Mathematical model for prediction of compressive strength of ternary blended cement concrete utilizing gene expression programming

Abstract

In order to encourage the utilization of rice husk ash (RHA), ceramic waste powder (CWP) and glass waste powder (GWP) in their ternary combinations in the production of concrete as well as determine the variability in the resulting compressive strength (CS), the application of soft-computing algorithms is of great interest. Therefore, this study developed a predictive model using the Gene Expression Programming (GEP) technique. The predictive model was trained and developed using extensive and trustworthy data of the compressive strength of ternary blended cement concrete comprising four (4) input parameters, including the RHA, CWP, GWP, and curing day (CD). The model was developed based on 70 % training datasets, and the model's accuracy was checked using 15 % testing datasets; the model's outcomes were validated using 15 % experimental datasets. The performance of the developed model was further assessed by applying statistical checks, comparing regression models and sensitivity analysis. The R-values in the training, testing and validation phases of GEP models are 0.95, 0.93 and 0.89 respectively with an objective function (OF) of 0.04. Based on the optimum GEP model, a closed-form mathematical equation is presented, proving to have excellent adaptability, predictive ability and capability of accurately estimating the compressive strength of concrete made from ternary blended cement mixes. Therefore, the outcomes of this study can help the construction sector predict the properties of pozzolanic concrete and manage scarce resources.

Keywords

• Pozzolanic concrete
• Ternary binders
• Machine learning
• Waste management
• Prediction
• Gene expression programming