Data-driven machine learning approaches for predicting permeability and corrosion risk in hybrid concrete incorporating blast furnace slag and fly ash

This study aims to identify the most suitable machine learning model for predicting the permeability and half-cell potentiometer test readings of hybrid concrete containing varying percentages of blast furnace slag and fly ash when exposed to a chloride-rich environment. The mix design adhered to IS 10262:2019 standards, and hybrid concrete beam specimens, incorporating protruding reinforcement bars, were fabricated with dimensions of 150 × 150 × 700 mm. Subsequently, these specimens underwent a 28-day curing process in water with 3.5% NaCl concentration, simulating chloride exposure that could induce corrosion by infiltrating the concrete towards embedded steel bars. The permeability and the compressive strength of the hybrid concrete were assessed by casting 36 cubes of 150 mm size after 28 days of curing. Half-cell potentiometer tests were conducted on the beam samples casted of hybrid concrete which were subjected to a chloride environment to generate the datasets used for the machine learning model to predict the permeability in concrete. The permeability values were predicted and compared using 3 machine learning models, i.e. Adaboost, random forest, and XGBoost. AdaBoost, in particular, demonstrated remarkable accuracy, showing strong correlations between the observed and predicted values. The findings of the study correlate and simplify the process of corrosion detection in concrete, ultimately aiding in the development of more accurate and robust corrosion monitoring systems.