PSI - Issue 70

Ch. Vihas et al. / Procedia Structural Integrity 70 (2025) 461–468

462

1. Introduction Concrete is one of the widely used construction materials globally. However, the production involves the extensive use of cement, which contributes significantly to the release of greenhouse gases. To reduce the environmental impact, one can use mineral admixtures, such as rice husk ash (RHA), natural pozzolans, Ground Granulated Blast Furnace Slag (GGBS), fly ash (FA), silica fume (SF), and metakaolin (Gogineni, A et al (2024), Alyami, M et al. (2024)). It not only helps from an environmental perspective by reducing the carbon footprint in concrete production but also improves the material's overall performance and durability. The addition of mineral admixtures into concrete is also a cost-effective option in construction projects, as they are residues released by industries (Sharma et al (2023)). In this study, GGBS is used as a replacement for cement partially to find the benefits of GGBS in concrete production. GGBS originates from the iron and steel industry, and its use in concrete has shown many benefits. It is observed that concrete incorporating GGBS typically shows greater strength, workability, and durability (Baral et al., (2024), Tran, N. T et al. (2024)). Additionally, GGBS also helps to reduce the risk of corrosion in steel reinforcement and reduces the occurrence of thermal cracking. It also reduces the alkali-silica reaction (ASR), which can cause concrete cracking. Machine learning models are used to identify relationships and patterns in the data (Zhang, Z, (2023), Rathakrishnan, R., et al. (2024)). Machine learning algorithms like support vector regression, artificial neural networks, and decision trees are used to analyze this data to uncover hidden connections between the materials and the concrete strength (Phoeuk et al., 2023)). Using the ML algorithms, one can predict the strength of concrete mixes that include mineral admixtures as a partial replacement for cement. By doing so, these models can optimize mix designs to achieve the strength, reduce material waste, and improve the sustainability of concrete. Ultimately, this method helps ensure more reliable and cost-effective concrete with fewer errors in predicting its performance, adding to safer and more efficient construction practices. The main objective of this study is to predict the compressive strength of concrete with GGBS as a replacement for cement. The aim is to determine whether incorporating GGBS into concrete will enhance the overall performance, making it more sustainable and durable than traditional concrete. Machine learning-based prediction of concrete strength incorporating GGBS, which represents modern solutions in the industry, aiming to enhance the efficiency and sustainability of mix design. The process begins by collecting data on various features such as the quantities of cement and GGBS, water-cement ratio, curing conditions, and the age of the concrete. For the current study, four ML models (Random Forest, Artificial Neural Networks, Support Vector Machine Learning and XG Boosting) are applied. These machine-learning models enable the prediction of the strength of concrete replaced with GGBS. In order to improve the performance of model hyperparameter tunig was applied. The hyperparameters are the most sensitive parameters which can control the behavior of the training process and influence how well the model fits the data. These parameters are varied from model to model. Hyperparameter tuning is the process of searching for the best set of hyperparameters to improve the model’s performance . The performance and models are compared using accuracy measures. 2. Methodology

Fig. 1 . Typical methodology of Machine learning models