Machine learning study of the mechanical properties of concretes containing waste foundry sand

Concrete is the most commonly used man-made material in buildings, pavements, and dams. The production of concrete requires large quantities of fine and coarse aggregates. To preserve the natural resources, the use of waste and by-product materials in concrete mixtures has been given lots of attentions. Fresh and hardened properties of the concrete mixtures containing waste foundry sand (WFS) as a partial or full replacement for fine aggregate have been the focus of several recent studies. The use of predictive models for the properties of the concretes can save in time and energy and provide information on scheduling the activities such as framework removal. In this study, M5P algorithm was used to model the compressive strength, modulus of elasticity, flexural strength, and splitting tensile strength of these concretes. For this purpose, a comprehensive dataset containing information on the mixture proportions and the values of the mechanical properties at different ages was collected from internationally published documents. The dataset was divided into two subsets of training data and testing data. Several performance metrics were used to evaluate the performance of the developed models. The results indicated that the proposed models can provide reliable predictions of the target mechanical properties. A more comprehensive dataset can further improve the performance of the models. (C) 2020 Elsevier Ltd. All rights reserved.