Reducing the wall thickness in the input (by subtracting the corrosion allowance) is very wrong. Firstly this reduces the weight of the system - which is incorrect. Secondly the thinner wall makes the system more flexible - which is non-conservative.

So manually reducing the wall thickness introduces two huge errors in the analysis of the system.

The "All_Cases_Corroded" option corrodes all load cases (by reducing A and Z as Dave discusses above).