Within the project preparation phase, experienced professionals manually map design information onto process information with the aim to develop realistic and practical schedules. Unfortunately, the mapping itself is neither part of any underlying data model nor is it supported by current scheduling tools. As a consequence the process of setting up the data model for a schedule is still not supported formally. Huhnt and Enge described a modelling technique that addresses the missing linkage between design and process information^[1]. The approach makes use of so called component types. These are template sub-processes that describe the fabrication procedure of typical building components. Decomposing the building into components and assigning a component type to each component allows for formal support while scheduling. Depending on the decomposition of the building into components and the complexity of the involved component types the specification effort differs. The question about optimal component types arises: Which layout of building components and component types results in minimal specification effort? This paper presents a branch and bound algorithm to determine optimal component types. For a given schedule, which has been modelled based on component types, all possible decompositions into sub-processes are determined. During the decomposition process the encountered configurations are compared. Those with minimal specification effort are registered. Theoretical and practical examples are examined and discussed.