from concrete waste. Various nano-additives such as nano-silica, and carbon nanotubes (CNTs) were used in different proportions, along with key mix parameters, including water, concrete waste, nanoparticles or superplasticizer. A Taguchi L8 orthogonal array was employed to identify the most significant factors affecting split tensile strength. Experimental results from the optimized mix combinations were then used to train an artificial neural network (ANN) model, enabling the accurate prediction of concrete performance under varied mix conditions. The ANN demonstrated high accuracy, with R2 values exceeding 0.95 and low error metrics across all predicted properties. The combined use of Taguchi DOE and ANN not only minimized the number of required experiments or evaluating interaction effects among factors but also provided a robust predictive tool for nano-concrete design and improving quality is through reducing variability. This hybrid approach provides a powerful framework for developing high-performance cementitious composites in the era of smart and sustainable construction materials, thereby accelerating the development of next-generation construction materials.