This paper deals with some important experiences gained from building a black-box model of a linz-donawitz (LD) steel converter. Steelmaking with an LD converter is a complex physico-chemical process where many variables have effects on the quality of the resulted steel. During the process a converter is filled with waste iron, melted pig iron and many additives, then it is blasted through with pure oxegyn to burn out the unwanted contamination. The complexity of the whole process and the fact that there are many effects that cannot be taken into consideration make this task difficult. It turned out that perhaps the most important step of the whole modeling task was the analysis of the large amount of data, the selection of relevant parameters and to find good strategy to deal with missing and biased data. The paper details the steps of data analysis, summarizes both the motivations of constructing several different neural models and the general experiences obtained through the whole project.