Generative Algorithmic Techniques for Architectural Design

Architectural design methodology is expanded through the ability to
create bespoke computational methods as integrated parts of the
design process. The rapid proliferation of digital production techniques
within building industry provides new means for establishing seamless
flows between digital form-generation and the realisation process. A
tendency in recent practice shows an increased focus on developing
unique tectonic solutions as a crucial ingredient in the design solution.
These converging trajectories form the contextual basis for this thesis.
In architectural design, digital tools have predominantly
been used for representation and specification, and as an efficient
replacement of equivalent manual tools. Recent developments
allow architects to integrate computation in the design process.
Thus, computation is used for generating geometry and information
concerning realisation. In this situation, the architect’s attention
is directed towards the form-generating rules, which then reflect
design intents, contextual parameters and production constraints. An
advantage of this approach is the capability of managing substantial
complexity in terms of geometry and performance. This is partly due
to the fact that underlying mathematical logic allows information to be
generated on many levels with respect to development stages and the
needs of the receiver.
A necessity for establishing an architectural framework for
adopting and discussing generative techniques is addressed in this
research. As such, a series of methods directed towards specific
types of geometric and architectural problems is developed, and form
the basis for discussing the potentials and implications related to each
problem type. These methods are often inspired by models developed
within natural sciences, biology in particular. The principles are further
developed to form new modes of articulation in architectural design.
Certain methods are contributions, which suggest a potential for
future use and development. Thus, a method is directed towards
bottom-up generation of surface topology through the use of an agentbased
logic. Another method embeds a negotiation between patternformation
and contextual parameters in the form-generating process.
A third example demonstrates a flow of information and matter from
form-generation to realisation through the use of generative tools and
feedback loops in the design development. Three additional methods
are described, and reference examples help to establish an overall
discussion across the algorithmic and architectonic principles. Here,
potentials and implications for architectural design are discussed.
Both concerning a method oriented level, where different approaches
are compared, and a general perspective with respect to architectural
design methodology.