Design of machine learning model for predicting the compressive strength of fabric fiber-reinforced Portland cement

Abstract

Mortar is a combination of cement, sand, and water used in civil engineering works; for joining building blocks, forming structure members, and plastering and masonry. Despite these vast uses, it is linked with structural failures mainly low compressive strength and brittle behavior under tensile stress. To address this, incorporating fabric fiber into the mortar mix has been explored to improve its mechanical performance. This study aims to predict mortar compressive strength a critical mechanical property of a mortar prepared with PC using three soft computing models i.e. feed-forward neural network (FFNN), support vector machine (SVM), and stepwise regression (SWR). The experimental data generated by adding fabric fiber (0%-2%) to the mortar mix and measuring compressive strength at 7, 14, and 28 days of curing age were used to train and test these models. The results revealed that SWR outperformed FFNN and SVM with a training and testing accuracy of 99.97% and 99.67% respectively. The results underscore the potential of advanced modeling techniques like SWR for enhancing mortar performance for the development of more reliable and durable construction materials for sustainable construction through the reuse of textile waste.