Mechanical properties prediction of lightweight coal gangue shotcrete

Abstract

Coal gangue is one of the most common types of solid waste worldwide, and its storage not only consumes land resources but also pollutes the water, air, and soil leading to resource waste. Therefore, to promote the utilization of coal gangue in shotcrete production and enhance its reusability, this study replaced coarse aggregates with coal gangue. The effect of the coal gangue aggregates-to-sand ratio (G/S), PVA fibre %wt to cement ratio (F/C), and the coal gangue aggregates (CGA) particle size on the working and mechanical properties, such as compressive strength (CS), splitting strength (SS), and density of lightweight coal gangue shotcrete (LCGS) were studied. A total of 504 specimens were manufactured and tested. The test outcomes were provided as testing and training datasets for six machine learning (ML) models in order to evaluate the prediction ability of each model. Additionally, the Partial Dependence Plot (PDP) analysis was used to visualize the results of the ML models and investigate the relationship between the variables and the outputs. In the modelling, using the Back Propagation Neural Network (BPNN), the high correlation coefficients for CS and density were obtained, while SS was determined using the Support Vector Machine (SVM), in which both models presented accurate and reliable predictions with the application of particle swarm optimization (PSO). Finally, a sensitivity analysis was carried out to assess the significance of the ranking input factor.