Optimization of reinforced cellular lightweight concrete beams under Cyclic loading: integrating experimental analysis and numerical simulations with regression modelling

Abstract

This study explores the optimization of reinforced cellular lightweight concrete (RCLC) beams under cyclic loading by integrating sustainable materials and advanced modelling techniques. Cement was partially replaced with limestone powder, and natural fine aggregates with recycled construction and demolition waste (CDW), to generate six concrete mixes. Mechanical behaviour was assessed using non-destructive tests (Ultrasonic Pulse Velocity and Rebound Hammer), along with flexural strength evaluation over 28 days. Results showed that moderate replacement levels, particularly in Mix N4, delivered optimal mechanical performance and internal uniformity. Furthermore, an Artificial Neural Network (ANN) model was developed using MATLAB to predict mechanical properties based on mix parameters. The model demonstrated strong generalization ability with a low mean squared error, proving its reliability for performance forecasting. This research supports sustainable construction by promoting waste reuse, minimizing carbon emissions, and validating machine learning techniques for material optimization.