Chloride-induced steel corrosion seriously affects the durability of reinforced concrete structures. Rubber concrete, an environmentally friendly construction material in which waste rubber is recycled as a concrete component, has demonstrated superior resistance to chloride-induced steel corrosion and the subsequent concrete deterioration. However, quantitative evaluation of the degree of deterioration in rubber concrete based on nondestructive detection is challenging due to the complexity of the material. In this paper, reinforced concrete specimens with rubber contents of 0, 10% and 20% are subjected to the electrochemically accelerated corrosion experiments and monitored by ultrasonic testing. Six machine learning models are trained by the data derived from the ultrasonic testing to predict the corrosion degree based on ultrasonic traits. The results show that the machine learning models except for the linear model can accurately and robustly predict the corrosion degree under the interference of outlier amplitude and size of training set. Furthermore, the corrosion-induced deterioration process is computed by mesoscale simulation based on the corrosion degree, so that the damages of specimens with different rubber contents are quantitatively evaluated. The experimental and computational studies prove that the addition of rubber into concrete effectively retards the corrosion of steel and the deterioration of concrete.


Civil, Architectural and Environmental Engineering


Shenzhen University, Grant 2020B1212060074

Keywords and Phrases

Corrosion degree; Machine learning; Mesoscale simulation; Robustness validation; Rubber concrete; Ultrasonic signals

International Standard Serial Number (ISSN)


Document Type

Article - Journal

Document Version

Final Version

File Type





© 2023 Elsevier, All rights reserved.

Publication Date

01 Apr 2022