The increasing demand for concrete due to factors such as population growth, urbanization, and infrastructure development has led to a scarcity of natural aggregates required for concrete production. This scarcity is exacerbated by the rapid industrialization, which generates various waste by-products posing environmental challenges for disposal. To address these issues, researchers are exploring the utilization of waste by-products in construction, offering an alternative to traditional aggregates.
Additionally, the lack of proper disposal system for residential and industrial wastes has caused serious threat to the ecosystem [1, 2]. Not only that, wastes such as fly ash, blast furnace and steel slag, rice husk ash, etc. pose a serious issue to the health of inhabiting living being [3]. One such waste material is Linz-Donawitz slag (LD Slag), a by-product of steel production, which can be repurposed as both coarse and fine aggregate in concrete by altering its physical form. The scarcity of natural aggregates, along with concerns about environmental impact, has prompted investigations into the feasibility of using LD slag as a sustainable alternative. LD slag is produced during the conversion of raw iron ore into steel products, and it is typically discarded as waste. However, researchers have found that LD slag possesses favorable properties for concrete production, showing potential as a substitute for natural aggregates [4]. Previous studies have also highlighted its other positive attributes, such as stable mechanical and physical properties, as well as improved concrete strength. Prasad [5] and Qasrawi [6] conducted compressive strength test on the concrete made with steel slag aggregate sand concluded that these aggregates can be used as a partial replacement of natural aggregates in concrete. Wang
[7] and Maslehuddin [8] have studied the behavior of concretes containing slag under various severe condition sand have concluded that the mechanical and durability aspects of these concretes are acceptable. The utilization of steel slag as an aggregate not only addresses the scarcity of natural resources but also contributes to waste management by repurposing industrial by-products.
Moreover, with the construction sector being a significant consumer of materials, especially in developing countries like India, the need for sustainable alternatives and waste management solutions becomes even more critical. The studies suggest that LD slag has the potential to serve as a reliable substitute for natural aggregates in concrete, enhancing the concrete's performance while reducing environmental impact. However, while research has shown promising results, further exploration and validation are needed to ensure the widespread implementation of LD slag in construction practices.
The world of machine learning (ML) tasks that involve finding relationships has changed a lot because of new and advanced techniques. Some important methods are gaussian progress regression (GPR), support vector machine (SVM), artificial neural networks (ANN), and random forest (RF). GEP helps create computer programs that can predict things better. SVMs are good at figuring out patterns in data, both for making predictions and sorting things into categories. ANN works like the human brain, using lots of data to learn and make predictions. RF has the advantage of bootstrapping in multiple regression trees. These methods have made a big difference in estimating values, saving resources and money. In recent studies, these techniques have been used to predict things like how strong the concrete is going to be. El-Mir [9] and Kumar et. al. [10] used SVM and GPR for predicting concrete properties and found that both models perform well for prediction. Abuodeh [11] and Naderpour [12] used ANN for predicting the compressive strength of concrete. Mangalathu [13] used machine learning models for identification of mode of failures of reinforced concrete shear walls. Rajkarunakaran [14] and Shaqadan [15] used random forest for predicting concrete strength. Rajkarunakaran [14] and Mangalathu [13] compared RF with other machine learning models and concluded that RF performs better than other ML models. This shows how they can help make better predictions for different things in a sustainable way.
The present study has been done to analyze the changes in physical and mechanical parameters of conventional concrete due to the replacement of natural aggregates with LD slag aggregates. Both the aggregates namely, natural fine aggregate (NFA) and natural coarse aggregates (NCA) were partially replaced with LD fine aggregates (LDFA) and LD coarse aggregates (LDCA) at an incremental increase of 20% respectively. The concrete mixes prepared were cast, cured and tested in order to get their compressive, split tensile and flexural strengths. These properties of LD slag aggregate concrete (LDSAC) were then compared with the properties of conventional concrete. Subsequently, RF was used to predict the compressive, split tensile and flexural strengths of the concrete based on the input of physical and mechanical parameters values of concrete.