423 U. Flemming and S. Van Wyk (eds.), CAAD Futures ‘93, pp. 423-435.  1993 Elsevier Science Publishers B.V. Printed in the Netherlands. A Hierarchical Model for Building Applications John R. Bedell Niklaus Kohler Institut für Industrielle Bauproduktion Universität Karlsruhe Englerstraße 7 7500 Karlsruhe, Germany Advanced computer-aided architectural applications must model buildings as multi-level compositions supporting distinct points of view. Hierarchies of encap- sulated, autonomous elements can be derived from ISO-STEP’s General AEC Ref- erence Model and configured for various applications. For analysis of life-cycle costs, we define a pyramid of evaluable production steps leading to the final build- ing; for optimization of renovation task schedules, a topological model of access paths and traffic flow. These separate viewpoints can be embedded in a single uni- fying structure permitting the communication and propagation of changes among its specialized aspects. Keywords: Design model, decision support system, object-oriented data model, building product model, STEP-GARM. 1 The Need for a General Building Model Recent work toward automation of design applications has required the creation of unconventional models to represent the design itself. Such a structure must be able to de- compose into separate nested components, dividing the design problem into more ap- proachable fragments while the application provides the designer with appropriate decision support. Building models have offered an especially rigorous domain for this re- search because of each instance’s complex, heterogenous nature. Each is in fact an inter- twining of structural systems and mechanical services from a number of disciplines with different points of view. While we cannot anticipate, let alone accommodate, every future application’s needs within a single all-embracing model, we can seek a common concep- tual foundation on which a variety of data models can eventually develop and communicate. We begin by briefly presenting our approach for the representation of building products as hierarchies of encapsulated, reusable modules, with simple, flexible connec- tions permitting the user a top-down or bottom-up approach to design. These we then ap- ply to two of our current projects in building product modelling: estimation of life cycle costs and scheduling of renovation tasks. We also consider the eventual unification of these structures into a multifaceted, extensible whole.