Abstract
The growing volume of industrial and electronic waste has intensified the need for sustainable material management strategies. Among these waste streams, cathode-ray-tube (CRT) glass is of particular concern due to its high density and lead-bearing composition, which typically contains about 20–25 wt.% lead oxide. Using recycled CRT glass (RCRT) as a fine aggregate in cementitious mixtures offers a practical means of reducing landfill disposal while enhancing mortar performance. However, the mechanical behavior of RCRT-containing mortars has not been sufficiently modeled, thereby constraining the optimized design of such sustainable mixtures. In this study, two white-box, soft-computing techniques, the Group Method of Data Handling (GMDH) and Gene Expression Programming (GEP), were developed to predict the compressive strength of mortars incorporating RCRT. The database consisted of 139 laboratory specimens, and the machine-learning models were trained using the following input variables: water-to-binder ratio (w/b), water content, cement content (CC), fly ash, sand content, RCRT content, and curing time (CT). The GMDH model demonstrated superior predictive performance, achieving an R(2) of 0.942 with RMSE and MAE values of 2.97 and 2.59, respectively. In contrast, the GEP model produced higher error levels (RMSE = 6.94 and MAE = 5.28). These findings indicate that transparent, data-driven modeling can capture the nonlinear interactions governing strength development in RCRT-modified mortars and provides a reliable basis for designing sustainable, dense, and mechanically efficient mixtures suitable for both conventional and radiation-shielding applications.