Journal Article
Integrated experimental, statistical, and ANN validation study on concrete deterioration under acid rain, carbonation, and chloride environments
Y. H. Sudeep; V. Vijaykumar; M. S. Ujawl; Christo George; S. Karthik; Rakesh Kumar; Gopal Bharamappa Bekkeri; Kiran K. Shetty
Innovative Infrastructure Solutions · Vol. 11, Issue 7 · 2026
Abstract
Concrete structures in coastal–industrial regions are simultaneously exposed to acid rain, carbonation, and chloride ingress; however, most durability studies investigate these mechanisms independently, limiting realistic service-life prediction. This study presents an integrated experimental, statistical, and artificial neural network (ANN)-based evaluation of M40-grade concrete subjected to synthetic acid rain, accelerated carbonation, and combined carbonation–chloride environments. Laboratory simulations were conducted for exposure durations of 20, 40, 60, and 80 h to replicate aggressive atmospheric conditions. Mechanical degradation was assessed through compressive and splitting tensile strength tests, while chemical deterioration was quantified using depth-wise pH profiling and carbonation depth measurements. Results revealed that compressive strength decreased by up to ~ 25% under acid rain exposure and showed greater deterioration under combined carbonation–chloride conditions. Splitting tensile strength exhibited higher sensitivity to degradation, with reductions exceeding 20% at prolonged exposure durations. Surface pH declined from 12.6 to approximately 8.8, confirming rapid alkalinity loss in the outer 1 cm zone. Carbonation depth increased progressively, reaching ~ 11.5 mm in the combined exposure environment. Regression analysis demonstrated strong inverse correlations (R 2 > 0.90) between carbonation depth and mechanical properties, while one-way ANOVA confirmed statistically significant differences across exposure durations ( p < 0.05). To enhance predictive capability, an ANN model was developed using experimental parameters as inputs, achieving high prediction accuracy (R 2 = 0.98 for training and R 2 = 0.96 for testing). The findings highlight synergistic deterioration effects of multi-environmental exposure and emphasize durability-based design strategies for reinforced concrete structures in aggressive environments.