In this study, the electrochemical noise produced by general and localized corrosion (pitting) in the coupling current between identical electrodes of ASTM A516 steel in 5M NaOH with or without added chloride ion was examined using Wavelet analysis (WA) as a means of differentiating between the two forms of attack. It was found that, under pitting corrosion conditions (the presence of 0.1M Cl-), there was no high-frequency energy contribution seen, but there was an energy contribution of the lowfrequency crystals. Additionally, visual time domain examination of the crystals produced by WA shows the individual contributions that were made by each frequency range toward events in the original signal. This allows for the identification of the onset and termination of individual corrosion types.
Currently, determination of the onset of various types of corrosion and the measurement of the severity or rate at which the corrosion occurs are important topics for many industries. In particular, a major challenge exists in relating electrochemical observations to corrosion rate, particularly when the data are collected in real time. Promising real time techniques include electrochemical noise analysis, in which the apparently stochastic (i.e., random) noise (EN data) in potential and current is analyzed to extract important corrosion parameters. For example, the ratio of the standard deviation in the potential noise divided by the standard deviation in current is identified as the noise resistance, which is often identified with the polarization resistance, from which the corrosion current density and hence the general corrosion rate can be calculated. However, this is strictly true only if the noise data are collected over an effectively infinite time, corresponding to zero frequency in the frequency domain. Compliance with this condition is rarely, if ever, tested, with the result that the polarization resistance is frequently under estimated.
Another technique that has gained popularity is wavelet analysis (WA), because of its ability to retain time information that is normally lost by Fourier transformation in the determination of the power spectral density (see below). Thus, WA, as used in this study, has the primary benefit of being collected in real time, and this collection can be done remotely and automatically. A major task facing the development of this type of monitoring?in addition to the creation of accurate, practical sensors?is the derivation of practical, useful information from the fluctuations in current and voltage, or related EN data. The derivation of useful information is the focus of this study.
