Statistical study of pit propagation in carbon steel under nuclear waste disposal conditions

Abstract

One aspect of a project aimed at specifying the metal allowance required to prevent the corrosion penetration of carbon steel nuclear waste containers over a 1000 year period in a geological repository is described. This involves a statistical analysis of long term pit growth data from potentiostatically controlled tests in 0·1 M NaHCO₃ + 1000 ppm Cl⁻solution at 90°C. In general it was found that these data fitted unlimited exponential distribution functions. This was surprising because it was anticipated that charge and mass transport effects would set an upper limit to pit depth. That this limit was not detected may be because it is large, or because the specimen surface area was insufficient to detect the ‘tail off” characteristic of a limited distribution. Using the exponential distributions it has been shown that for probabilities less than 0·2 the pit depth x(mm) having a probability P_A(x) of occurring in a surface area A increases with time t(h) according to the expression: x = 9·5 × 10⁻⁵ t + 1·28 × 10⁻² to^0·38 {In[Aρ/ρA(x)]} where ρ is the experimentally measured pit density, which should not vary significantly with exposure time. The non-statistical term in this expression is interpreted as indicating that the pits are followed by what is effectively a general corrosion front, consisting of a high density of shallow pits.