ML Based Design and Optimisation of High Strength Concrete with GGBS and Silica Fume Using Aggregates Characteristics

Abstract

The high strength and sustainable construction of high strength concrete with low environmental impact have been the focal point of current research due to the high carbon footprint of ordinary Portland cement. The use of supplementary cementitious materials, such as ground granulated blast furnace slag (GGBS) and silica fume, in the production of high strength concrete is a common approach that increases the mechanical properties of the concrete while enhancing its sustainability. However, the interaction between the binder, the aggregate, and the properties of the concrete can be complex, and therefore the conventional trial-and-error approach for the design of concrete mixtures can be time-consuming and not effective. In this study, a machine learning framework was developed for the prediction and design of high strength concrete that contains GGBS and silica fume, and takes into account the detailed characteristics of the aggregate. A comprehensive database that contains information on the mixture proportions, the curing conditions, the properties of the binder, and the characteristics of the aggregate, including the grading curve, the specific gravity, the water absorption, the crushing strength, the impact strength, and the shape indices of the aggregate, was created and used for the training and the evaluation of the models. Four machine learning models, namely the artificial neural network model, the support vector machine model, the random forest model, and the gradient boosting model, were used to predict the performance indicators of the concrete. The performance indicators of the models were evaluated by using the coefficient of determination (R2), the mean absolute error (MAE), and the root mean square error (RMSE). In addition, an explainable artificial intelligence approach was used to investigate the effects of the input variables on the predicted strength of the concrete, and to identify the interactions between the variables. The results of the models were also used for a multi-objective design approach in order to find the optimum mixture composition that satisfies the strength, workability, cost, and environmental impact of the concrete. The proposed data-driven approach provides an efficient and effective design methodology for sustainable high strength concrete, and highlights the significance of the characteristics of the aggregate on the properties of GGBS–silica fume based concrete systems.