Statistical evaluation of corrosion of sialon in burner rig simulated combustion atmospheres

Samples of a sialon with low levels of aluminium substitution were exposed, for times up to 500 h, to a combustion atmosphere simulated in a low-velocity burner rig at 1300 °C. Corrosion was evaluated by metallographic measurement of multiple polished sections. It was found that the extent of corrosion varied systematically along the length of each sample, which has been explained as being the result of creep of the corrosion scale towards the base of the sample. The data from a number of measurement points along a sample were combined using a normalizing procedure so that a sufficiently large data set could be acquired for statistical analysis. Extreme value statistics were used to analyse the section loss data and the extreme value parameters μe and σe were determined and correlated with time of exposure. The most likely extreme value, μe, was found to follow a parabolic relationship with time which is in conflict with mass change results, where the parabolic rate constant decreased at longer times. The time-dependent correlation for the extreme value parameters was used, in conjunction with the reproductive property of the extreme value model, to predict the most likely maximum extreme value at longer times and for larger sample sizes. The results of this prediction were compared with data for superalloys corroded in similar atmospheres at 900 °C. The behaviour of the sialon was found to be comparable with that of uncoated superalloys.