Analysis of Experimental Study and Machine Learning Methods for Strength Prediction in Cenospheres Concrete

This research investigates the impact of replacing fine aggregates with cenospheres (0-60%) on concrete properties, integrating experimental methods and machine learning techniques. Cenospheres, a by-product of coal combustion, are used as lightweight fillers to enhance sustainability in construction materials. The experimental phase includes tests on compressive, tensile, and flexural strength to evaluate mechanical performance. Microstructural analysis, including Scanning Electron Microscopy (SEM) and X-ray Diffraction (XRD), is conducted to study morphological changes within the concrete matrix. Additionally, machine learning models, particularly Random Forest, are employed to predict concrete strength based on experimental data, offering a more accurate and automated approach to evaluating material performance. The experimental results reveal a gradual decline in mechanical strength with increasing cenosphere content due to reduced density, though this is compensated by lighter weight and improved workability. The SEM analysis reveals significant changes in the cementitious matrix with higher cenosphere content, notably in calcium silicate hydrate (C-S-H) formation and porosity. XRD confirms the presence of pozzolanic activity with unreacted silica phases. The study’s machine learning component proves effective in predicting strength outcomes, validating its potential in material design. This work demonstrates the potential for cenosphere use in concrete as a sustainable alternative while highlighting the importance of balancing mechanical performance with environmental benefits.