Document Type : Original Article
Authors
1
M.Sc. Student, Civil Engineering Department, Faculty of Engineering, University of Sistan and Baluchestan, Zahedan, Iran
2
Associate Professor, Civil Engineering Department, Faculty of Engineering, University of Sistan and Baluchestan, Zahedan, Iran
3
Professor, Civil Engineering Department, Faculty of Engineering, University of Sistan and Baluchestan, Zahedan, Iran
4
Assistant Professor, Civil Engineering Department, Faculty of Engineering, University of Sistan and Baluchestan, Zahedan, Iran
10.22065/jsce.2026.554755.3849
Abstract
Geopolymer mortars have attracted considerable attention as sustainable and environmentally friendly alternatives to ordinary Portland cement due to their lower greenhouse gas emissions and the use of natural aluminosilicate resources. However, developing simple, reliable and practical models for predicting their compressive strength remains an important challenge in materials engineering. In this study, a combined approach based on the Taguchi design of experiments and Gene Expression Programming was used to develop a predictive relationship for the compressive strength of geopolymer mortar containing Taftan pozzolan. The Taguchi method was applied to optimize the experimental design and reduce the number of required laboratory tests, while Gene Expression Programming was used to establish a nonlinear relationship between the main parameters and compressive strength. Three key variables including the water-to-solid ratio, silica-to-sodium ratio and cement replacement percentage were considered at four different levels. Based on the Taguchi orthogonal array, sixteen mix designs were prepared and tested for compressive strength at 28 days. The results indicated that the water-to-solid ratio had the greatest influence on compressive strength. The developed model showed good predictive capability, with correlation coefficient 0.97, Nash–Sutcliffe efficiency 0.93, mean absolute percentage error 11.48 percent and root mean square error 3.90, demonstrating reliable prediction accuracy.
Keywords
Subjects
