Architektur Routine(n)

Abstract

For quite some time now the architectural design process has to a large extent been driven by and developed with the aid of computers. The assessment of buildings, on the other hand, has to this day remained an exclusively human domain. A variety of design parameters, which for instance are found in the space allocation programs of competition briefs, could potentially be captured and processed by a machine and thereby generate an added value for the work- and learning-environments of architects. The main focus of this work is the development of an open system that enables the machine-driven evaluation of a building design. Such a system requires a continuous and uninterrupted formalization of the whole design cycle – from the verbal description of architectural qualities to the capturing of relevant aspects of specific design solutions presented via a Building Information Model (BIM).
First, the EXTENSIBLE SPACE ALLOCATION LANGUAGE (XSAL) is introduced: the proposal of an open standard to describe the building criteria based on the meta-language XML. Building specifications noted in XSAL can be read and understood by humans and its inherent information can also be processed by machines. The second emphasis of this work lies in the development of a suitable building-datamodel. Due to constantly rising technical requirements for buildings, most of today’s Building Information Models (BIM) have grown to become highly complex data-bases. The author proposes a much simpler associative building-datamodel for the evaluation of design solutions. Rather than drawing from civil engineering, engineering and construction, this model relates to the notion of ‘design thinking‘ and to the way architecture is described.
The ARCHITEKTUR ROUTINE(n), a collection of algorithms to detect and explore qualities of interest or of concern, are the third and connecting element between the building specification and the digital model. The ALGORITHMIC EXPLICATION was developed in order to find a more precise language for describing a design task and to convert the often vague terms of our everyday language into quantitative (measurable) and consequently computable form. Re-iterative and repeated use of this method creates an explicit and expandable vocabulary, which is not based on the a priori inter-subjective idea but which enables a dialogue between man and machine. Finally, ARCHILLES PRO, a software developed by the author, is used to demonstrate how the potential of such an explicit vocabulary can be integrated into the work-flow of everyday practice and how, through the definition of fitness functions, it can be applied to generative design methods. Furthermore, a new form of a language based Data-Mining is introduced to Building Information Models.