After rolling, the plates will cool down in a delayed manner either in free
piles or stacked in bins. In the case of delayed cooling, hydrogen will ha
ve the opportunity for effusion. The efficiency of effusion is thereby cont
rolled by the time evolution of the plate temperatures in the bin. In order
to get information on that evolution, the stack-in and stack-out temperatu
res were measured. As there are different parameters influencing the temper
ature evolution, a neural network approach was used for quantification of p
arameter dependencies. By that it was possible to quantify the influence of
each parameter separately. The result of that investigation was used for p
rocess optimization.