Functional product models in design so far were built up by asking for the main function of a product first and then breaking the main function down into sub-functions, until an appropriate level of granularity was reached. A problem of this approach is that a product often provides even more functions which are not subordinate to the main function. This paper presents an approach how these functions can be captured in a reliable and structured way, using the requirements of a product as a superstructure. The improvements of this new approach are demonstrated by predicting change propagation based on an improved functional product model and comparing it to the results of a change prediction using the conventional functional product model.@en