Using Artificial Neural Networks to Model the Environmental Dependence of Pit Growth in Aluminum Alloys

High strength aluminum alloys (specifically AA7075-T651, Al-5.6Zn-2.5Mg-1.6Cu) used in aerospace applications are highly susceptible to localized corrosion. Pits have been shown to nucleate fatigue cracks and since these alloys are subject to cyclic loading in service, characterizing pit growth as a precursor to fatigue cracking is necessary. To achieve this, AA7075 samples were constantly immersed in solutions of varying temperature, pH, and chloride concentration and serially removed after 1 hr, 1 day, and 1 month exposure time. In addition to this, two different orientations were investigated. These samples were then assessed by optical profilometry to quantify the pit dimensions at each condition. Each pit depth distribution was fit to a two-parameter Weibull and artificial neural networks (ANN) were utilized to model both Weibull parameters as well as the maximum pit depth as a function of temperature, solution pH, [Cl^-], exposure time, and orientation. Pit depth kinetics were extracted from the ANN and fit to a power law.