Abstract
Rising natural resource consumption leads to increased hazardous gas emissions, necessitating the concrete industry's focus on sustainable alternatives like palm oil fuel ash (POFA) to replace cement. Also, advanced machine learning (ML) techniques can uncover previously unreported insights about the effects of POFA that may be missing from the literature. Hence, this study investigates the influence of varying concentrations of POFA on fresh and mechanical characteristics with quantifying ML approaches and microstructural performance, as well as the environmental impact of structural concrete. For this, cement substitutions of 5 %, 15 %, 25 %, 35 %, and 45 % (by weight of cement) were utilized. POFA enhanced the overall concrete workability, with slump increments ranging from approximately 9 %-55 % and compacting factor increments of 4 %-12 %. Mechanical performance of POFA concrete improved up to 25 % replacement levels, with the highest enhancements observed in compressive (4.5 %), splitting tensile (36 %), and flexural (31 %) strength, for the mix containing 15 % POFA. The finer particle size of POFA improved microstructural performance by reducing porosity, aligning with the enhanced mechanical strength. The environmental impact of POFA was assessed by measuring eCO(2) emissions, revealing a potential reduction of up to 44 %. Incorporating 5 %-15 % POFA yielded ideal mechanical performance results, significantly enhancing sustainability and cost-effectiveness. Regarding the ML approach, it can be observed that a low regression coefficient (R(2)) contrasts sharply with the higher R(2) values for the random forest (RF) and the ensemble model, indicating satisfactory precision prediction with experimental results.