Non-destructive investigation of corrosion current density in steel reinforced concrete by artificial neural networks

Corrosion of the steel reinforcement in concrete is an important problem for the civil engineering. Inspection techniques are needed to assess the corrosion in order to protect and repair concrete structures. Many studies were performed to establish a series of corrosion rate assessment methods. A method of providing a direct evaluation of the corrosion rate by corrosion current density measurement is Linear Polarisation Resistance (LPR). The main drawback is that it requires a localised breakout of the concrete cover. The corrosion of the steel reinforcement is monitored by measuring the resistivity of the concrete. The purpose of this paper is to use the resistivity four-probe method and galvanostatic resistivity measurement together with neural networks to assess the corrosion rate of steel in concrete without a direct connection to the reinforcement. Three parameters determined by two non-destructive resistivity methods together with the air temperature were employed as input variables, and corrosion current density, predicted by the destructive LPR method, acted as the output variable. The results shows that it is possible to predict corrosion current density in steel reinforced concrete by using the model based on artificial neural networks on the basis of parameters determined by two non-destructive resistivity measurement techniques.

Highlight
► The concrete resistivity measurement for the assessment of the corrosion rate by neural networks has been applied.
► Resistivity parameters and the air temperature were employed as input variables.
► The corrosion current density was employed as output variable.
► The prediction of the corrosion current density by neural networks is a viable method.
► Neural network model has a theoretical significance in the monitoring for corrosion current density.