Development of an artificial neural network model to predict waste marble powder demand in eco-efficient self-compacting concrete

Abstract

The marble industry produces large amount of waste at almost every stage of marble processing. This waste is always uncontrollably discharged into open areas. Therefore, the consumption of Waste Marble Powder (WMP) in concrete is very important and will provide both an economic gain for concrete industries and an opportunity to achieve eco-efficient concrete production. Self-Compacting Concrete (SCC) is mostly preferred concrete type which uses the WMP as powdered material. Thus, to increase WMP usage in the production of eco-efficient SCC, it is aimed in this study to develop a model that can predict WMP demand. To develop this model, Artificial Neural Network (ANN), an artificial intelligence method, was preferred. An ANN model was developed using a comprehensive dataset that included eco-efficient SCC mixture compositions, workability measurements, and compressive strengths. The ANN model with seven inputs and one output as WMP was successfully trained and managed to produce the correct outputs to both validation and test datasets. Mix design of SCC with aimed properties can be a long process compared to conventional concrete. The proposed ANN model could reduce time loss in the production process. ANN's success is expected to facilitate the production of eco-efficient SCC with WMP.