EVA Extreme Value Analysis is a statistical method used to predict extreme or rare events. Examples of rare events include extreme floods and snowfalls, high wind-speeds, extreme temperatures, large fluctuations in exchange rates, and market crashes. In NDT it is used to predict corrosion in areas that cannot be inspected based on data from areas that can be accessed.
If the corrosion is uniform then standard statistical distributions can be used to predict the average wall thickness loss. However, when the corrosion is non-uniform with localised pits then EVA has to be applied. The use of average pit depth will not allow prevention of loss of containment as this will occur when a single fast-growing pit grows through-wall.
Standard corrosion mapping data is collected in a number of different areas of the component. The maximum values of wall loss in these areas are put together to generate a new data set. Applying EVA to these extreme values allows a prediction to be made of the likely size of pits in an uninspected area, as long as it sees the same conditions as the inspected areas.

For more information on the use of EVA in NDT see:

What the hec?! articles are not intended to be the definitive account on the topic or acronym in question. Readers’ comments and contributions are welcomed. Email: ndtnews@bindt.org