Mechanical performance prediction of air-entrained mortar by coupling electrochemical testing with back propagation neural network modelling

Abstract

Accurate analysis of electrochemical properties, along with a profound understanding of the intrinsic mechanisms governing variations in the electrochemical system, holds pivotal significance in predicting the mechanical performance of air-entrained mortar. This study employs resistivity and electrochemical impedance spectroscopy (EIS) techniques to elucidate the electrochemical attributes of air-entrained mortar. A novel equivalent circuit (EC) model, accounting for the incorporation of bubbles, is developed to derive electrochemical parameters for air-entrained mortar. The underlying mechanisms dictating alterations in the electrochemical parameters are revealed through microscopic characterization techniques. Additionally, a backpropagation (BP) neural network model is constructed to forecast the mechanical performance of air-entrained mortar. The results demonstrate that air-entraining admixtures (AEA)-introduced bubbles exhibit insulating properties, substantially increasing the electrical resistivity (ρ) and the bulk electrical resistivity (ρbulk) of air-entrained mortars. The developed EC model, which considers the air effect, aligns well with the test results in Nyquist plots. The insulating properties of AEA-introduced bubbles is attributed to the formation of a bubble shell, influenced by bubble film barriers and the sulfonate-mediated promotion and delay of Ca3Al2O6 and Ca3SiO5 hydration. The BP neural network model showcases exceptional predictive capabilities, with correlations surpassing 0.96. Consequently, the findings of this study can be applied to predict the mechanical properties of air-entrained mortar in practical engineering applications.