Experimental and GEP-based sensitivity analysis of water and chloride ingress in concrete with varying mix designs and curing conditions

Abstract

This study presents an experimental and gene expression programming (GEP)-based investigation into concrete resistance to water and chloride ingress under diverse mix designs and curing conditions. A large-scale experimental campaign comprising 250 surface water absorption tests (SWAT) and penetration depth measurements was conducted, covering variations in binder types, water-to-binder ratios, aggregate types, mineral additives, and curing regimes. Water absorption rates were monitored at 2- and 10-minute intervals, followed by systematic assessments of water, chloride penetration depths, and initial absorption rates. Utilizing this dataset, robust GEP models were developed, demonstrating high predictive accuracy for SWAT, penetration depths, and initial water absorption rates. A key novelty of this study is employing GEP for comprehensive sensitivity analysis, revealing the relative influence of each parameter under varied conditions. The analysis identified that while binder types exhibited moderate sensitivity, water-to-binder ratios, mineral additives, aggregate types, and especially curing methods critically governed concrete’s resistance to water and chloride ingress. Notably, high-alite cement significantly enhanced water resistance, while slag incorporation effectively reduced chloride ingress, with curing quality remaining a dominant factor in mitigating both. These combined experimental and GEP-driven insights provide valuable guidance for optimizing concrete mix designs and curing practices to improve durability performance under aggressive environmental exposures.