Reliability based evaluation of commonly applied corrosion mitigation techniques

By identifying possible degradation threats and assessing remaining asset serviceability corrosion risk assessment is normally used to assign priorities for corrosion control and monitoring such as timely application of coatings, cathodic protection or ultrasonic inspection programs. Since these methods rely on probabilistic modelling of limited representable data, it follows that prediction of information reliability is of interest in modern asset management. The present paper interprets literature statistical data for (i) pitting corrosion depth as a function of time, (ii) dry film thickness, (iii) ultrasonic wall loss measurements and (iv) cathodic protection current fluctuations. It is shown that representation and predictions can be highly influenced by the choice of distribution used to fit the data. This has direct implications for future likelihood predictions. The reasons are discussed.