A step towards the production of eco-friendly non-proprietary ultra-high performance fiber reinforced concrete: Experimental assessment and machine learning quantification

This study investigates the mechanical performance of eco-friendly non-proprietary ultra-high-performance fiberreinforced concrete (UHPC) using locally available cementitious materials and evaluates the effect of ambient and thermal curing regimes over short- and long-term durations. Key mechanical properties, including compressive strength, first-peak flexural strength, and modulus of elasticity (MOE), were experimentally assessed. Results demonstrated that thermal curing significantly accelerated early-age strength development, with short-term compressive and flexural strength enhancements ranging from 77% to 90%, and long-term gains of 10% to 30% compared to ambient curing. MOE exhibited consistent growth across both curing methods, achieving values up to 52 GPa. Predictive equations for MOE and first-peak flexural strength were established to aid in design applications. Furthermore, two machine learning models&#8212;Random Forest (RF) and k-Nearest Neighbors (KNN)&#8212;were employed to predict mechanical performance. The RF model outperformed KNN across all metrics, achieving correlation coefficients (R²) between 0.93 and 0.99 and minimal error values (RMSE < 1.21 for compressive strength). These findings validate the potential of non-proprietary UHPC as a sustainable alternative to commercial mixes and provide a predictive framework for its behavior under varying curing conditions, advancing its practical implementation in structural applications.