Warqaa Muhammad Bahaaddin 1
Serwan kamal Jalal 2
Naema Sadeeq Saleh 3
Lawend Kamal Askar 3
Sevar Dilkhaz Salahaddin 2
Ahmed Mohammed Ahmed 2
Salim T.Yousif 4

https://aeis.bilijipub.com/article_243517.html

This study evaluates how coarse aggregate size and combined admixture regimes influence compressive strength in high-performance concrete. A total of 120 cube-level observations were assembled from four mixes, five aggregate sizes of 38, 25, 19, 12.5, and 9.5 mm, and two curing ages, 7 and 28 days, with three specimens per design cell. The mixes followed a 1:1.16:2.02 proportion, cement: sand: aggregate and a nominal water-to-cement ratio of 0.32, while Mix2-Mix4 incorporated 5/0.5, 7/0.7, and 10/0.9 silica-fume/superplasticizer regimes, respectively. Classical and HC3-robust ANOVA confirmed statistically significant effects of mix regime, curing age, and aggregate size on compressive strength, while residual diagnostics indicated acceptable normality but some heteroscedasticity, motivating the robust inference check. To address reproducibility and AI scope, the revised study also includes a case-level supplementary dataset and an AI/ML prediction module. Under GroupKFold validation by design cell, Gradient Boosting achieved the strongest predictive performance, MAE 3.46 MPa, RMSE 4.92 MPa, R2 0.885. Variable Importance Permutation indicated the important parameters as silica fume ratio, curing period, superplasticizer ratio, and aggregate size. The highest measured strength was 83.48 MPa at 28 days for the 12.5 mm aggregate with the 10% silica fume and 0.9% superplasticizer regime. Because silica fume and superplasticizer were co-varied in the experimental design, Mix2-Mix4 are interpreted as combined admixture regimes rather than isolated single-factor effects.