The curriculum of the Faculty has been radically revised over the last two years and is now based on the concept of "Problem-Based Learning". The subject matter taught is divided thematically into specific issues that are taught in six week blocks. The vehicles for these blocks are specially selected and adapted case studies prepared by teams of staff members. These provide a focus for integrating specialist subjects around a studio based design theme. In the case of second year this studio is largely computer-based: many drawings are produced by computer and several specially written computer applications are used in association with the specialist inputs.

This paper describes the "block structure" used in second year, giving examples of the special computer programs used, but also raises a number of broader educational issues. Introduction of the block system arose as a method of curriculum integration in response to difficulties emerging from the independent functioning of strong discipline areas in the traditional work groups. The need for a greater level of selfdirected learning was recognised as opposed to the "passive information model" of student learning in which the students are seen as empty vessels to be filled with knowledge - which they are then usually unable to apply in design related contexts in the studio. Furthermore, the value of electives had been questioned: whilst enabling some diversity of choice, they may also be seen as diverting attention and resources from the real problems of teaching architecture.

Assuming the availability of a more general building data model, one must define life and fire safety features of a building before any automatic checking can be performed. Object oriented data structures are beginning to be applied to design objects, since they allow the type versatility demanded by design applications. As one generates a functional view of the main data model, the software user must provide domain specific information. A functional view is defined as the process of generating domain specific data structures from a more general purpose data model, such as defining egress routes from wall or room object data structure. Typically in the early design phase of a project, these are related to the emergency egress design features of a building. Certain decisions such as where to provide sprinkler protection or the location of protected egress ways must be made early in the process.

The term "model" in the above paragraph has been used in various ways and in this context is defined as any representation through which design intent is expressed. This includes accurate/ rational or abstract drawings (2- dimensional and 3-dimensional), physical models (realistic and abstract) and computer models (solid, void and virtual reality). The various models that fall within the categories above have been derived from the need to "view" the proposed design in various ways in order to support intuitive reasoning about the proposal and for evaluation purposes. For example, a 2-dimensional drawing of a floor plan is well suited to support reasoning about spatial relationships and circulation patterns while scaled 3-dimensional models facilitate reasoning about overall form, volume, light, massing etc. However, the common denominator of all architectural design projects (if the intent is to construct them in actual scale, physical form) are the discrete building elements from which the design will be constructed. It is proposed that a single computational model representing individual components supports all of the above "models" and facilitates "viewing"' the design according to the frame of reference of the viewer.

Furthermore, it is the position of the authors that all reasoning stems from this rudimentary level of modeling individual components.

The concept of component representation has been derived from the fact that a "real" building (made from individual components such as nuts, bolts and bar joists) can be "viewed" differently according to the frame of reference of the viewer. Each individual has the ability to infer and abstract from the assemblies of components a variety of different "models" ranging from a visceral, experiential understanding to a very technical, physical understanding. The component concept has already proven to be a valuable tool for reasoning about assemblies, interferences between components, tracing of load path and numerous other component related applications. In order to validate the component-based modeling concept this effort will focus on the development of spatial understanding from the component-based model. The discussions will, therefore, center about the representation of individual components and the development of spatial models and spatial reasoning from the component model. In order to frame the argument that spatial modeling and reasoning can be derived from the component representation, a review of the component-based modeling concept will precede the discussions of spatial issues.