FORECASTING MECHANICAL PROPERTIES OF CONCRETE CONTAINING SECONDARY RAW MATERIALS

Abstract

Climate change due to the significant release of CO2 from cement industries has become a critical issue worldwide. However, secondary cementitious materials (SCMs), can be used as eco-friendly cement alternatives. Most of the previously conducted studies primarily rely on either experimental investigations or simple regression models to find out the optimal mixture design of concrete made with SCMs. However, in the experimental approach, few tests could be performed for optimization due to time limitations and the availability of resources. On the other hand, simple machine learning (ML) models can’t be relied on without extensive validation. Therefore, to overcome these limitations this study aims to use three ML techniques, such as artificial neural network (ANN), adaptive neuro-fuzzy interface system (ANFIS), and gene expression programming (GEP) method with experimental validation for forecasting the compressive strength (CS) and tensile strength (TS) of concrete incorporated with ground granulated blast furnace slag (GGBS) and silica fume (SF) as SCMs. A comprehensive dataset containing the eight most influential inputs of concrete with CS and TS as outputs was collected from the literature and used for model development. The efficiency of the developed model was evaluated using statistical measures and experimental validation. Additionally, sensitivity and parametric analysis were carried out to identify the coherence of developed models with experimental studies. Comparative analysis showed that ANFIS models surpassed other models with higher R2 and lower errors. Conversely, GEP demonstrated enhanced performance compared to ANFIS and ANN concerning the nearness of statistical measures among the training, validation, and testing sets. Further, GEP also gives predictive formulas that can be utilized for the pre-design of concrete mixtures made with SF and GGBS. Sensitivity and parametric analysis showed strong relevance with experimental studies validating the model's performance. Thus, the recommended models are reliable and can be used to promote the sustainable use of industrial wastes (SF and GGBS) in concrete.