Quantitative analysis and modeling of mechanical properties of geopolymer concrete incorporating nanomaterials

Geopolymer-based composites have attracted growing interest as sustainable alternatives to ordinary Portland cement, particularly when modified with nanomaterials to improve mechanical and microstructural performance. This study presents a structured, database-supported synthesis and empirical modeling of geopolymer composites incorporating different nanomaterials. Peer-reviewed experimental studies were selected using predefined inclusion and exclusion criteria, and mixture-level data were extracted, checked, and standardized into a unified database comprising 229 experimental records from 50 reference entries. Records containing incomplete, duplicated, inconsistent, or unsuitable numerical data were excluded from the corresponding regression analyses, and missing values were not estimated or interpolated. The database covered different precursor systems, including fly ash, ground granulated blast-furnace slag, metakaolin, volcanic tuff, red-mud-blended systems, and blended precursors, as well as nanomaterials such as nano-silica, nano-clay, nano-titania, graphene oxide, carbon nanotubes, multi-walled carbon nanotubes, waste glass nanopowder, and nano-metakaolin. The effects of nanomaterial type and dosage on compressive strength, splitting tensile strength, flexural strength, and modulus of elasticity were evaluated. The results showed that nanomaterials can enhance mechanical performance through filler action, nucleation effects, pore refinement, matrix densification, and crack-bridging mechanisms. However, the degree of improvement depends strongly on nanomaterial type and dosage, dispersion quality, precursor composition, curing conditions, and mixture characteristics, while excessive dosages may reduce performance due to particle agglomeration and reduced workability. Separate property-specific datasets containing complete paired observations were used to develop database-specific empirical relationships correlating compressive strength with splitting tensile strength, flexural strength, and modulus of elasticity. Unlike conventional OPC-based equations and relationships developed for individual geopolymer systems, the proposed models were calibrated using a unified database representing multiple nanomaterial types and precursor systems. Linear, power, and logarithmic models were examined, and their performance was evaluated using R 2 , RMSE, MAE, and leave-one-out cross-validation. No single regression form consistently provided the best performance for all investigated properties. The linear model showed competitive performance for splitting tensile strength, whereas the logarithmic model produced the lowest overall prediction errors for flexural strength and modulus of elasticity. Nevertheless, moderate-to-low fitting and cross-validation coefficients indicate that compressive strength alone cannot fully represent the complex behavior of nanomaterial-modified geopolymer composites. Therefore, the proposed equations should be considered preliminary, database-specific empirical tools applicable only within the material systems and numerical ranges represented by the property-specific datasets. The primary contribution of this study is a database-supported quantitative synthesis and benchmarking exercise demonstrating the limitations of compressive-strength-only correlations for nanomaterial-modified geopolymer composites.