DIGITAL DESIGN STRATEGIES
WAYS TO THINK, CONCEIVE AND IMPLEMENT THE DIGITAL IN ARCHITECTURAL
DESIGN
Author: Jon Arteta Grisaleña
Adapted abstract of PhD Thesis:
“The Paradigm of Complexity in Architectural and Urban Design”
University of Alcala, 2017
(Chapter 3: Digital Tools)
ABSTRACT:
The text describes main digital design tools and strategies in the contemporary context,
grouping them into three main categories - representational methods, parametric
methods and generative methods. The objective is to offer a clear and didactic text that
helps to understand the foundations of digital architecture, showing its influence on the
conception and development of the architectural project.
KEYWORDS: digital architecture, digital tools, algorithmic design, parametric design,
generative design
1
_INTRODUCTION: TOWARDS A THEORY OF DIGITAL DESIGN
In a general way, we can define the digital architecture as the one which uses
digital tools in some of the phases of the architectural project. This will in turn imply the
need to translate (at some point) the architectural ideas into a programming language,
so that the communication between designer and machine can be produced. Every
design that makes use of the computer (nowadays practically 100% of the architectural
projects) should consider, then, a strategy to relate architecture and digitality.
In most cases, this link is usually established in an unconscious or predefined
way, limiting the digital to the use of commercial software such as Autocad or similar.
This does not mean, however, that the relationship between architecture and digitality
should always occur like this. In fact, by neglecting or unconsciously assuming this
relationship between architecture and digitality, we give up exploring a field full of
possibilities for innovation and the development of new architectural proposals.
It is necessary, therefore, to build a "theory of digital design", a theory that
explores this field of possibilities and investigates the different ways of communicating
the digital and the architectural. It will be necessary then to go into the "black box" of
computing, in oder to have a wider and deeper understanding of the digital and its
implications for architecture.
Some authors such as Kostas Terzidis, Rivka Oxman, Nick Dunn and Toni
Kotnik, among others, have deepened in this topic, providing visions and fundamental
conceptual proposals for the development of a true knowledge of digital design. Our job
will be to synthesize and reinterpret some of these ideas in order to establish a general
map of what would be the main trends or "digital design strategies" linked to
architecture. Each one of the strategies will reveal a way of thinking, of conceiving and
of implementing the digital within the architectural design.
2
A_REPRESENTATIONAL METHODS
This type of methods refers to the use of the computer exclusively as a
drawing or representation tool. Digital tools are conceived as a resource to facilitate the
graphic representation of architectural forms.
In this case the communication between designer and computer is produced
through commercial CAD packages, which allow the user to represent geometric
shapes from a certain number of prefixed commands. Although each command or
action will involve a process of computation or digital calculation inside the machine,
these processes will remain hidden from the user, whose function is limited to the
control and manipulation of the resulting forms.
This kind of strategies promotre the figure of the designer as the only creative
agent, from whose mind arise the ideas and geometric shapes that must later be
translated into digital graphics. The designer maintains full control over the resulting
shape in each and every step of the design, obeying the visual and formal logic typical
of traditional paper techniques.
Figure 3.12: Conceptual scheme of representative digital design strategies
Author: Jon Arteta
The use of traditional CAD programs will offer two main advantages: on one
hand, they allow to automate and speed up the mechanical and repetitive tasks always
present in design work. This will be possible thanks to resources such as copy / cut /
paste commands, blocks or external references in classic CAD programs, or
automation between 2D and 3D views characteristic of BIM platforms.
3
On the other hand, digital representation tools allow drawing and manipulating
complex geometries that are difficult to handle on paper. The possibility of drawing and
visualizing figures in 3D and the incorporation of geometric resources such as NURBS
(Non-Uniform Rational B-Splines), will allow the designer to access new levels of
freedom in terms of development and experimentation. with new architectural forms.
Figure 13: Frank O. Gehry represents a singular case within the representative strategies, since
previously to the manipulation of the digital model, the architect develops sketches and physical
models that are later scanned (in 3D) and transformed into digital graphic representations and
NURBS surfaces. Gehry was one of the first to employ the "Catia" aerospace engineering
software for architectural design. Source: images from various publications.
The shift from Representational to Algorithmic logic
Representational strategies can be considered as the most usual in the current
architectural context, although they should not be considered as the main development
path of the "digital architecture". In fact, authors such as K. Terzidis or T. Kotnik will be
especially critical of this way of working, coming to consider it as non-digital (Terzidis,
2006, Kotnik, 2010). For these authors, representational strategies constitute an
excessively superficial approach to the digital domain, an approach on which the
designer never becomes aware of the computational and algorithmic processes implicit
in the design process.
From a personal point of view, I think that representational techniques have
had an important influence on architectural design and experimentation, and therefore
they should be considered as an important part of digital architecture, regardless of
conceptual subtleties.
4
On the other hand, we have Algorithmic strategies, in which the designer
enters the "black box" of computer, participating directly in the design of algorithms and
computational languages. This implies a new way of approaching the digital and the
use of the computer, which ceases to be an auxiliary tool to become a true "co-worker”.
As [Link] said in his 1960 article, man and machine should be understood as
entities with different and complementary skills and abilities, and there is an important
potential for collaboration between them. Algorithmic strategies will focus precisely on
the exploration of this potential, in order to advance in the development of new
methodologies and project strategies.
Conceiving the computer as a co-worker involves certain degree of uncertainty
in the final result of the design, accepting the contingencies derived from calculation
and computational operations. This, in turn, leads to a new way of conceiving the
architectural project itself; it will no longer be an exclusive creation of human inspiration
and creativity, but rather an experiment, an exploration closer to scientific methodology.
To communicate the man and the machine, it is necessary that the messages
are expressed in a common language, a language that both are capable of
understanding and coding / decoding. The algorithm is the way to express any
message in this common language, so we should consider it as "the mediator between
the human mind and the computer processing capacity" (Terzidis, 2006; 16). The
design of algorithms is a basic task in all design strategies linked to programming and
computing, hence the name of "algorithmic strategies". There will be many and very
diverse ways to program and use algorithms in the architectural design, being able to
distinguish mainly two great types of strategies, developed below.
5
B_PARAMETRIC METHODS
“Parametric design allows the designer to define relationships between
elements or groups of elements and assign values or expressions to organize and
control these definitions” (Dunn, 2012; 54).
Figure 3.14: Conceptual scheme of parametric digital design strategies
Author: Jon Arteta
Unlike representational strategies, in which digitally represented elements
tend to be independent of each other, in parametric strategies the different elements
and components will be highly interconnected through relational laws. Thus, the
designer not only devises the elements or basic components of the system, but also
must design and program the links between them through mathematical functions
The attached image is a screenshot of the Grasshopper software (Rhino plug-
in), in which we can see the design of the relationships between elements, following a
characteristic scheme in the form of a tree. Some authors call this scheme "body plan"
(De Landa, 2002) or "metadesign" (Spuybroek, 2010).
Relational functions or laws will depend, in turn, on a series of internal
variables or parameters that will allow the same relational law to manifest itself in
different ways. Depending on the values assigned to these parameters, the model
acquires different dimensions and forms. So, we can say that the "body plan" is a
6
generic model, a field of possibilities whose final manifestation will depend on the
values assigned to the different parameters. In the attached grasshopper image we can
check this phenomenon, with its formal consequences
Figura 3.15: Example of parametric model, made with Grasshopper software.
Author: Jon Arteta
Another fundamental characteristic of the parametric method is that the model
tends to work as a whole, so the change in a parameter will affect all the associated
elements (chain reaction), resulting in a complete (and “instantaneous”) update of the
whole model.
The architect Daniel Davis, author of the doctoral thesis entitled "Modelled on
Software Engineering; Flexible Parametric Models in the Practice of Architecture"
(Davis, 2013), dedicates some paragraphs to describe the particularities of parametric
systems, clarifying all the possible misunderstandings that arise around this type of
methods. According to Davis, "a parametric model is unique, not because it has
parameters (all design, by definition, has parameters), not because it changes (other
design representations change), not because it is a tool or a style of architecture, a
parametric model is unique for what it does but rather for how it was created. A
parametric model is created by a designer explicitly stating how outcomes derive from
a set of parameters. " (Davis, 2013; 31). Thus, Davis emphasizes the importance that
7
the relationships between the different elements of the parametric system are
expressed explicitly, through mathematical functions (through computer code)
Digital tools and “performance-oriented” design
As we have just seen, the design of a parametric architecture can be divided
into two phases: the first consists in the elaboration of the "body plan" or generic
relationship scheme, while the second part consists in the assignment of concrete
values to the existing parameters, in order to reach a final result with a specific
geometry. The designer will have total freedom to determine these values, being able
to dictate them intuitively, or to consider values from tables or external databases. It is
precisely in this second case where the greatest interest of parametric architecture lies;
it is possible to connect the laws of formal genesis (relational laws) with data related to
external phenomena, such as sun, wind, noise, topography, etc. This design strategy
allows to create sensitive geometries, directly connected to the surrounding
phenomena. This is known as "performance-oriented design" or "performative"
design.1.
This aspect is explained by Professor Michael Weinstock through a biological
analogy, assimilating the process of development and growth of living beings to the
process of parametric design (Hensel, Menges & Weinstock, 2010). At present we
know that the form adopted by a certain living being does not constitute a direct and
pure translation of the information contained in its genetic code (genotype), but a
manifestation of this genetics adapted by the forces of the environment in which it lives
(phenotype). The genesis of the final form will be based on the interaction between
internal generic laws and the singularities of the environment, a phenomenon that will
be perfectly identifiable within the parametric architecture, with the interaction between
the bodyplan and the environmentally sensitive parameters.
As in living beings, the generation of the parametric form is conceived as an
evolutionary process, a process in which architecture varies its form to adapt to the
different requirements of the project. As Rivka Oxman indicates, this design strategy
"has transformed the concept of form into the concept of formation" (Oxman R., 2006;
249). This will involve the transition from an "objective logic" (based on pure, rigid
forms and with clearly defined dimensions) to a much more free and flexible "process
logic", in which the formal guidelines are defined by a kind of genetic code, by an open
1 To develop the “Performance- Oriented Design” it will be necessary to conceive the
architecture as an active agent, as an element that directly modifies the cahraceristics of its
environment. Thus, the “performative design” will consist in designing the architecture taking
into account its environmental consequences, its associated behaviors. For this, it will be
fundamental to know the environment and experiment with it, simulating it to obtain operational
data that will help us to direct the design of architectural geometries.
8
"dispositional" scheme; a scheme that describes a mechanism, a process of formal
generation or morphogenesis.
Figure 3.16: Example of "performative" design using sunlight data
Connecting geometries and “performative” functions in a logical and operational
way will lead us to a new understanding the relationship between form and function in
architecture, replacing the typical confrontations of the Modern Movement with an
integrating logic, with formal and functional precepts generating a mutual feedback
loop.
Parametric tools and “interactive design”
It is posible to incorporate the time factor within the parametric schemes,
giving rise to "interactive design". In interactive designs some parameters can vary
their value over time, since they are permanently connected to sensors or mechanisms
that collect information from different phenomena in real time. These phenomena can
be very diverse: changes in sunlight or atmospheric phenomena, information about the
movements/physical actions of people, etc. Interactive designs are usually considered
as experiments located at the limit of the discipline, blurring the boundaries between
architecture and machine. This will lead to debate and question many concepts of the
architectural discipline, such as the idea of authorship, the concept of finished product,
the evolutive capacity of architecture, etc.
9
Figure 3.17: Photograph of the HypoSurface project, developed by Mark Goulthorpe / dECOi.
Source: [Link]
Finally, note that there may be different degrees of experimentality within the
parametric design. If the designer presents a scheme with very simple and obvious
interrelations, with parameters defined quantitatively by himself, the parametric design
will be very close to what we have previously defined as representational strategies. If,
on the contrary, relationships become more complex and the parameters are defined
from large amounts of data managed by the computer, the designer will no longer have
complete control over the final result, which will remain open to a certain level of
contingency and surprise. In this last case it is where the computational and algorithmic
nature of the parametric design is most clearly manifested.
Parametric design strategies can be developed using different comercial
software, such as Grasshopper (plug-in for Rhinoceros), Generative Components
(Bentley) or Dynamo (Revit), among others.
10
C_GENERATIVE METHODS
In the generative strategies, the designer proposes an algorithm or basic
computer instruction that, when executed by the computer recursively, leads to the
generation of the architectural form.
Figure 3.18: Conceptual scheme of generative digital design strategies
Author: Jon Arteta
Developing a generative strategy is equivalent to proposing a scientific
experiment, in which the researcher defines three basic aspects: the elements that take
part in the experiment, the initial disposition of these elements, and the generic law that
determines how these elements should interact with each other (algorithm). To know
the final result, the researcher must execute the experiment, which consists of
calculating and simulating all the possible interactions between the base elements.
This is where the computer and its computing capacity become essential tools.
In parametric strategies, the designer builds and controls all the relations
between the elements of the project; the architect explicitly designs all the steps and
branches of the relationship tree. In generative strategies, on the other hand, the
designer only controls the initial conditions; all the intermediate stages necessary to
define the final form are developed in an implicit or hidden way, inside the computer.
The degree of uncertainty and experimentality increases, but always under the
certainty that, whatever the result obtained, this will be fully consistent with the
conditions or laws established at the beginning.
11
In generative strategies it is common for highly complex forms to emerge from
relatively simple initial laws, breaking the linearity of human intuition and rationality. In
this type of strategy is where the need for the computer as a means of design becomes
more patent, as a tool capable of exploring territories hitherto inaccessible to the
human mind.
The generative processes described in the algorithms can be of artificial
origin, that is, devised by the mind of the designer and based on purely intellectual
criteria (as in the case of the Toyo Ito’s pavilion for the Serpentine Gallery, image
below), or of natural origin, understood as direct translations of processes and real
phenomena present in nature. In this second case, generally the most common, the
formulation of the algorithm is a consequence of the observation, quantification and
formalization of real phenomena.
Figure 3.19_Serpentine Gallery Pavilion 2002, designed by Toyo Ito.
The algorithm that gives rise to this pavilion is inspired by purely geometric criteria, and is based
on the recursive repetition of the same pattern of tracing, subdivision and intersection between
lines. Source: composition made by the author with images from different publications.
Regardless of their origin, generative algorithms can be distinguished by their
function, existing two large groups:
12
Process-Oriented Generative Algorithms
"Process-oriented generative algorithms" are those whose development is not
linked to any specific purpose or objective, but to the mere fact of reproducing and
simulating a given process or phenomenon. This will be the case, for example, of most
of the so-called "complexity algorithms", commonly used for the study and analysis of
different "complex phenomena" present in real life. The attached table shows some of
them, with their respective explanations and related software tools.
Figure 3.20: Summary table of some of the main "complex algorithms", elaborated based on the
information collected in Reynoso, 2010
13
Optimization Algorithms
Optimization algorithms are defined by a clearly defined goal. The purpose of
these algorithms is the optimization of processes, the discovery of the most appropriate
solution among all possible options.
In the field of architecture, these resources are used to create forms that
optimize the structural behavior, the degree of sun exposure, etc. It should be noted
that these "form-finding" processes are not exclusive to the digital age, since they were
already present much earlier in the works of masters such as Antoni Gaudí or Frei
Otto. The big difference is that now, thanks to digital media, it is possible to explore a
greater number of phenomena.
Figure 3.21_Florence New Station. Competition proposal 2003.
Project designed by Arata Isozaki with the collaboration of engineer Mutsuro Sasaki.
The shape of the structure is designed with the optimization method designed by Sasaki and
called "Extended Evolutionary Structural Optimization method" (Extended ESO method). This
method will be based on the growth process of the roots of the "subtropical Banyan Tree"
Source: Sakamoto, 2008; 108-115
From the computational point of view, the optimization algorithms are
characterized by having a more sophisticated structure, since the law of generation is
complemented with evaluation and communication loops. This allows iterations to be
made, advancing progressively towards results closer and closer to the final goal
14
Figure 3.22_ Outline of the generic operation of an optimization algorithm
Author: Jon Arteta
There are many types of optimization algorithms (selective search, ant colony,
etc.), each of them with its own singularities and fields of application. In the field of
architecture, one of the most used algorithms in recent times is the "genetic algorithm".
Devised by John Holland in the 70s, this algorithm is based on an analogy with the
principles of genetics and natural selection; it uses recombination, mutation and
selection processes as a means to achieve optimal results. Its application in the
architectural field has been claimed by diverse authors such as the architect John
Frazer, the engineer Mutsuro Sasaki or the philosopher Manuel de Landa, among
others.
As indicated by Terzidis, the use of algorithmic optimization tools in
architecture does not mean that the computer will directly present a specific or
definitive form; architectural problems are always under-specified, so the algorithms will
generate a more limited field of possibilities, a range of options on which the architect
must subsequently apply his own criteria and preferences.
Generative design strategies can be developed using comercial software such
as “Galapagos” (developed as a plug-in for Grasshopper/Rhino), or “Generative
Design” (Autodesk), among others.
15
_NOTES AND CONCLUSIONS
In this section we have distinguished different strategies and digital tools
within the field of architectural design. The objective is to clarify basic concepts and
methods from a theoretical point of view, although it is necessary to indicate that, when
using them to respond to complex architectural requirements, different methods are
usually combined, giving rise to mixed or hybrid strategies.
One of the most evident examples of this process of hybridization is the
implementation of parametric resources within typical representational programs. This
is the case of BIM platforms such as Revit or Archicad, in which it is possible to
establish parametric relationships between different elements. For example, it is
possible to link the position and dimensions of the pillars to the position of the structural
axes, to the distance between slabs, etc. In any case, these parametric commands are
intended for the mechanization of drawing processes, saving mechanical work, without
major implications at project level. This implies that these tools continue to be linked to
a merely representative function.
This kind of programs, however, also have more ambitious updates, such as
the connection with simulation and algorithmic calculation programs. The objective is to
have information about the behavior or "performance" of the building while it is
designed, in order to inform design decisions during the process. It is necessary to
emphasize, however, that these improvements are still far from being a fully operative
and easy-to-use, but latest improvements point clearly towards this direction.
Apart from the changes in commercial software packages, the combination of
different digital design methods is increasing in the design methodology of architects,
especially in those areas related to research. A representative example of this type of
practice is the "Desert Pavilion" project, developed by the architect Andres Harris,
which brings together different techniques, both analogue and digital, as a way to
obtain a final form. As we can see in the picture below, Harris's proposal combines
physical models, parametric modeling, algorithmic simulation, digital representation,
etc.
16
Figure 3.23: “Desert Pavilion” Project, by Andres Harris
Source: [Link]
The architect Rivka Oxman talks about exploring the capabilities of the
computer to accompany the designer not only in the representation or graphic phase of
the project, but in all the phases involved in the architectural design: ideation, decision-
making, evaluation of results, development of technical aspects, etc. The digital offers
a wide field of experimentation, and the combination of different strategies in a free and
creative way seems the most appropriate way to achieve a digital architecture that is
operational and adapted to the increasingly complex architectural and urban problems.
Through this text we have introduced some of the main foundations of
contemporary digital design, presenting its challenges and possibilities. From now on, it
will be necessary to continue researching and exploring the potentials of these tools, in
order to achieve more creative, operational and efficient architectures and design
processes.
17
BIBLIOGRAPHY
Bundy, A. (2007). Computational Thinking is Pervasive. Journal of Scientific and Practical
Computing, 67-69.
Davis, D. (2013). Modelled on Software Engineering: Flexible Parametric Models in the Practice
of Architecture. Tesis Doctoral. RMIT University.
Dunn, N. (2012). Proyecto y Construcción Digital en Arquitectura. Blume.
Frazer, J. (1995). An Evolutionary Architecture. Architectural Association.
Gausa, M. (2010). OPEN: Espacio, Tiempo, Información. Barcelona: Actar.
Hensel, M. (2013). Performance-Oriented Architecture. Wiley.
Hensel, M., & Menges, A. (2008). AD "Versatility and Vicissitude". Wiley.
Hensel, M., Menges, A., & Weinstock, M. (2006). AD Techniques and technologies in
morphogenetic design. Wiley-Academy.
Hensel, M., Menges, A., & Weinstock, M. (2010). Emergent Technologies and Design. Towards
a biological paradigm for architecture. Routledge.
Kotnik, T. (2010). Digital Design as Exploration of Computable Functions. International Journal
of Architectural Computing, 8(1).
Lawson, B. (2005). How Designers Think. Architectural Press / Elsevier.
Licklider, L. (1960). Man- Computer Symbiosis. IRE Transactions on Human Factors in
Electronics, HFE-1, 4-11.
Mark, E., Mark, E., Oxman, R., & Martens, B. (2001). The Ideal Computer Curriculum.
Architectural Information Management, 19th CAAD Conference Proceedings (págs. 4-
11). Helsinki: Penttila.
Ortega, L. (2009). La digitalización toma el mando. GG.
Oxman, R. (2006). Theory and Design in the First Digital Age. Design Studies (27), 229-265.
Oxman, R., & Oxman, R. (2010). The New Structuralism. Design, engineering and architectural
technologies. Architectural Design, New Structuralism, 15-23.
Reynoso, C. (2010). Análisis y Diseño de la Ciudad Compleja. SB.
Sakamoto, T. (2008). From Control to Design. Parametric/Algorithmic Architecture. Actar.
Spuybroek, L. (2010). NOX Diagrams. AD Reader. The Diagrams of Architecture, 270-281.
Terzidis, K. (2006). Algorithmic Architecture. Routledge.
18
WEBSITES
[Link]
[Link]
[Link]
[Link]
[Link]
[Link]
[Link]
19
