The study of corrosion chemistry of aluminosilicate refractories in potlini
ngs has historically focused on a single corrosive agent: either molten sod
ium fluoride or metallic sodium. Each approach can explain only a fraction
of in-field observations. This article presents a unified approach, taking
into account the effect of both corrosion agents, on aluminosilicates. Alon
g with the model, mathematical tools have been developed to help interpret
experimental results as well as to make predictions of in-service behaviors
. Those predictions are based on the use of corrosion maps, which vary acco
rding to the sodium ratio [R-Na = N-Na(M)/(N-NaF+N-Na(M))], as well as othe
r parameters. Predictions made with the model match almost perfectly with t
he results obtained in a wide range of corrosion conditions. Depending on t
he quantity of fluorides that a lining lets percolate and the amount of met
allic sodium diffusing, cells can be divided into two distinct groups: wet
or dry. This article discusses the wet/dry distinction, including industria
l aspects.