Pitting potential modelling of EN 1.4404 stainless steel

The development of a reliable tool to know the behaviour of stainless steel on localized corrosion is necessary to achieve a proper selection of the material in the structures design. In order to know the behavioural pattern of stainless steel on localized corrosion, a total of 60 samples of austenitic stainless steel were subjected to polarization tests. Based on experimental results, support vector machines are proposed to estimate pitting potential values analysing different environmental factors: chloride ion concentration, pH and temperature. The influence of parameters selection in the presented model is analysed in order to get the optimal structure of the pitting potential estimator. Correlation coefficient and root mean squared error are proposed as performance criteria to measure the deviation between experimental and predicted values. Results show that the presented model provides a great estimation capability, becoming a useful tool to compare the corrosion resistance of this material depending on the environmental conditions.